Super-squash branch 'main' using huggingface_hub

.gitattributes

@@ -0,0 +1,46 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text
+figure_furcoat.jpg filter=lfs diff=lfs merge=lfs -text
+figure_hoodie.jpg filter=lfs diff=lfs merge=lfs -text
+figure_carwindow.jpg filter=lfs diff=lfs merge=lfs -text
+figure_subway.jpg filter=lfs diff=lfs merge=lfs -text
+figure_lace.jpg filter=lfs diff=lfs merge=lfs -text
+figure_strength.jpg filter=lfs diff=lfs merge=lfs -text
+figure_oldprint.jpg filter=lfs diff=lfs merge=lfs -text
+figure_vs_furcoat.jpg filter=lfs diff=lfs merge=lfs -text
+figure_vs_hoodie.jpg filter=lfs diff=lfs merge=lfs -text
+figure_vs_freckles_8mp.jpg filter=lfs diff=lfs merge=lfs -text
+figure_vs_headphones_8mp.jpg filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,346 @@
+---
+license: other
+license_name: qwen-research-license
+base_model: Qwen/Qwen-Image-2.1
+pipeline_tag: image-to-image
+tags:
+  - lora
+  - image-editing
+  - restoration
+  - upscale
+  - qwen-image
+  - qwen-image-2.1
+  - comfyui
+  - ai-toolkit
+---
+
+# Qwen Image 2.1 Restore LoRA
+
+Qwen Image 2.1 can upscale a small or worn picture by itself, but it overdoes it. Skin turns
+gritty, pores and freckles show up that were not there, faces look older, and the picture shifts a
+few pixels. I trained this LoRA to soften that harsh upscale. With it the result is a little
+softer and still detailed, it makes up less, and it stays true to the picture you gave it.
+
+I tested it on 37 pictures I had shrunk myself, so each result could be compared with the
+full-size picture. With the LoRA the result sat about 0.3 px off. Without it, more than 2 px.
+
+It is a redraw, not a recovery. The fine detail is made up to fit. Plain Qwen 2.1 draws more
+texture; this LoRA trades some of that for staying true to the picture.
+
+> **Blurry photos need a lower strength.** If your picture is out of focus, shaky, or a frame from
+> a video, set the LoRA to **0.5**. At 1.0 it stays true to the blur and the result comes back
+> soft. A small or compressed picture that is in focus is fine at 1.0.
+
+No trigger word. Start at strength 1.0, and lower it for a blurry photo or for more of Qwen's own
+texture (more on that below). Enlarge your picture to about 2 megapixels first (both sides
+divisible by 32), give it as image 1, and ask:
+
+```
+Restore the photo: remove blur, noise and compression artifacts and recover sharp, natural detail, keeping the same composition, people and colors.
+```
+
+For scratched or faded prints:
+
+```
+Restore the old photo: remove scratches, dust, fading and grain and recover sharp, natural detail and color, keeping the same composition and people.
+```
+
+I use it with [Viggle Turbo](https://huggingface.co/Viggle/Qwen-Image-2.1-viggle-turbo) at 8 steps.
+It also works at 25 steps without Turbo. That comes out a little closer and softer.
+
+Also on Civitai: the [LoRA page](https://civitai.com/models/2989697/qwen-image-21-restore-lora) and my
+[Qwen Image 2.1 Tiled Upscale](https://civitai.com/models/2988629/qwen-image-21-tiled-upscale) workflow, which uses this LoRA for results over 2 megapixels.
+
+![A woman in a pink fur coat under neon signs. Without the LoRA Qwen covers her face in pores and freckles that are not in the picture; with it the skin stays hers](figure_furcoat.jpg)
+
+*In each picture: the small input, the result without the LoRA and with it, and the marked section
+enlarged underneath. All made with the workflow in this repo, same prompt and seed, 8 steps with
+Viggle Turbo, about 2 MP, shown as rendered. "Without the LoRA" is the same workflow with the
+Restore row switched off. "px off" is how far the result sits from the full-size picture the input
+was shrunk from. None of these pictures are in the training set.*
+
+![A selfie in a hoodie. Without the LoRA Qwen adds wrinkles and pores and she looks older; with it she looks like the picture](figure_hoodie.jpg)
+
+![A woman resting on a car window. Without the LoRA her skin comes out rough and heavily freckled; with it she keeps the few freckles she has](figure_carwindow.jpg)
+
+![A man on a subway with earphones. Without the LoRA he gets rough skin and stubble that is not in the picture; with it he stays the same guy](figure_subway.jpg)
+
+![A close portrait in a lace top. Without the LoRA the freckles multiply and the whole picture gets darker; with it the face and the light stay](figure_lace.jpg)
+
+## Old prints
+
+For a scratched or faded print, use the old-photo request above. On four held-out old prints (pictures it never
+trained on, damaged the way the training pairs were), color and brightness sat 15 % off the clean picture before the
+restore, 7 % off after it with the normal request and 4 % off with the old-photo request. Scratches, dust and grain
+went with either request.
+
+A kind of fading it has not seen can stay. In a second test with a different fade (two pictures, in tiles), the
+scratches went and the faded color stayed.
+
+## Strength is a dial
+
+![The same picture without the LoRA and with it at 0.25, 0.5, 0.75 and 1.0: the skin gets softer and the picture sits closer to the input at each step](figure_strength.jpg)
+
+At 0 you get all the harsh detail Qwen adds. Turn the LoRA up and the picture gets softer and
+stays truer to your input.
+
+- **1.0** for a small or compressed picture that is in focus. This is what the pictures above use.
+- **0.75** for a little more of Qwen's texture. On 29 test pictures it still held the picture in
+  place (0.4 px off, against 0.3 px at 1.0).
+- **0.5** for a blurry photo. It is about halfway back to plain Qwen (1.3 px off, against 2.2 px
+  without the LoRA).
+
+**Blurry photos are where it matters most.** I ran 26 real low-resolution photos (old selfies and
+phone pictures) at 1.0, 0.75 and 0.5. On the 6 blurriest ones, 1.0 came back soft, 0.75 barely
+helped, and 0.5 came back sharp. On the other 20 all three worked, and 1.0 was the smoothest. The
+face stayed closest to the input at 1.0 and 0.75 (face match 0.94) and a little less at 0.5 (0.91).
+
+Redrawing a soft photo bigger also helped. Three of them, taken to 8 megapixels in tiles, came
+back sharp with the LoRA still at 1.0.
+
+## Files
+
+| file | notes |
+|---|---|
+| `qwen-image-2.1-restore.safetensors` | **step 1,500, start here.** The closest in my test, and the one I liked best on my test pictures. Skin comes out a little smooth. |
+| `qwen-image-2.1-restore-1250.safetensors` | step 1,250. Draws more fine detail and keeps more skin texture on noisy phone pictures. A little less close. |
+
+Rank 32, ComfyUI key format, loads onto the Comfy-Org Qwen Image 2.1 weights.
+
+## What I measured
+
+Eight of my own pictures, shrunk to 512 or 560 px on the long side and saved as JPEG (quality 70).
+The full-size picture is "the original" below. Every number is measured against it, and the model
+never sees it. Each small picture was enlarged to about 2 MP and restored with the same seed, LoRA
+at 1.0, CFG 1, euler / simple. Nothing is done to the result afterwards.
+
+| | no LoRA, 8 steps | with it, 8 steps | no LoRA, 25 steps | with it, 25 steps |
+|---|---|---|---|---|
+| how far the result sits from the original (worst corner, median) | 2.35 px | 0.26 px | 1.59 px | 0.15 px |
+| the same, worst of the eight | 3.83 px | 0.38 px | 3.25 px | 0.26 px |
+| close to the original (SSIM, 1 = identical) | 0.578 | 0.766 | 0.743 | 0.812 |
+| PSNR | 23.2 dB | 28.0 dB | 26.3 dB | 29.3 dB |
+| color and brightness shift | 2.05 % | 0.97 % | 1.72 % | 1.04 % |
+| same face (face recognition, 1 = identical) | 0.821 | 0.905 | 0.837 | 0.901 |
+| fine detail, against the original's | 121 % | 78 % | 58 % | 36 % |
+
+The 8-step columns use Viggle Turbo. The 25-step columns do not.
+
+- Plain Qwen 2.1 does much better at 25 steps than at 8. The LoRA helps most at 8 steps, and at
+  both it is what keeps the picture in place.
+- Fine detail over 100 % is detail that was made up: at 8 steps plain Qwen 2.1 draws more grain
+  than the original has. With the LoRA there is less than the original has, so results are a little
+  smooth, more so at 25 steps.
+- For scale, the small picture just enlarged scores 0.822 and 29.5 dB, because a blurry picture is
+  close on average. It has 3 % of the original's fine detail.
+- Same face is on seven of the eight. The face detector finds no face in the eighth original.
+
+**29 more pictures**, the same test at 8 steps: 0.28 px off with the LoRA against 2.16 px
+without (it moved less on all 29), same face 0.88 against 0.79 (closer on all 28 that have a
+face), color and brightness shift 0.94 % against 1.88 %, SSIM 0.777 against 0.611.
+
+**Every kind of damage.** 57 held-out pairs, damaged the way the training pairs were, on pictures it never trained
+on. One pass, 8 steps, the normal request. SSIM is against the clean picture, px off is the median.
+
+| kind of damage | pictures | SSIM, the damaged picture | SSIM, no LoRA | SSIM, with it | px off, no LoRA | px off, with it |
+|---|---|---|---|---|---|---|
+| small web picture | 15 | 0.772 | 0.524 | 0.716 | 2.53 | 0.31 |
+| phone smoothing | 5 | 0.836 | 0.565 | 0.857 | 2.65 | 0.17 |
+| video still | 5 | 0.771 | 0.578 | 0.745 | 2.77 | 0.50 |
+| saved and re-saved | 5 | 0.816 | 0.560 | 0.794 | 2.21 | 0.24 |
+| blur and noise | 5 | 0.496 | 0.521 | 0.678 | 2.48 | 0.24 |
+| low-resolution JPEG | 5 | 0.777 | 0.531 | 0.763 | 3.16 | 0.23 |
+| heavy mix | 5 | 0.735 | 0.646 | 0.746 | 1.28 | 0.25 |
+| old print | 4 | 0.551 | 0.437 | 0.749 | 2.46 | 0.19 |
+| banding | 4 | 0.915 | 0.463 | 0.843 | 2.34 | 0.17 |
+| low-resolution PNG | 4 | 0.873 | 0.636 | 0.847 | 3.39 | 0.15 |
+| all | 57 | 0.756 | 0.544 | 0.762 | 2.70 | 0.23 |
+
+With the LoRA the result was closer to the clean picture than without it on all 57, and it moved less on all 57. On
+several kinds the damaged picture itself scores higher than any restore, because a soft picture is close on average.
+
+**Strength**, on the 29 pictures above, without the LoRA / at 0.5 / at 0.75 / at 1.0: px off 2.16 / 1.31 /
+0.37 / 0.28, same face 0.794 / 0.858 / 0.875 / 0.880, color and brightness shift 1.88 / 1.26 / 1.05 /
+0.94 %, fine detail against the original's 102 / 81 / 77 / 72 %.
+
+The step 1,250 file on the first test at 8 steps: 0.26 px, 0.750, 27.3 dB, same face 0.899, fine
+detail 108 %.
+
+**Drawings.** It was trained on photos, and it holds on drawings too. On six test drawings (anime,
+a poster with lettering, watercolor, a 3D cartoon, manga, an oil painting), shrunk the same way, it
+kept the style and the lettering: SSIM 0.714 against 0.491 without the LoRA, 0.27 px off against
+2.65.
+
+I also put 12 portraits through three kinds of old phone and camera damage (36 pictures,
+colors matched back to the input afterwards). Same face was 0.76 without the LoRA and 0.85 with it.
+The damaged picture itself scores 0.84, so it sharpens these and keeps the person. It does not get
+closer to them than the damaged picture was.
+
+## Against other tools
+
+Before I released this I ran the other tools I could find for the same job, on the same pictures, with the same
+measurements. They are not all built for the same thing. Some are made to add detail. This LoRA is made to stay on the
+picture. So the numbers answer one question: how close does the result stay to the original?
+
+Tested on 2026-10-07:
+
+- [Elusarca's Qwen 2.1 Detail Enhancer](https://huggingface.co/reverentelusarca/elusarcas-qwen-2.1-detail-enhancer-lora)
+  v1.0, at 1.0 with the prompt from its author's workflow. Run two ways: in the same graph as this LoRA, and in the
+  author's own [2K upscale workflow](https://huggingface.co/reverentelusarca/qwen-image-2.1-workflows) as published.
+- [Deblur anything](https://civitai.com/models/2975487) by Luntrix, v1, at 1.0 with its trigger phrase.
+- SeedVR2 (3B), with ComfyUI's own nodes.
+
+Every number per picture is in [`comparison_numbers.csv`](comparison_numbers.csv).
+
+**29 small pictures**, the same test as above: shrunk, saved as JPEG, restored at about 2 MP, 8 steps unless noted.
+
+| | off the original | same face | color and brightness shift | SSIM |
+|---|---|---|---|---|
+| Qwen Image 2.1 alone | 2.16 px | 0.79 | 1.88 % | 0.611 |
+| SeedVR2 | 0.42 px | 0.82 | 0.44 % | 0.771 |
+| Deblur anything | 0.39 px | 0.85 | 1.42 % | 0.702 |
+| Detail Enhancer, same graph as this LoRA | 2.07 px | 0.76 | 2.94 % | 0.673 |
+| Detail Enhancer, its own workflow, 25 steps | 12.45 px | 0.71 | 4.16 % | 0.590 |
+| **this LoRA** | **0.28 px** | **0.88** | 0.94 % | **0.777** |
+
+- By SSIM this LoRA stayed closer to the original than the Detail Enhancer and Deblur anything on all 29 pictures
+  (28 of 29 against the Detail Enhancer in its own workflow), and than SeedVR2 on 18 of 29. It kept the face better
+  than the Detail Enhancer on 28 of 28, than Deblur anything on 26 of 28 and than SeedVR2 on 27 of 28.
+- SeedVR2 is the fastest by far, about 3 seconds against 11, and it keeps color best.
+- The Detail Enhancer's own workflow gives the model the picture at 1 MP and draws on a new canvas. That is why its
+  result sits 12 px off.
+
+![A woman in a pink fur coat under neon signs, from a 314 px wide JPEG. The Detail Enhancer's workflow smooths and redraws her face and moves the framing, Deblur anything covers her skin in pores and freckles, the Restore LoRA keeps her](figure_vs_furcoat.jpg)
+
+![A selfie in a black hoodie, from a 389 px JPEG. The Detail Enhancer's workflow smooths her face and moves the framing, Deblur anything paints a pattern on the hoodie, the Restore LoRA keeps the picture](figure_vs_hoodie.jpg)
+
+**At 25 steps without Turbo**, 13 of those pictures:
+
+| | off the original | same face | color and brightness shift | SSIM |
+|---|---|---|---|---|
+| Qwen Image 2.1 alone | 2.09 px | 0.85 | 1.58 % | 0.793 |
+| Detail Enhancer, same graph as this LoRA | 1.99 px | 0.79 | 3.20 % | 0.762 |
+| Detail Enhancer, its own workflow | 12.32 px | 0.73 | 4.17 % | 0.655 |
+| **this LoRA** | **0.17 px** | **0.90** | **0.76 %** | **0.860** |
+
+**Real photos.** 31 real low-resolution photos (the 26 from the strength section and 5 more), one pass at about
+2 MP, 8 steps. There is no sharp version of these, so each result is compared with its own input. Fine detail is
+against this LoRA at 1.0.
+
+| | same face | color and brightness shift | off the input | fine detail |
+|---|---|---|---|---|
+| Qwen Image 2.1 alone | 0.85 | 3.93 % | 2.72 px | 2.5 x |
+| SeedVR2 | 0.79 | 0.53 % | 0.28 px | 1.9 x |
+| Detail Enhancer | 0.73 | 7.33 % | 2.85 px | 2.6 x |
+| Deblur anything | 0.94 | 1.08 % | 0.33 px | 1.0 x |
+| this LoRA at 0.5 | 0.91 | 2.25 % | 0.98 px | 1.6 x |
+| this LoRA at 0.75 | 0.94 | 1.14 % | 0.33 px | 1.2 x |
+| this LoRA at 1.0 | 0.94 | 0.88 % | 0.25 px | 1.0 x |
+
+- On real photos the Detail Enhancer gives the sharpest picture, and it changes the face the most.
+- Deblur anything stays as close as this LoRA does.
+- At 1.0 this LoRA changes a soft photo very little in one pass at 2 MP. That is the blurry-photo case from the
+  strength section: lower it, or let the picture grow.
+
+**In tiles, to 4 and 8 MP.** Seven pictures with grain, noise, heavy JPEG or half the size, through my
+[Qwen Image 2.1 Tiled Upscale](https://civitai.com/models/2988629/qwen-image-21-tiled-upscale) workflow. SeedVR2 ran at the same output size.
+
+| | same face | color and brightness shift | off the original | SSIM | time |
+|---|---|---|---|---|---|
+| SeedVR2 | 0.87 | 0.67 % | 0.22 px | 0.813 | 7 s |
+| Detail Enhancer, its own workflow (to 4 MP) | 0.70 | 4.21 % | 12.88 px | 0.551 | 66 s |
+| tiles, no LoRA | 0.82 | 1.30 % | 0.89 px | 0.576 | 43 s |
+| tiles, Detail Enhancer | 0.77 | 1.83 % | 0.58 px | 0.682 | 43 s |
+| **tiles, this LoRA** | **0.88** | 0.72 % | **0.16 px** | 0.793 | 40 s |
+
+- SeedVR2 ties this LoRA here and is six times faster. It was sharper on the grainy picture and level on the noisy
+  one. On one heavy JPEG it drew the block edges as cracks in the skin.
+
+![A neon portrait saved as a heavy JPEG, taken to 8 MP. SeedVR2 draws the JPEG blocks as cracks under the eye, the Detail Enhancer's workflow changes her face and hair, the Restore LoRA in tiles brings the freckles back](figure_vs_freckles_8mp.jpg)
+
+![A grainy portrait with headphones, taken to 8 MP. SeedVR2 is the sharpest here, the Detail Enhancer's workflow is softer, the Restore LoRA in tiles sits between them](figure_vs_headphones_8mp.jpg)
+
+One seed per picture, one RTX 5090. If you make one of these tools and think I ran it wrong, tell me and I will run it
+again.
+
+## What it does not do
+
+- It does not bring back what the camera saw. A face stays the same person, not the same pixels.
+- A blurry photo comes back soft at 1.0. Lower the strength to 0.5 for those (see above).
+- Small faces are redrawn. A face about 100 px tall in the picture you give it can come back as a
+  slightly different person, and a much smaller one is a guess.
+- It was trained at 1 to 2 megapixels. Keep each pass at about 2 MP. For a bigger result, redraw
+  the picture in overlapping pieces of that size. My [Tiled Upscale](https://civitai.com/models/2988629/qwen-image-21-tiled-upscale) workflow does that.
+- It is for small or damaged pictures. Don't expect much from a picture that is already sharp at
+  the size you run it at.
+- A big picture that is noisy or worn should be shrunk first. In one test an 8 MP noisy JPEG
+  redrawn at its own size came back gritty. Taken down to 2 MP, restored, and then taken back up
+  in tiles, it came back clean.
+- Small lettering can come back misspelled.
+- One pass is enough. A second pass over the result at the same size came back crunchy.
+- Don't stack it with my [Consistency LoRA](https://huggingface.co/ausboss/Qwen-Image-2.1-Consistency-LoRA).
+  An earlier run of this LoRA got worse on faces that way, and this one holds the picture by
+  itself.
+- The prompts it was trained on are English.
+
+## Prompt
+
+The first instruction above was on about two thirds of the training pairs. The others say the same thing
+in other words.
+
+About 60 % of the captions also carried ` Scene: ` and a description of the clean picture, so you
+can add one after the instruction. On an earlier run of this LoRA it made no measurable difference,
+and a wrong description did harm: "an old man with a gray beard" painted stubble on a young woman.
+The instruction alone is enough.
+
+## ComfyUI
+
+[`workflows/qwen-image-2.1-restore.json`](workflows/qwen-image-2.1-restore.json) has it set up:
+load your image and press Run. It is one pass at about 2 megapixels, with a Fast switch (8 steps
+with Viggle Turbo, or 25 steps without it). It uses my nodes
+([AusBoss](https://github.com/ausboss/ComfyUI-AusBoss), 2.8.0 or newer) and needs ComfyUI 0.38
+or newer.
+
+The same workflow is on the [Civitai LoRA page](https://civitai.com/models/2989697/qwen-image-21-restore-lora), inside its example pictures.
+
+For a result over 2 megapixels, my [Qwen Image 2.1 Tiled Upscale](https://civitai.com/models/2988629/qwen-image-21-tiled-upscale)
+workflow redraws the picture in tiles with this LoRA (AusBoss nodes 2.9.0 or newer). The pack's
+[Upscale + Restore example](https://github.com/ausboss/ComfyUI-AusBoss/tree/main/example_workflows) is the same graph.
+
+To add the LoRA to your own Qwen Image 2.1 edit graph:
+
+1. A LoRA loader right after the model loader, strength **1.0** to start, **0.5** for a blurry
+   photo.
+2. Resize your picture to about 2 MP, both sides divisible by 32.
+3. **Text Encode Qwen Image 2.1**: the resized picture as `image_1`, `resolution` **0**, and the
+   instruction as the prompt.
+4. **KSampler** on the encoder's `latent` output: CFG 1, `euler` / `simple`, denoise 1. 8 steps
+   with Viggle Turbo at 1.0, or 25 steps without it.
+5. **VAE Decode**, then **Split Image with Alpha** (the Qwen 2.1 VAE decodes RGBA).
+
+## How it was made
+
+- 1,209 pairs from 786 clean pictures. A pair is a clean picture and the same picture damaged the
+  way pictures get damaged: shrunk 3 to 6 times and saved as JPEG, WebP or PNG like a web picture
+  (324 pairs), low-resolution JPEG and PNG, blur and noise, saving and re-saving, phone smoothing,
+  old prints, video stills, banding, and a heavy mix of these.
+- The clean pictures are 661 [Unsplash](https://unsplash.com/license) photos (via
+  [unsplash-lite](https://huggingface.co/datasets/1aurent/unsplash-lite)) and 125 pictures of my
+  own, most of them made with Qwen Image 2.1.
+- Captions: an instruction line, and on 60 % of the pairs ` Scene: ` plus a description of the
+  clean picture written by Qwen3-VL 8B. Caption dropout 0.
+- This is the second run. In the first one every video-still pair sat about 12 px off its clean
+  picture, and the later checkpoints learned to move the picture. For this run those pairs were
+  rebuilt, and every damaged picture was measured against its clean one before training.
+- [ostris/ai-toolkit](https://github.com/ostris/ai-toolkit), `arch: qwen_image_2`, Comfy-Org INT8
+  convrot base, reference kept at the target's size (`match_target_res`), rank 32 / alpha 32,
+  AdamW8bit, LR 1e-4 constant, batch 1, `shift` timesteps, `resolution: 1408` (targets of 1 to 2 MP
+  on the 32 px grid), a checkpoint every 250 steps.
+- I tested every checkpoint from step 1,000 on pictures with a known original. It got closer up to
+  step 1,500 and fell apart by step 1,750 (gritty skin on every face I looked at), so 1,500 is the
+  last good one.
+
+## License
+
+A LoRA for Qwen Image 2.1, which is released under the Qwen Research License; use of the base
+model, and of this LoRA with it, follows that license.

comparison_numbers.csv

@@ -0,0 +1,573 @@
+test,picture,method,ssim,same_face,color_shift_percent,px_off,seconds
+one pass at 2 MP,picture 01,input enlarged the plain way,0.893,0.956,0.33,,
+one pass at 2 MP,picture 01,"Qwen Image 2.1 alone, 8 steps",0.632,0.889,1.84,4.43,10.0
+one pass at 2 MP,picture 01,SeedVR2 3B,0.817,0.938,0.41,0.45,2.8
+one pass at 2 MP,picture 01,"Deblur anything, 8 steps",0.572,0.912,7.42,0.54,10.7
+one pass at 2 MP,picture 01,"Detail Enhancer, same graph, 8 steps",0.701,0.856,2.53,4.03,10.9
+one pass at 2 MP,picture 01,"Detail Enhancer, its own workflow, 25 steps",0.689,0.851,4.48,14.32,58.1
+one pass at 2 MP,picture 01,"Restore LoRA at 0.5, 8 steps",0.706,0.932,1.37,2.75,15.8
+one pass at 2 MP,picture 01,"Restore LoRA at 0.75, 8 steps",0.761,0.934,1.1,1.34,10.8
+one pass at 2 MP,picture 01,"Restore LoRA at 1.0, 8 steps",0.794,0.951,0.92,0.41,48.2
+one pass at 2 MP,picture 01,"Qwen Image 2.1 alone, 25 steps",0.812,0.924,1.47,4.1,22.8
+one pass at 2 MP,picture 01,"the Detail Enhancer's workflow without the LoRA, 25 steps",0.74,0.929,2.49,7.64,43.0
+one pass at 2 MP,picture 01,"Detail Enhancer, same graph, 25 steps",0.799,0.882,2.47,4.91,20.8
+one pass at 2 MP,picture 01,"Restore LoRA at 1.0, 25 steps",0.873,0.96,0.67,0.28,20.8
+one pass at 2 MP,picture 02,input enlarged the plain way,0.899,0.934,0.29,,
+one pass at 2 MP,picture 02,"Qwen Image 2.1 alone, 8 steps",0.626,0.899,1.75,1.29,10.0
+one pass at 2 MP,picture 02,SeedVR2 3B,0.784,0.885,0.37,0.44,2.7
+one pass at 2 MP,picture 02,"Deblur anything, 8 steps",0.643,0.875,1.91,0.48,10.7
+one pass at 2 MP,picture 02,"Detail Enhancer, same graph, 8 steps",0.691,0.864,2.5,1.0,10.7
+one pass at 2 MP,picture 02,"Detail Enhancer, its own workflow, 25 steps",0.758,0.783,3.06,5.19,32.8
+one pass at 2 MP,picture 02,"Restore LoRA at 0.5, 8 steps",0.698,0.93,1.53,0.58,11.0
+one pass at 2 MP,picture 02,"Restore LoRA at 0.75, 8 steps",0.723,0.934,1.24,0.44,10.8
+one pass at 2 MP,picture 02,"Restore LoRA at 1.0, 8 steps",0.737,0.936,1.07,0.36,10.0
+one pass at 2 MP,picture 02,"Qwen Image 2.1 alone, 25 steps",0.838,0.901,1.18,0.9,20.8
+one pass at 2 MP,picture 02,"the Detail Enhancer's workflow without the LoRA, 25 steps",0.78,0.868,1.69,6.97,32.8
+one pass at 2 MP,picture 02,"Detail Enhancer, same graph, 25 steps",0.826,0.833,2.41,0.78,20.8
+one pass at 2 MP,picture 02,"Restore LoRA at 1.0, 25 steps",0.86,0.931,0.73,0.15,20.8
+one pass at 2 MP,picture 03,input enlarged the plain way,0.922,0.928,0.34,,
+one pass at 2 MP,picture 03,"Qwen Image 2.1 alone, 8 steps",0.749,0.795,1.8,2.78,10.0
+one pass at 2 MP,picture 03,SeedVR2 3B,0.875,0.891,0.38,0.35,2.8
+one pass at 2 MP,picture 03,"Deblur anything, 8 steps",0.78,0.866,1.32,0.42,10.7
+one pass at 2 MP,picture 03,"Detail Enhancer, same graph, 8 steps",0.788,0.783,2.82,1.37,10.7
+one pass at 2 MP,picture 03,"Detail Enhancer, its own workflow, 25 steps",0.76,0.752,3.1,8.74,32.8
+one pass at 2 MP,picture 03,"Restore LoRA at 0.5, 8 steps",0.813,0.884,1.11,1.33,10.9
+one pass at 2 MP,picture 03,"Restore LoRA at 0.75, 8 steps",0.84,0.915,0.92,0.36,10.8
+one pass at 2 MP,picture 03,"Restore LoRA at 1.0, 8 steps",0.858,0.916,0.77,0.2,10.0
+one pass at 2 MP,picture 03,"Qwen Image 2.1 alone, 25 steps",0.846,0.863,2.51,2.09,20.8
+one pass at 2 MP,picture 03,"the Detail Enhancer's workflow without the LoRA, 25 steps",0.851,0.837,2.29,3.52,32.9
+one pass at 2 MP,picture 03,"Detail Enhancer, same graph, 25 steps",0.823,0.796,3.54,0.98,20.8
+one pass at 2 MP,picture 03,"Restore LoRA at 1.0, 25 steps",0.903,0.915,0.75,0.11,20.8
+one pass at 2 MP,picture 04,input enlarged the plain way,0.93,0.933,0.29,,
+one pass at 2 MP,picture 04,"Qwen Image 2.1 alone, 8 steps",0.701,0.826,2.13,4.68,10.0
+one pass at 2 MP,picture 04,SeedVR2 3B,0.869,0.832,0.39,0.24,2.8
+one pass at 2 MP,picture 04,"Deblur anything, 8 steps",0.75,0.696,1.52,0.51,10.8
+one pass at 2 MP,picture 04,"Detail Enhancer, same graph, 8 steps",0.716,0.734,5.56,13.8,10.8
+one pass at 2 MP,picture 04,"Detail Enhancer, its own workflow, 25 steps",0.822,0.761,8.69,12.45,49.3
+one pass at 2 MP,picture 04,"Restore LoRA at 0.5, 8 steps",0.762,0.899,1.46,3.34,10.9
+one pass at 2 MP,picture 04,"Restore LoRA at 0.75, 8 steps",0.796,0.921,1.17,2.03,10.8
+one pass at 2 MP,picture 04,"Restore LoRA at 1.0, 8 steps",0.839,0.918,1.02,0.6,10.0
+one pass at 2 MP,picture 04,"Qwen Image 2.1 alone, 25 steps",0.842,0.877,1.31,4.41,20.8
+one pass at 2 MP,picture 04,"the Detail Enhancer's workflow without the LoRA, 25 steps",0.811,0.847,3.16,9.23,49.3
+one pass at 2 MP,picture 04,"Detail Enhancer, same graph, 25 steps",0.713,0.767,7.83,33.29,20.8
+one pass at 2 MP,picture 04,"Restore LoRA at 1.0, 25 steps",0.915,0.937,0.55,0.23,20.8
+one pass at 2 MP,picture 05,input enlarged the plain way,0.822,0.814,0.24,,
+one pass at 2 MP,picture 05,"Qwen Image 2.1 alone, 8 steps",0.686,0.708,1.65,0.89,10.0
+one pass at 2 MP,picture 05,SeedVR2 3B,0.781,0.674,0.4,0.42,2.8
+one pass at 2 MP,picture 05,"Deblur anything, 8 steps",0.754,0.789,0.85,0.21,10.7
+one pass at 2 MP,picture 05,"Detail Enhancer, same graph, 8 steps",0.743,0.727,2.86,1.73,10.8
+one pass at 2 MP,picture 05,"Detail Enhancer, its own workflow, 25 steps",0.507,0.559,4.14,12.32,49.2
+one pass at 2 MP,picture 05,"Restore LoRA at 0.5, 8 steps",0.765,0.814,1.19,0.56,10.8
+one pass at 2 MP,picture 05,"Restore LoRA at 0.75, 8 steps",0.799,0.82,0.98,0.3,10.8
+one pass at 2 MP,picture 05,"Restore LoRA at 1.0, 8 steps",0.813,0.819,0.87,0.27,10.0
+one pass at 2 MP,picture 05,"Qwen Image 2.1 alone, 25 steps",0.804,0.779,1.02,0.52,20.8
+one pass at 2 MP,picture 05,"the Detail Enhancer's workflow without the LoRA, 25 steps",0.586,0.679,3.29,9.91,49.1
+one pass at 2 MP,picture 05,"Detail Enhancer, same graph, 25 steps",0.789,0.754,2.3,1.76,20.8
+one pass at 2 MP,picture 05,"Restore LoRA at 1.0, 25 steps",0.834,0.839,0.67,0.2,20.9
+one pass at 2 MP,picture 06,input enlarged the plain way,0.891,0.922,0.42,,
+one pass at 2 MP,picture 06,"Qwen Image 2.1 alone, 8 steps",0.678,0.852,2.05,2.44,10.0
+one pass at 2 MP,picture 06,SeedVR2 3B,0.837,0.895,0.48,0.36,2.8
+one pass at 2 MP,picture 06,"Deblur anything, 8 steps",0.78,0.869,1.0,0.43,10.7
+one pass at 2 MP,picture 06,"Detail Enhancer, same graph, 8 steps",0.739,0.826,3.55,2.75,10.7
+one pass at 2 MP,picture 06,"Detail Enhancer, its own workflow, 25 steps",0.734,0.769,3.95,7.62,32.8
+one pass at 2 MP,picture 06,"Restore LoRA at 0.5, 8 steps",0.787,0.914,1.26,1.64,10.8
+one pass at 2 MP,picture 06,"Restore LoRA at 0.75, 8 steps",0.833,0.911,0.89,0.5,10.8
+one pass at 2 MP,picture 06,"Restore LoRA at 1.0, 8 steps",0.848,0.929,0.8,0.32,10.0
+one pass at 2 MP,picture 06,"Qwen Image 2.1 alone, 25 steps",0.818,0.853,1.57,2.09,21.6
+one pass at 2 MP,picture 06,"the Detail Enhancer's workflow without the LoRA, 25 steps",0.766,0.872,2.24,5.23,32.8
+one pass at 2 MP,picture 06,"Detail Enhancer, same graph, 25 steps",0.798,0.786,3.47,2.69,20.7
+one pass at 2 MP,picture 06,"Restore LoRA at 1.0, 25 steps",0.884,0.897,0.91,0.15,21.5
+one pass at 2 MP,picture 07,input enlarged the plain way,0.859,0.81,0.38,,
+one pass at 2 MP,picture 07,"Qwen Image 2.1 alone, 8 steps",0.637,0.664,1.67,3.12,10.0
+one pass at 2 MP,picture 07,SeedVR2 3B,0.808,0.629,0.47,0.31,2.8
+one pass at 2 MP,picture 07,"Deblur anything, 8 steps",0.779,0.743,0.83,0.28,10.7
+one pass at 2 MP,picture 07,"Detail Enhancer, same graph, 8 steps",0.652,0.631,2.3,4.59,10.7
+one pass at 2 MP,picture 07,"Detail Enhancer, its own workflow, 25 steps",0.556,0.656,4.38,5.16,38.9
+one pass at 2 MP,picture 07,"Restore LoRA at 0.5, 8 steps",0.764,0.787,1.09,1.84,10.7
+one pass at 2 MP,picture 07,"Restore LoRA at 0.75, 8 steps",0.816,0.795,0.9,0.37,10.7
+one pass at 2 MP,picture 07,"Restore LoRA at 1.0, 8 steps",0.831,0.807,0.81,0.27,10.0
+one pass at 2 MP,picture 07,"Qwen Image 2.1 alone, 25 steps",0.743,0.693,1.91,3.18,23.3
+one pass at 2 MP,picture 07,"the Detail Enhancer's workflow without the LoRA, 25 steps",0.637,0.632,3.43,3.27,38.9
+one pass at 2 MP,picture 07,"Detail Enhancer, same graph, 25 steps",0.667,0.61,2.33,6.24,20.7
+one pass at 2 MP,picture 07,"Restore LoRA at 1.0, 25 steps",0.86,0.792,0.89,0.13,23.9
+one pass at 2 MP,picture 08,input enlarged the plain way,0.767,0.931,0.5,,
+one pass at 2 MP,picture 08,"Qwen Image 2.1 alone, 8 steps",0.53,0.849,2.63,3.97,10.0
+one pass at 2 MP,picture 08,SeedVR2 3B,0.736,0.888,0.63,0.44,2.7
+one pass at 2 MP,picture 08,"Deblur anything, 8 steps",0.62,0.893,1.86,0.88,10.7
+one pass at 2 MP,picture 08,"Detail Enhancer, same graph, 8 steps",0.596,0.801,4.44,1.92,10.7
+one pass at 2 MP,picture 08,"Detail Enhancer, its own workflow, 25 steps",0.595,0.806,4.25,5.81,38.9
+one pass at 2 MP,picture 08,"Restore LoRA at 0.5, 8 steps",0.624,0.882,1.89,3.3,10.7
+one pass at 2 MP,picture 08,"Restore LoRA at 0.75, 8 steps",0.682,0.894,1.54,2.07,10.8
+one pass at 2 MP,picture 08,"Restore LoRA at 1.0, 8 steps",0.72,0.901,1.3,1.29,10.0
+one pass at 2 MP,picture 08,"Qwen Image 2.1 alone, 25 steps",0.681,0.874,1.78,3.91,22.1
+one pass at 2 MP,picture 08,"the Detail Enhancer's workflow without the LoRA, 25 steps",0.616,0.875,2.89,5.23,38.9
+one pass at 2 MP,picture 08,"Detail Enhancer, same graph, 25 steps",0.689,0.81,4.09,1.63,20.7
+one pass at 2 MP,picture 08,"Restore LoRA at 1.0, 25 steps",0.772,0.897,1.05,0.37,21.1
+one pass at 2 MP,picture 09,input enlarged the plain way,0.848,0.911,0.35,,
+one pass at 2 MP,picture 09,"Qwen Image 2.1 alone, 8 steps",0.688,0.86,1.7,1.32,10.0
+one pass at 2 MP,picture 09,SeedVR2 3B,0.821,0.879,0.47,0.52,2.8
+one pass at 2 MP,picture 09,"Deblur anything, 8 steps",0.769,0.905,1.06,0.32,10.8
+one pass at 2 MP,picture 09,"Detail Enhancer, same graph, 8 steps",0.725,0.813,3.04,1.68,10.7
+one pass at 2 MP,picture 09,"Detail Enhancer, its own workflow, 25 steps",0.656,0.762,3.64,25.35,49.2
+one pass at 2 MP,picture 09,"Restore LoRA at 0.5, 8 steps",0.781,0.914,1.04,0.47,10.8
+one pass at 2 MP,picture 09,"Restore LoRA at 0.75, 8 steps",0.811,0.919,0.91,0.27,10.8
+one pass at 2 MP,picture 09,"Restore LoRA at 1.0, 8 steps",0.826,0.921,0.83,0.23,10.0
+one pass at 2 MP,picture 09,"Qwen Image 2.1 alone, 25 steps",0.792,0.869,1.63,0.69,21.1
+one pass at 2 MP,picture 09,"the Detail Enhancer's workflow without the LoRA, 25 steps",0.693,0.871,3.56,8.49,49.1
+one pass at 2 MP,picture 09,"Detail Enhancer, same graph, 25 steps",0.763,0.827,2.28,1.99,20.7
+one pass at 2 MP,picture 09,"Restore LoRA at 1.0, 25 steps",0.842,0.918,0.89,0.1,20.9
+one pass at 2 MP,picture 10,input enlarged the plain way,0.841,0.933,0.27,,
+one pass at 2 MP,picture 10,"Qwen Image 2.1 alone, 8 steps",0.609,0.831,1.71,3.73,10.0
+one pass at 2 MP,picture 10,SeedVR2 3B,0.801,0.903,0.36,0.34,3.0
+one pass at 2 MP,picture 10,"Deblur anything, 8 steps",0.715,0.883,1.31,0.49,10.7
+one pass at 2 MP,picture 10,"Detail Enhancer, same graph, 8 steps",0.673,0.834,2.64,3.04,10.7
+one pass at 2 MP,picture 10,"Detail Enhancer, its own workflow, 25 steps",0.613,0.787,2.85,17.43,55.4
+one pass at 2 MP,picture 10,"Restore LoRA at 0.5, 8 steps",0.71,0.869,1.23,1.72,10.7
+one pass at 2 MP,picture 10,"Restore LoRA at 0.75, 8 steps",0.769,0.883,1.02,0.8,10.8
+one pass at 2 MP,picture 10,"Restore LoRA at 1.0, 8 steps",0.799,0.897,0.87,0.34,10.0
+one pass at 2 MP,picture 10,"Qwen Image 2.1 alone, 25 steps",0.721,0.887,1.95,3.29,20.8
+one pass at 2 MP,picture 10,"the Detail Enhancer's workflow without the LoRA, 25 steps",0.642,0.904,2.48,7.9,55.3
+one pass at 2 MP,picture 10,"Detail Enhancer, same graph, 25 steps",0.718,0.854,2.12,3.08,20.8
+one pass at 2 MP,picture 10,"Restore LoRA at 1.0, 25 steps",0.819,0.933,0.97,0.18,20.8
+one pass at 2 MP,picture 11,input enlarged the plain way,0.875,0.945,0.27,,
+one pass at 2 MP,picture 11,"Qwen Image 2.1 alone, 8 steps",0.645,0.866,1.7,1.13,10.0
+one pass at 2 MP,picture 11,SeedVR2 3B,0.812,0.917,0.33,0.49,2.7
+one pass at 2 MP,picture 11,"Deblur anything, 8 steps",0.582,0.916,2.6,0.49,10.7
+one pass at 2 MP,picture 11,"Detail Enhancer, same graph, 8 steps",0.712,0.85,2.75,1.04,10.7
+one pass at 2 MP,picture 11,"Detail Enhancer, its own workflow, 25 steps",0.666,0.81,3.25,17.73,45.3
+one pass at 2 MP,picture 11,"Restore LoRA at 0.5, 8 steps",0.723,0.915,1.16,0.56,10.8
+one pass at 2 MP,picture 11,"Restore LoRA at 0.75, 8 steps",0.761,0.938,1.0,0.43,10.8
+one pass at 2 MP,picture 11,"Restore LoRA at 1.0, 8 steps",0.787,0.943,0.89,0.34,10.0
+one pass at 2 MP,picture 11,"Qwen Image 2.1 alone, 25 steps",0.809,0.911,1.18,0.73,20.7
+one pass at 2 MP,picture 11,"the Detail Enhancer's workflow without the LoRA, 25 steps",0.704,0.898,1.81,2.41,45.2
+one pass at 2 MP,picture 11,"Detail Enhancer, same graph, 25 steps",0.818,0.881,2.43,0.9,20.8
+one pass at 2 MP,picture 11,"Restore LoRA at 1.0, 25 steps",0.847,0.946,0.65,0.19,20.8
+one pass at 2 MP,picture 12,input enlarged the plain way,0.889,0.851,0.22,,
+one pass at 2 MP,picture 12,"Qwen Image 2.1 alone, 8 steps",0.675,0.762,2.07,1.62,10.0
+one pass at 2 MP,picture 12,SeedVR2 3B,0.798,0.712,0.36,0.44,2.7
+one pass at 2 MP,picture 12,"Deblur anything, 8 steps",0.777,0.825,1.08,0.23,10.7
+one pass at 2 MP,picture 12,"Detail Enhancer, same graph, 8 steps",0.765,0.736,3.43,0.57,10.7
+one pass at 2 MP,picture 12,"Detail Enhancer, its own workflow, 25 steps",0.678,0.508,3.73,5.85,63.4
+one pass at 2 MP,picture 12,"Restore LoRA at 0.5, 8 steps",0.783,0.814,1.25,0.45,10.8
+one pass at 2 MP,picture 12,"Restore LoRA at 0.75, 8 steps",0.816,0.837,1.02,0.27,10.8
+one pass at 2 MP,picture 12,"Restore LoRA at 1.0, 8 steps",0.84,0.838,0.92,0.19,10.0
+one pass at 2 MP,picture 12,"Qwen Image 2.1 alone, 25 steps",0.838,0.755,1.62,0.89,20.8
+one pass at 2 MP,picture 12,"the Detail Enhancer's workflow without the LoRA, 25 steps",0.755,0.657,2.35,4.12,63.4
+one pass at 2 MP,picture 12,"Detail Enhancer, same graph, 25 steps",0.838,0.709,4.0,0.62,20.8
+one pass at 2 MP,picture 12,"Restore LoRA at 1.0, 25 steps",0.892,0.838,0.51,0.11,20.8
+one pass at 2 MP,picture 13,input enlarged the plain way,0.845,0.934,0.35,,
+one pass at 2 MP,picture 13,"Qwen Image 2.1 alone, 8 steps",0.578,0.828,2.17,2.54,10.0
+one pass at 2 MP,picture 13,SeedVR2 3B,0.804,0.843,0.51,0.4,2.7
+one pass at 2 MP,picture 13,"Deblur anything, 8 steps",0.749,0.883,1.25,0.39,10.7
+one pass at 2 MP,picture 13,"Detail Enhancer, same graph, 8 steps",0.66,0.78,2.65,3.52,10.7
+one pass at 2 MP,picture 13,"Detail Enhancer, its own workflow, 25 steps",0.482,0.734,4.65,18.31,51.1
+one pass at 2 MP,picture 13,"Restore LoRA at 0.5, 8 steps",0.727,0.89,1.52,1.31,10.7
+one pass at 2 MP,picture 13,"Restore LoRA at 0.75, 8 steps",0.804,0.91,1.17,0.36,10.7
+one pass at 2 MP,picture 13,"Restore LoRA at 1.0, 8 steps",0.819,0.908,1.08,0.29,10.0
+one pass at 2 MP,picture 13,"Qwen Image 2.1 alone, 25 steps",0.762,0.855,1.45,1.91,20.7
+one pass at 2 MP,picture 13,"the Detail Enhancer's workflow without the LoRA, 25 steps",0.561,0.848,3.3,9.29,51.2
+one pass at 2 MP,picture 13,"Detail Enhancer, same graph, 25 steps",0.669,0.806,2.29,9.11,20.7
+one pass at 2 MP,picture 13,"Restore LoRA at 1.0, 25 steps",0.873,0.922,0.61,0.17,20.7
+one pass at 2 MP,picture 14,input enlarged the plain way,0.688,0.876,0.34,,
+one pass at 2 MP,picture 14,"Qwen Image 2.1 alone, 8 steps",0.48,0.783,1.95,2.01,10.0
+one pass at 2 MP,picture 14,SeedVR2 3B,0.672,0.799,0.56,0.49,2.7
+one pass at 2 MP,picture 14,"Deblur anything, 8 steps",0.673,0.843,0.96,0.27,10.7
+one pass at 2 MP,picture 14,"Detail Enhancer, same graph, 8 steps",0.58,0.742,2.83,1.42,10.7
+one pass at 2 MP,picture 14,"Detail Enhancer, its own workflow, 25 steps",0.382,0.73,5.33,11.73,55.2
+one pass at 2 MP,picture 14,"Restore LoRA at 0.5, 8 steps",0.64,0.851,1.28,1.03,10.7
+one pass at 2 MP,picture 14,"Restore LoRA at 0.75, 8 steps",0.709,0.866,1.27,0.28,10.7
+one pass at 2 MP,picture 14,"Restore LoRA at 1.0, 8 steps",0.72,0.864,1.23,0.21,10.0
+one pass at 2 MP,picture 15,input enlarged the plain way,0.721,0.836,0.35,,
+one pass at 2 MP,picture 15,"Qwen Image 2.1 alone, 8 steps",0.536,0.717,1.71,2.11,10.0
+one pass at 2 MP,picture 15,SeedVR2 3B,0.683,0.717,0.49,0.4,2.7
+one pass at 2 MP,picture 15,"Deblur anything, 8 steps",0.676,0.805,0.96,0.26,10.7
+one pass at 2 MP,picture 15,"Detail Enhancer, same graph, 8 steps",0.612,0.602,1.9,1.93,10.7
+one pass at 2 MP,picture 15,"Detail Enhancer, its own workflow, 25 steps",0.454,0.565,4.78,22.45,47.3
+one pass at 2 MP,picture 15,"Restore LoRA at 0.5, 8 steps",0.657,0.797,1.12,1.23,10.7
+one pass at 2 MP,picture 15,"Restore LoRA at 0.75, 8 steps",0.702,0.831,1.07,0.31,10.7
+one pass at 2 MP,picture 15,"Restore LoRA at 1.0, 8 steps",0.713,0.833,1.03,0.24,10.0
+one pass at 2 MP,picture 16,input enlarged the plain way,0.801,0.967,0.3,,
+one pass at 2 MP,picture 16,"Qwen Image 2.1 alone, 8 steps",0.566,0.911,1.65,4.78,10.0
+one pass at 2 MP,picture 16,SeedVR2 3B,0.764,0.928,0.36,0.5,2.8
+one pass at 2 MP,picture 16,"Deblur anything, 8 steps",0.59,0.926,1.51,0.63,10.7
+one pass at 2 MP,picture 16,"Detail Enhancer, same graph, 8 steps",0.651,0.896,2.22,5.01,10.7
+one pass at 2 MP,picture 16,"Detail Enhancer, its own workflow, 25 steps",0.585,0.835,4.25,61.72,47.2
+one pass at 2 MP,picture 16,"Restore LoRA at 0.5, 8 steps",0.639,0.946,1.22,3.62,10.7
+one pass at 2 MP,picture 16,"Restore LoRA at 0.75, 8 steps",0.7,0.947,1.04,1.11,10.7
+one pass at 2 MP,picture 16,"Restore LoRA at 1.0, 8 steps",0.732,0.952,0.97,0.28,10.0
+one pass at 2 MP,picture 17,input enlarged the plain way,0.784,,0.34,,
+one pass at 2 MP,picture 17,"Qwen Image 2.1 alone, 8 steps",0.575,,1.67,1.74,10.0
+one pass at 2 MP,picture 17,SeedVR2 3B,0.739,,0.44,0.4,3.3
+one pass at 2 MP,picture 17,"Deblur anything, 8 steps",0.699,,0.94,0.24,10.7
+one pass at 2 MP,picture 17,"Detail Enhancer, same graph, 8 steps",0.651,,1.85,2.08,10.7
+one pass at 2 MP,picture 17,"Detail Enhancer, its own workflow, 25 steps",0.531,,3.56,12.53,47.2
+one pass at 2 MP,picture 17,"Restore LoRA at 0.5, 8 steps",0.703,,1.05,0.97,10.7
+one pass at 2 MP,picture 17,"Restore LoRA at 0.75, 8 steps",0.754,,0.89,0.22,10.7
+one pass at 2 MP,picture 17,"Restore LoRA at 1.0, 8 steps",0.772,,0.78,0.16,10.0
+one pass at 2 MP,picture 18,input enlarged the plain way,0.731,0.949,0.32,,
+one pass at 2 MP,picture 18,"Qwen Image 2.1 alone, 8 steps",0.525,0.865,2.37,0.78,10.0
+one pass at 2 MP,picture 18,SeedVR2 3B,0.67,0.893,0.47,0.34,5.3
+one pass at 2 MP,picture 18,"Deblur anything, 8 steps",0.609,0.93,1.57,0.27,10.6
+one pass at 2 MP,picture 18,"Detail Enhancer, same graph, 8 steps",0.543,0.825,3.28,1.77,10.6
+one pass at 2 MP,picture 18,"Detail Enhancer, its own workflow, 25 steps",0.403,0.81,3.9,10.4,55.2
+one pass at 2 MP,picture 18,"Restore LoRA at 0.5, 8 steps",0.636,0.909,1.79,0.44,10.6
+one pass at 2 MP,picture 18,"Restore LoRA at 0.75, 8 steps",0.664,0.922,1.65,0.3,10.7
+one pass at 2 MP,picture 18,"Restore LoRA at 1.0, 8 steps",0.679,0.922,1.54,0.24,10.0
+one pass at 2 MP,picture 19,input enlarged the plain way,0.759,0.913,0.4,,
+one pass at 2 MP,picture 19,"Qwen Image 2.1 alone, 8 steps",0.542,0.83,2.05,3.07,10.0
+one pass at 2 MP,picture 19,SeedVR2 3B,0.725,0.875,0.47,0.36,3.1
+one pass at 2 MP,picture 19,"Deblur anything, 8 steps",0.684,0.892,1.03,0.4,10.7
+one pass at 2 MP,picture 19,"Detail Enhancer, same graph, 8 steps",0.623,0.742,2.88,2.07,10.7
+one pass at 2 MP,picture 19,"Detail Enhancer, its own workflow, 25 steps",0.515,0.726,4.9,38.26,41.0
+one pass at 2 MP,picture 19,"Restore LoRA at 0.5, 8 steps",0.649,0.889,1.37,1.64,10.7
+one pass at 2 MP,picture 19,"Restore LoRA at 0.75, 8 steps",0.698,0.9,1.18,1.01,10.7
+one pass at 2 MP,picture 19,"Restore LoRA at 1.0, 8 steps",0.725,0.897,1.08,0.51,10.0
+one pass at 2 MP,picture 20,input enlarged the plain way,0.82,0.927,0.35,,
+one pass at 2 MP,picture 20,"Qwen Image 2.1 alone, 8 steps",0.644,0.854,2.17,1.98,10.0
+one pass at 2 MP,picture 20,SeedVR2 3B,0.792,0.888,0.41,0.33,2.8
+one pass at 2 MP,picture 20,"Deblur anything, 8 steps",0.701,0.901,1.48,0.57,10.7
+one pass at 2 MP,picture 20,"Detail Enhancer, same graph, 8 steps",0.707,0.802,2.8,0.94,10.7
+one pass at 2 MP,picture 20,"Detail Enhancer, its own workflow, 25 steps",0.644,0.719,3.66,10.69,41.0
+one pass at 2 MP,picture 20,"Restore LoRA at 0.5, 8 steps",0.734,0.919,1.34,1.15,10.7
+one pass at 2 MP,picture 20,"Restore LoRA at 0.75, 8 steps",0.759,0.929,1.0,0.37,10.8
+one pass at 2 MP,picture 20,"Restore LoRA at 1.0, 8 steps",0.773,0.925,0.85,0.24,10.0
+one pass at 2 MP,picture 21,input enlarged the plain way,0.739,0.806,0.49,,
+one pass at 2 MP,picture 21,"Qwen Image 2.1 alone, 8 steps",0.528,0.626,2.0,2.91,10.0
+one pass at 2 MP,picture 21,SeedVR2 3B,0.722,0.726,0.53,0.46,2.8
+one pass at 2 MP,picture 21,"Deblur anything, 8 steps",0.676,0.776,0.98,0.36,10.7
+one pass at 2 MP,picture 21,"Detail Enhancer, same graph, 8 steps",0.606,0.671,2.55,2.36,10.7
+one pass at 2 MP,picture 21,"Detail Enhancer, its own workflow, 25 steps",0.57,0.67,3.23,17.89,34.9
+one pass at 2 MP,picture 21,"Restore LoRA at 0.5, 8 steps",0.658,0.736,1.28,1.59,10.7
+one pass at 2 MP,picture 21,"Restore LoRA at 0.75, 8 steps",0.713,0.787,1.01,0.51,10.7
+one pass at 2 MP,picture 21,"Restore LoRA at 1.0, 8 steps",0.726,0.801,0.93,0.41,10.0
+one pass at 2 MP,picture 22,input enlarged the plain way,0.831,0.841,0.42,,
+one pass at 2 MP,picture 22,"Qwen Image 2.1 alone, 8 steps",0.696,0.696,1.79,1.94,126.1
+one pass at 2 MP,picture 22,SeedVR2 3B,0.82,0.789,0.48,0.3,2.9
+one pass at 2 MP,picture 22,"Deblur anything, 8 steps",0.793,0.786,0.79,0.24,10.7
+one pass at 2 MP,picture 22,"Detail Enhancer, same graph, 8 steps",0.726,0.604,2.55,2.62,10.7
+one pass at 2 MP,picture 22,"Detail Enhancer, its own workflow, 25 steps",0.483,0.583,6.13,23.74,40.9
+one pass at 2 MP,picture 22,"Restore LoRA at 0.5, 8 steps",0.79,0.772,1.09,1.12,10.8
+one pass at 2 MP,picture 22,"Restore LoRA at 0.75, 8 steps",0.823,0.792,0.88,0.34,10.8
+one pass at 2 MP,picture 22,"Restore LoRA at 1.0, 8 steps",0.837,0.8,0.83,0.26,10.0
+one pass at 2 MP,picture 23,input enlarged the plain way,0.799,0.752,0.35,,
+one pass at 2 MP,picture 23,"Qwen Image 2.1 alone, 8 steps",0.603,0.635,1.6,2.07,10.0
+one pass at 2 MP,picture 23,SeedVR2 3B,0.751,0.652,0.46,0.43,2.9
+one pass at 2 MP,picture 23,"Deblur anything, 8 steps",0.729,0.702,0.79,0.4,10.7
+one pass at 2 MP,picture 23,"Detail Enhancer, same graph, 8 steps",0.687,0.658,2.48,4.88,10.7
+one pass at 2 MP,picture 23,"Detail Enhancer, its own workflow, 25 steps",0.61,0.582,5.6,8.91,39.0
+one pass at 2 MP,picture 23,"Restore LoRA at 0.5, 8 steps",0.714,0.728,1.03,1.57,10.7
+one pass at 2 MP,picture 23,"Restore LoRA at 0.75, 8 steps",0.761,0.766,0.87,0.56,10.8
+one pass at 2 MP,picture 23,"Restore LoRA at 1.0, 8 steps",0.777,0.776,0.8,0.38,10.0
+one pass at 2 MP,picture 24,input enlarged the plain way,0.813,0.828,0.33,,
+one pass at 2 MP,picture 24,"Qwen Image 2.1 alone, 8 steps",0.584,0.795,1.82,6.75,10.0
+one pass at 2 MP,picture 24,SeedVR2 3B,0.771,0.812,0.41,0.46,3.1
+one pass at 2 MP,picture 24,"Deblur anything, 8 steps",0.74,0.89,0.87,0.57,10.7
+one pass at 2 MP,picture 24,"Detail Enhancer, same graph, 8 steps",0.68,0.732,3.69,3.52,10.7
+one pass at 2 MP,picture 24,"Detail Enhancer, its own workflow, 25 steps",0.599,0.76,4.27,18.85,41.0
+one pass at 2 MP,picture 24,"Restore LoRA at 0.5, 8 steps",0.729,0.827,0.97,2.49,10.7
+one pass at 2 MP,picture 24,"Restore LoRA at 0.75, 8 steps",0.773,0.855,0.81,0.92,10.7
+one pass at 2 MP,picture 24,"Restore LoRA at 1.0, 8 steps",0.787,0.856,0.75,0.57,10.0
+one pass at 2 MP,picture 25,input enlarged the plain way,0.798,0.732,0.33,,
+one pass at 2 MP,picture 25,"Qwen Image 2.1 alone, 8 steps",0.703,0.651,1.8,1.43,10.0
+one pass at 2 MP,picture 25,SeedVR2 3B,0.797,0.627,0.44,0.33,5.2
+one pass at 2 MP,picture 25,"Deblur anything, 8 steps",0.777,0.693,1.13,0.29,10.7
+one pass at 2 MP,picture 25,"Detail Enhancer, same graph, 8 steps",0.729,0.607,3.61,1.54,10.7
+one pass at 2 MP,picture 25,"Detail Enhancer, its own workflow, 25 steps",0.572,0.471,4.65,19.94,47.3
+one pass at 2 MP,picture 25,"Restore LoRA at 0.5, 8 steps",0.769,0.73,1.14,0.85,10.7
+one pass at 2 MP,picture 25,"Restore LoRA at 0.75, 8 steps",0.804,0.751,0.78,0.25,10.7
+one pass at 2 MP,picture 25,"Restore LoRA at 1.0, 8 steps",0.818,0.756,0.72,0.25,12.0
+one pass at 2 MP,picture 26,input enlarged the plain way,0.834,0.889,0.36,,
+one pass at 2 MP,picture 26,"Qwen Image 2.1 alone, 8 steps",0.659,0.837,1.58,3.51,10.0
+one pass at 2 MP,picture 26,SeedVR2 3B,0.804,0.866,0.41,0.49,5.7
+one pass at 2 MP,picture 26,"Deblur anything, 8 steps",0.745,0.871,1.05,0.54,10.7
+one pass at 2 MP,picture 26,"Detail Enhancer, same graph, 8 steps",0.697,0.788,2.72,3.35,10.7
+one pass at 2 MP,picture 26,"Detail Enhancer, its own workflow, 25 steps",0.611,0.745,4.25,26.98,40.9
+one pass at 2 MP,picture 26,"Restore LoRA at 0.5, 8 steps",0.73,0.874,1.07,2.49,10.7
+one pass at 2 MP,picture 26,"Restore LoRA at 0.75, 8 steps",0.774,0.895,0.86,0.44,10.7
+one pass at 2 MP,picture 26,"Restore LoRA at 1.0, 8 steps",0.791,0.9,0.76,0.32,10.0
+one pass at 2 MP,picture 27,input enlarged the plain way,0.783,0.897,0.34,,
+one pass at 2 MP,picture 27,"Qwen Image 2.1 alone, 8 steps",0.58,0.849,1.56,2.89,10.0
+one pass at 2 MP,picture 27,SeedVR2 3B,0.742,0.807,0.41,0.39,5.5
+one pass at 2 MP,picture 27,"Deblur anything, 8 steps",0.707,0.891,0.9,0.35,10.7
+one pass at 2 MP,picture 27,"Detail Enhancer, same graph, 8 steps",0.671,0.817,2.79,1.38,10.7
+one pass at 2 MP,picture 27,"Detail Enhancer, its own workflow, 25 steps",0.558,0.684,2.46,8.97,47.2
+one pass at 2 MP,picture 27,"Restore LoRA at 0.5, 8 steps",0.699,0.894,0.98,1.52,10.7
+one pass at 2 MP,picture 27,"Restore LoRA at 0.75, 8 steps",0.755,0.894,0.84,0.27,10.7
+one pass at 2 MP,picture 27,"Restore LoRA at 1.0, 8 steps",0.765,0.898,0.79,0.17,20.0
+one pass at 2 MP,picture 28,input enlarged the plain way,0.722,0.862,0.37,,
+one pass at 2 MP,picture 28,"Qwen Image 2.1 alone, 8 steps",0.51,0.725,2.1,2.16,10.0
+one pass at 2 MP,picture 28,SeedVR2 3B,0.666,0.743,0.47,0.51,5.5
+one pass at 2 MP,picture 28,"Deblur anything, 8 steps",0.648,0.854,1.2,0.32,10.7
+one pass at 2 MP,picture 28,"Detail Enhancer, same graph, 8 steps",0.595,0.716,3.38,2.15,10.7
+one pass at 2 MP,picture 28,"Detail Enhancer, its own workflow, 25 steps",0.523,0.676,2.78,8.77,41.0
+one pass at 2 MP,picture 28,"Restore LoRA at 0.5, 8 steps",0.654,0.834,1.37,0.64,10.7
+one pass at 2 MP,picture 28,"Restore LoRA at 0.75, 8 steps",0.685,0.86,1.21,0.31,10.7
+one pass at 2 MP,picture 28,"Restore LoRA at 1.0, 8 steps",0.698,0.864,1.14,0.25,10.0
+one pass at 2 MP,picture 29,input enlarged the plain way,0.732,0.911,0.38,,
+one pass at 2 MP,picture 29,"Qwen Image 2.1 alone, 8 steps",0.55,0.82,1.87,1.83,18.0
+one pass at 2 MP,picture 29,SeedVR2 3B,0.69,0.861,0.48,0.53,5.3
+one pass at 2 MP,picture 29,"Deblur anything, 8 steps",0.654,0.887,1.11,0.33,10.6
+one pass at 2 MP,picture 29,"Detail Enhancer, same graph, 8 steps",0.595,0.796,2.62,1.37,10.6
+one pass at 2 MP,picture 29,"Detail Enhancer, its own workflow, 25 steps",0.554,0.768,2.76,9.78,40.9
+one pass at 2 MP,picture 29,"Restore LoRA at 0.5, 8 steps",0.65,0.867,1.35,1.12,10.7
+one pass at 2 MP,picture 29,"Restore LoRA at 0.75, 8 steps",0.69,0.895,1.1,0.44,10.7
+one pass at 2 MP,picture 29,"Restore LoRA at 1.0, 8 steps",0.704,0.898,0.99,0.32,10.0
+tiles at 4 or 8 MP,"daisies, soft noisy JPEG",input enlarged the plain way,0.753,0.919,0.47,,
+tiles at 4 or 8 MP,"daisies, soft noisy JPEG",SeedVR2 3B,0.796,0.932,0.71,0.15,8.8
+tiles at 4 or 8 MP,"daisies, soft noisy JPEG","Detail Enhancer, its own workflow (to 4 MP)",0.403,0.737,5.07,12.88,63.6
+tiles at 4 or 8 MP,"daisies, soft noisy JPEG","tiles, no LoRA",0.594,0.898,1.14,0.99,45.9
+tiles at 4 or 8 MP,"daisies, soft noisy JPEG","tiles, Detail Enhancer",0.714,0.848,1.65,0.41,46.5
+tiles at 4 or 8 MP,"daisies, soft noisy JPEG","tiles, Restore LoRA",0.789,0.922,0.71,0.14,43.218
+tiles at 4 or 8 MP,"field, half a megapixel",input enlarged the plain way,0.895,0.944,0.4,,
+tiles at 4 or 8 MP,"field, half a megapixel",SeedVR2 3B,0.886,0.862,0.44,0.26,5.4
+tiles at 4 or 8 MP,"field, half a megapixel","Detail Enhancer, its own workflow (to 4 MP)",0.609,0.709,5.63,18.48,63.4
+tiles at 4 or 8 MP,"field, half a megapixel","tiles, no LoRA",0.612,0.849,1.61,1.4,23.3
+tiles at 4 or 8 MP,"field, half a megapixel","tiles, Detail Enhancer",0.731,0.78,2.37,0.58,23.3
+tiles at 4 or 8 MP,"field, half a megapixel","tiles, Restore LoRA",0.877,0.933,0.64,0.16,20.841
+tiles at 4 or 8 MP,"freckles, heavy JPEG",input enlarged the plain way,0.755,0.855,0.97,,
+tiles at 4 or 8 MP,"freckles, heavy JPEG",SeedVR2 3B,0.774,0.851,1.01,0.15,5.8
+tiles at 4 or 8 MP,"freckles, heavy JPEG","Detail Enhancer, its own workflow (to 4 MP)",0.52,0.716,5.4,19.63,63.4
+tiles at 4 or 8 MP,"freckles, heavy JPEG","tiles, no LoRA",0.481,0.78,1.78,0.67,44.1
+tiles at 4 or 8 MP,"freckles, heavy JPEG","tiles, Detail Enhancer",0.616,0.748,1.87,0.46,44.3
+tiles at 4 or 8 MP,"freckles, heavy JPEG","tiles, Restore LoRA",0.719,0.844,1.11,0.24,42.776
+tiles at 4 or 8 MP,"rain, soft noisy JPEG",input enlarged the plain way,0.723,,0.4,,
+tiles at 4 or 8 MP,"rain, soft noisy JPEG",SeedVR2 3B,0.787,,0.43,0.12,8.7
+tiles at 4 or 8 MP,"rain, soft noisy JPEG","Detail Enhancer, its own workflow (to 4 MP)",0.561,,3.28,7.62,63.6
+tiles at 4 or 8 MP,"rain, soft noisy JPEG","tiles, no LoRA",0.492,,1.07,0.89,45.8
+tiles at 4 or 8 MP,"rain, soft noisy JPEG","tiles, Detail Enhancer",0.645,,1.36,1.24,46.2
+tiles at 4 or 8 MP,"rain, soft noisy JPEG","tiles, Restore LoRA",0.766,,0.58,0.15,43.531
+tiles at 4 or 8 MP,"stairwell, heavy JPEG",input enlarged the plain way,0.719,0.708,0.88,,
+tiles at 4 or 8 MP,"stairwell, heavy JPEG",SeedVR2 3B,0.736,0.768,0.89,0.25,5.5
+tiles at 4 or 8 MP,"stairwell, heavy JPEG","Detail Enhancer, its own workflow (to 4 MP)",0.501,0.479,3.43,10.85,63.4
+tiles at 4 or 8 MP,"stairwell, heavy JPEG","tiles, no LoRA",0.544,0.72,1.17,1.04,46.4
+tiles at 4 or 8 MP,"stairwell, heavy JPEG","tiles, Detail Enhancer",0.612,0.682,1.51,1.04,46.5
+tiles at 4 or 8 MP,"stairwell, heavy JPEG","tiles, Restore LoRA",0.732,0.798,0.83,0.52,43.827
+tiles at 4 or 8 MP,"teal portrait, noisy",input enlarged the plain way,0.38,0.794,0.53,,
+tiles at 4 or 8 MP,"teal portrait, noisy",SeedVR2 3B,0.832,0.886,0.66,0.23,8.2
+tiles at 4 or 8 MP,"teal portrait, noisy","Detail Enhancer, its own workflow (to 4 MP)",0.577,0.756,4.22,26.37,63.1
+tiles at 4 or 8 MP,"teal portrait, noisy","tiles, no LoRA",0.644,0.806,0.98,0.59,46.5
+tiles at 4 or 8 MP,"teal portrait, noisy","tiles, Detail Enhancer",0.732,0.775,1.59,0.53,46.6
+tiles at 4 or 8 MP,"teal portrait, noisy","tiles, Restore LoRA",0.823,0.89,0.55,0.16,45.004
+tiles at 4 or 8 MP,"headphones, grainy",input enlarged the plain way,0.37,0.892,0.37,,
+tiles at 4 or 8 MP,"headphones, grainy",SeedVR2 3B,0.877,0.926,0.55,0.22,8.4
+tiles at 4 or 8 MP,"headphones, grainy","Detail Enhancer, its own workflow (to 4 MP)",0.686,0.793,2.45,7.56,82.5
+tiles at 4 or 8 MP,"headphones, grainy","tiles, no LoRA",0.665,0.861,1.37,0.64,46.4
+tiles at 4 or 8 MP,"headphones, grainy","tiles, Detail Enhancer",0.722,0.804,2.44,0.64,46.6
+tiles at 4 or 8 MP,"headphones, grainy","tiles, Restore LoRA",0.844,0.9,0.62,0.09,43.938
+"real photo, against its own input",photo 01,"Qwen Image 2.1 alone, 8 steps",,0.871,12.37,1.65,
+"real photo, against its own input",photo 01,SeedVR2 3B,,0.905,0.63,0.44,
+"real photo, against its own input",photo 01,"Detail Enhancer, 8 steps",,0.829,15.7,1.21,
+"real photo, against its own input",photo 01,"Deblur anything, 8 steps",,0.959,1.1,0.24,
+"real photo, against its own input",photo 01,Restore LoRA at 0.5,,0.918,11.18,0.78,
+"real photo, against its own input",photo 01,Restore LoRA at 0.75,,0.971,1.14,0.28,
+"real photo, against its own input",photo 01,Restore LoRA at 1.0,,0.973,0.76,0.24,
+"real photo, against its own input",photo 02,"Qwen Image 2.1 alone, 8 steps",,0.67,1.76,1.41,
+"real photo, against its own input",photo 02,SeedVR2 3B,,0.601,0.38,0.56,
+"real photo, against its own input",photo 02,"Detail Enhancer, 8 steps",,0.511,5.79,3.51,
+"real photo, against its own input",photo 02,"Deblur anything, 8 steps",,0.93,0.65,0.39,
+"real photo, against its own input",photo 02,Restore LoRA at 0.5,,0.865,0.84,0.86,
+"real photo, against its own input",photo 02,Restore LoRA at 0.75,,0.883,0.84,0.57,
+"real photo, against its own input",photo 02,Restore LoRA at 1.0,,0.84,0.75,0.48,
+"real photo, against its own input",photo 03,"Qwen Image 2.1 alone, 8 steps",,0.831,1.55,1.42,
+"real photo, against its own input",photo 03,SeedVR2 3B,,0.8,0.32,0.34,
+"real photo, against its own input",photo 03,"Detail Enhancer, 8 steps",,0.673,3.55,3.25,
+"real photo, against its own input",photo 03,"Deblur anything, 8 steps",,0.953,0.74,0.3,
+"real photo, against its own input",photo 03,Restore LoRA at 0.5,,0.931,0.91,0.74,
+"real photo, against its own input",photo 03,Restore LoRA at 0.75,,0.935,0.54,0.36,
+"real photo, against its own input",photo 03,Restore LoRA at 1.0,,0.925,0.48,0.3,
+"real photo, against its own input",photo 04,"Qwen Image 2.1 alone, 8 steps",,,2.37,1.3,
+"real photo, against its own input",photo 04,SeedVR2 3B,,,0.47,0.38,
+"real photo, against its own input",photo 04,"Detail Enhancer, 8 steps",,,6.62,3.85,
+"real photo, against its own input",photo 04,"Deblur anything, 8 steps",,,0.79,0.24,
+"real photo, against its own input",photo 04,Restore LoRA at 0.5,,,1.14,0.86,
+"real photo, against its own input",photo 04,Restore LoRA at 0.75,,,0.59,0.31,
+"real photo, against its own input",photo 04,Restore LoRA at 1.0,,,0.54,0.26,
+"real photo, against its own input",photo 05,"Qwen Image 2.1 alone, 8 steps",,0.931,1.91,3.59,
+"real photo, against its own input",photo 05,SeedVR2 3B,,0.968,0.35,0.46,
+"real photo, against its own input",photo 05,"Detail Enhancer, 8 steps",,0.788,4.1,2.84,
+"real photo, against its own input",photo 05,"Deblur anything, 8 steps",,0.978,0.62,0.42,
+"real photo, against its own input",photo 05,Restore LoRA at 0.5,,0.968,1.31,2.57,
+"real photo, against its own input",photo 05,Restore LoRA at 0.75,,0.975,0.95,0.81,
+"real photo, against its own input",photo 05,Restore LoRA at 1.0,,0.977,0.83,0.31,
+"real photo, against its own input",photo 06,"Qwen Image 2.1 alone, 8 steps",,0.809,2.59,2.72,
+"real photo, against its own input",photo 06,SeedVR2 3B,,0.925,0.37,0.22,
+"real photo, against its own input",photo 06,"Detail Enhancer, 8 steps",,0.723,4.49,2.69,
+"real photo, against its own input",photo 06,"Deblur anything, 8 steps",,0.919,0.75,0.51,
+"real photo, against its own input",photo 06,Restore LoRA at 0.5,,0.865,1.33,1.91,
+"real photo, against its own input",photo 06,Restore LoRA at 0.75,,0.876,0.92,1.31,
+"real photo, against its own input",photo 06,Restore LoRA at 1.0,,0.883,0.89,1.36,
+"real photo, against its own input",photo 07,"Qwen Image 2.1 alone, 8 steps",,0.861,3.05,4.27,
+"real photo, against its own input",photo 07,SeedVR2 3B,,0.875,0.44,0.33,
+"real photo, against its own input",photo 07,"Detail Enhancer, 8 steps",,0.79,4.06,1.7,
+"real photo, against its own input",photo 07,"Deblur anything, 8 steps",,0.959,0.89,0.33,
+"real photo, against its own input",photo 07,Restore LoRA at 0.5,,0.926,1.54,3.02,
+"real photo, against its own input",photo 07,Restore LoRA at 0.75,,0.939,0.96,1.61,
+"real photo, against its own input",photo 07,Restore LoRA at 1.0,,0.934,0.77,0.79,
+"real photo, against its own input",photo 08,"Qwen Image 2.1 alone, 8 steps",,0.82,1.95,3.65,
+"real photo, against its own input",photo 08,SeedVR2 3B,,0.675,0.42,0.21,
+"real photo, against its own input",photo 08,"Detail Enhancer, 8 steps",,0.784,11.03,80.02,
+"real photo, against its own input",photo 08,"Deblur anything, 8 steps",,0.901,1.31,0.34,
+"real photo, against its own input",photo 08,Restore LoRA at 0.5,,0.867,1.48,1.97,
+"real photo, against its own input",photo 08,Restore LoRA at 0.75,,0.931,0.82,0.33,
+"real photo, against its own input",photo 08,Restore LoRA at 1.0,,0.943,0.59,0.22,
+"real photo, against its own input",photo 09,"Qwen Image 2.1 alone, 8 steps",,0.889,3.31,3.01,
+"real photo, against its own input",photo 09,SeedVR2 3B,,0.788,0.48,0.22,
+"real photo, against its own input",photo 09,"Detail Enhancer, 8 steps",,0.691,6.58,1.14,
+"real photo, against its own input",photo 09,"Deblur anything, 8 steps",,0.955,1.13,0.25,
+"real photo, against its own input",photo 09,Restore LoRA at 0.5,,0.926,1.9,0.98,
+"real photo, against its own input",photo 09,Restore LoRA at 0.75,,0.947,0.9,0.22,
+"real photo, against its own input",photo 09,Restore LoRA at 1.0,,0.944,0.6,0.14,
+"real photo, against its own input",photo 10,"Qwen Image 2.1 alone, 8 steps",,0.899,6.02,5.41,
+"real photo, against its own input",photo 10,SeedVR2 3B,,0.719,0.7,0.38,
+"real photo, against its own input",photo 10,"Detail Enhancer, 8 steps",,0.857,8.66,5.93,
+"real photo, against its own input",photo 10,"Deblur anything, 8 steps",,0.958,1.07,0.55,
+"real photo, against its own input",photo 10,Restore LoRA at 0.5,,0.952,1.74,1.99,
+"real photo, against its own input",photo 10,Restore LoRA at 0.75,,0.976,0.68,0.21,
+"real photo, against its own input",photo 10,Restore LoRA at 1.0,,0.981,0.58,0.14,
+"real photo, against its own input",photo 11,"Qwen Image 2.1 alone, 8 steps",,0.905,6.84,4.8,
+"real photo, against its own input",photo 11,SeedVR2 3B,,0.728,0.64,0.27,
+"real photo, against its own input",photo 11,"Detail Enhancer, 8 steps",,0.748,10.0,14.8,
+"real photo, against its own input",photo 11,"Deblur anything, 8 steps",,0.899,3.49,0.67,
+"real photo, against its own input",photo 11,Restore LoRA at 0.5,,0.953,3.88,0.65,
+"real photo, against its own input",photo 11,Restore LoRA at 0.75,,0.965,2.16,0.16,
+"real photo, against its own input",photo 11,Restore LoRA at 1.0,,0.973,1.73,0.08,
+"real photo, against its own input",photo 12,"Qwen Image 2.1 alone, 8 steps",,0.819,1.27,1.99,
+"real photo, against its own input",photo 12,SeedVR2 3B,,0.632,0.58,0.26,
+"real photo, against its own input",photo 12,"Detail Enhancer, 8 steps",,0.7,3.04,4.58,
+"real photo, against its own input",photo 12,"Deblur anything, 8 steps",,0.945,0.89,0.35,
+"real photo, against its own input",photo 12,Restore LoRA at 0.5,,0.923,0.83,0.39,
+"real photo, against its own input",photo 12,Restore LoRA at 0.75,,0.943,0.62,0.25,
+"real photo, against its own input",photo 12,Restore LoRA at 1.0,,0.948,0.45,0.14,
+"real photo, against its own input",photo 13,"Qwen Image 2.1 alone, 8 steps",,0.895,1.47,4.9,
+"real photo, against its own input",photo 13,SeedVR2 3B,,0.827,0.48,0.12,
+"real photo, against its own input",photo 13,"Detail Enhancer, 8 steps",,0.689,7.49,12.41,
+"real photo, against its own input",photo 13,"Deblur anything, 8 steps",,0.965,0.82,0.47,
+"real photo, against its own input",photo 13,Restore LoRA at 0.5,,0.944,1.14,3.86,
+"real photo, against its own input",photo 13,Restore LoRA at 0.75,,0.964,0.83,2.2,
+"real photo, against its own input",photo 13,Restore LoRA at 1.0,,0.973,0.6,0.46,
+"real photo, against its own input",photo 14,"Qwen Image 2.1 alone, 8 steps",,0.809,2.08,3.1,
+"real photo, against its own input",photo 14,SeedVR2 3B,,0.768,0.51,0.23,
+"real photo, against its own input",photo 14,"Detail Enhancer, 8 steps",,0.744,3.0,1.39,
+"real photo, against its own input",photo 14,"Deblur anything, 8 steps",,0.941,0.55,0.13,
+"real photo, against its own input",photo 14,Restore LoRA at 0.5,,0.872,1.11,1.18,
+"real photo, against its own input",photo 14,Restore LoRA at 0.75,,0.92,0.69,0.29,
+"real photo, against its own input",photo 14,Restore LoRA at 1.0,,0.94,0.65,0.17,
+"real photo, against its own input",photo 15,"Qwen Image 2.1 alone, 8 steps",,0.938,3.27,2.46,
+"real photo, against its own input",photo 15,SeedVR2 3B,,0.746,0.88,0.16,
+"real photo, against its own input",photo 15,"Detail Enhancer, 8 steps",,0.845,5.2,2.36,
+"real photo, against its own input",photo 15,"Deblur anything, 8 steps",,0.954,1.74,0.49,
+"real photo, against its own input",photo 15,Restore LoRA at 0.5,,0.957,1.78,0.59,
+"real photo, against its own input",photo 15,Restore LoRA at 0.75,,0.976,1.28,0.24,
+"real photo, against its own input",photo 15,Restore LoRA at 1.0,,0.977,0.85,0.14,
+"real photo, against its own input",photo 16,"Qwen Image 2.1 alone, 8 steps",,0.907,1.69,2.41,
+"real photo, against its own input",photo 16,SeedVR2 3B,,0.823,0.53,0.27,
+"real photo, against its own input",photo 16,"Detail Enhancer, 8 steps",,0.832,3.14,2.68,
+"real photo, against its own input",photo 16,"Deblur anything, 8 steps",,0.926,1.39,0.27,
+"real photo, against its own input",photo 16,Restore LoRA at 0.5,,0.921,1.11,0.48,
+"real photo, against its own input",photo 16,Restore LoRA at 0.75,,0.933,0.95,0.22,
+"real photo, against its own input",photo 16,Restore LoRA at 1.0,,0.932,0.93,0.2,
+"real photo, against its own input",photo 17,"Qwen Image 2.1 alone, 8 steps",,0.729,2.54,5.15,
+"real photo, against its own input",photo 17,SeedVR2 3B,,0.558,0.69,0.28,
+"real photo, against its own input",photo 17,"Detail Enhancer, 8 steps",,0.647,6.87,7.47,
+"real photo, against its own input",photo 17,"Deblur anything, 8 steps",,0.917,0.89,0.38,
+"real photo, against its own input",photo 17,Restore LoRA at 0.5,,0.87,1.55,2.98,
+"real photo, against its own input",photo 17,Restore LoRA at 0.75,,0.934,0.76,1.44,
+"real photo, against its own input",photo 17,Restore LoRA at 1.0,,0.932,0.64,0.46,
+"real photo, against its own input",photo 18,"Qwen Image 2.1 alone, 8 steps",,0.861,3.53,5.76,
+"real photo, against its own input",photo 18,SeedVR2 3B,,0.769,0.68,0.26,
+"real photo, against its own input",photo 18,"Detail Enhancer, 8 steps",,0.655,13.12,4.76,
+"real photo, against its own input",photo 18,"Deblur anything, 8 steps",,0.955,1.9,0.41,
+"real photo, against its own input",photo 18,Restore LoRA at 0.5,,0.929,1.36,0.8,
+"real photo, against its own input",photo 18,Restore LoRA at 0.75,,0.966,0.78,0.17,
+"real photo, against its own input",photo 18,Restore LoRA at 1.0,,0.966,0.78,0.15,
+"real photo, against its own input",photo 19,"Qwen Image 2.1 alone, 8 steps",,0.853,3.26,3.8,
+"real photo, against its own input",photo 19,SeedVR2 3B,,0.841,0.51,0.15,
+"real photo, against its own input",photo 19,"Detail Enhancer, 8 steps",,0.718,6.62,2.0,
+"real photo, against its own input",photo 19,"Deblur anything, 8 steps",,0.956,0.77,0.31,
+"real photo, against its own input",photo 19,Restore LoRA at 0.5,,0.907,1.45,1.3,
+"real photo, against its own input",photo 19,Restore LoRA at 0.75,,0.932,0.89,0.41,
+"real photo, against its own input",photo 19,Restore LoRA at 1.0,,0.959,0.71,0.25,
+"real photo, against its own input",photo 20,"Qwen Image 2.1 alone, 8 steps",,0.892,10.04,3.3,
+"real photo, against its own input",photo 20,SeedVR2 3B,,0.783,0.6,0.28,
+"real photo, against its own input",photo 20,"Detail Enhancer, 8 steps",,0.765,18.2,2.33,
+"real photo, against its own input",photo 20,"Deblur anything, 8 steps",,0.941,1.28,0.24,
+"real photo, against its own input",photo 20,Restore LoRA at 0.5,,0.949,6.92,0.61,
+"real photo, against its own input",photo 20,Restore LoRA at 0.75,,0.965,4.18,0.17,
+"real photo, against its own input",photo 20,Restore LoRA at 1.0,,0.975,0.87,0.12,
+"real photo, against its own input",photo 21,"Qwen Image 2.1 alone, 8 steps",,0.628,2.01,4.71,
+"real photo, against its own input",photo 21,SeedVR2 3B,,0.731,0.53,0.29,
+"real photo, against its own input",photo 21,"Detail Enhancer, 8 steps",,0.58,3.84,3.49,
+"real photo, against its own input",photo 21,"Deblur anything, 8 steps",,0.905,0.69,0.32,
+"real photo, against its own input",photo 21,Restore LoRA at 0.5,,0.79,1.22,2.08,
+"real photo, against its own input",photo 21,Restore LoRA at 0.75,,0.885,0.89,0.93,
+"real photo, against its own input",photo 21,Restore LoRA at 1.0,,0.886,0.84,0.33,
+"real photo, against its own input",photo 22,"Qwen Image 2.1 alone, 8 steps",,0.89,10.13,4.69,
+"real photo, against its own input",photo 22,SeedVR2 3B,,0.875,0.45,0.42,
+"real photo, against its own input",photo 22,"Detail Enhancer, 8 steps",,0.834,18.02,2.85,
+"real photo, against its own input",photo 22,"Deblur anything, 8 steps",,0.939,1.38,0.68,
+"real photo, against its own input",photo 22,Restore LoRA at 0.5,,0.938,5.47,3.02,
+"real photo, against its own input",photo 22,Restore LoRA at 0.75,,0.962,1.63,0.7,
+"real photo, against its own input",photo 22,Restore LoRA at 1.0,,0.975,0.86,0.26,
+"real photo, against its own input",photo 23,"Qwen Image 2.1 alone, 8 steps",,0.902,1.56,2.4,
+"real photo, against its own input",photo 23,SeedVR2 3B,,0.891,0.5,0.18,
+"real photo, against its own input",photo 23,"Detail Enhancer, 8 steps",,0.831,4.22,2.51,
+"real photo, against its own input",photo 23,"Deblur anything, 8 steps",,0.971,1.19,0.3,
+"real photo, against its own input",photo 23,Restore LoRA at 0.5,,0.961,1.04,0.62,
+"real photo, against its own input",photo 23,Restore LoRA at 0.75,,0.972,0.91,0.31,
+"real photo, against its own input",photo 23,Restore LoRA at 1.0,,0.977,0.94,0.23,
+"real photo, against its own input",photo 24,"Qwen Image 2.1 alone, 8 steps",,0.879,2.06,1.48,
+"real photo, against its own input",photo 24,SeedVR2 3B,,0.74,0.64,0.17,
+"real photo, against its own input",photo 24,"Detail Enhancer, 8 steps",,0.766,5.85,22.43,
+"real photo, against its own input",photo 24,"Deblur anything, 8 steps",,0.922,1.31,0.32,
+"real photo, against its own input",photo 24,Restore LoRA at 0.5,,0.939,1.33,0.43,
+"real photo, against its own input",photo 24,Restore LoRA at 0.75,,0.965,0.87,0.18,
+"real photo, against its own input",photo 24,Restore LoRA at 1.0,,0.974,0.66,0.07,
+"real photo, against its own input",photo 25,"Qwen Image 2.1 alone, 8 steps",,0.753,3.87,4.77,
+"real photo, against its own input",photo 25,SeedVR2 3B,,0.827,0.71,0.37,
+"real photo, against its own input",photo 25,"Detail Enhancer, 8 steps",,0.647,9.54,1.26,
+"real photo, against its own input",photo 25,"Deblur anything, 8 steps",,0.904,1.19,0.53,
+"real photo, against its own input",photo 25,Restore LoRA at 0.5,,0.805,2.64,3.99,
+"real photo, against its own input",photo 25,Restore LoRA at 0.75,,0.865,1.56,0.95,
+"real photo, against its own input",photo 25,Restore LoRA at 1.0,,0.857,1.53,0.6,
+"real photo, against its own input",photo 26,"Qwen Image 2.1 alone, 8 steps",,,1.87,0.69,
+"real photo, against its own input",photo 26,SeedVR2 3B,,,0.9,0.54,
+"real photo, against its own input",photo 26,"Detail Enhancer, 8 steps",,,5.27,1.04,
+"real photo, against its own input",photo 26,"Deblur anything, 8 steps",,,1.16,0.22,
+"real photo, against its own input",photo 26,Restore LoRA at 0.5,,,1.35,0.45,
+"real photo, against its own input",photo 26,Restore LoRA at 0.75,,,1.31,0.52,
+"real photo, against its own input",photo 26,Restore LoRA at 1.0,,,1.25,0.47,
+"real photo, against its own input",photo 27,"Qwen Image 2.1 alone, 8 steps",,,3.7,2.43,
+"real photo, against its own input",photo 27,SeedVR2 3B,,,0.31,0.67,
+"real photo, against its own input",photo 27,"Detail Enhancer, 8 steps",,,7.55,1.88,
+"real photo, against its own input",photo 27,"Deblur anything, 8 steps",,,0.53,0.46,
+"real photo, against its own input",photo 27,Restore LoRA at 0.5,,,2.54,1.23,
+"real photo, against its own input",photo 27,Restore LoRA at 0.75,,,1.05,0.82,
+"real photo, against its own input",photo 27,Restore LoRA at 1.0,,,0.95,0.59,
+"real photo, against its own input",photo 28,"Qwen Image 2.1 alone, 8 steps",,0.927,3.73,2.53,
+"real photo, against its own input",photo 28,SeedVR2 3B,,0.932,0.67,0.34,
+"real photo, against its own input",photo 28,"Detail Enhancer, 8 steps",,0.714,4.28,2.18,
+"real photo, against its own input",photo 28,"Deblur anything, 8 steps",,0.939,1.45,0.33,
+"real photo, against its own input",photo 28,Restore LoRA at 0.5,,0.95,3.31,2.57,
+"real photo, against its own input",photo 28,Restore LoRA at 0.75,,0.964,2.14,1.27,
+"real photo, against its own input",photo 28,Restore LoRA at 1.0,,0.973,1.87,0.42,
+"real photo, against its own input",photo 29,"Qwen Image 2.1 alone, 8 steps",,,8.72,1.69,
+"real photo, against its own input",photo 29,SeedVR2 3B,,,0.34,0.38,
+"real photo, against its own input",photo 29,"Detail Enhancer, 8 steps",,,5.75,5.78,
+"real photo, against its own input",photo 29,"Deblur anything, 8 steps",,,0.59,0.2,
+"real photo, against its own input",photo 29,Restore LoRA at 0.5,,,1.65,0.74,
+"real photo, against its own input",photo 29,Restore LoRA at 0.75,,,1.18,0.21,
+"real photo, against its own input",photo 29,Restore LoRA at 1.0,,,1.22,0.18,
+"real photo, against its own input",photo 30,"Qwen Image 2.1 alone, 8 steps",,,7.63,2.09,
+"real photo, against its own input",photo 30,SeedVR2 3B,,,0.37,0.3,
+"real photo, against its own input",photo 30,"Detail Enhancer, 8 steps",,,9.11,4.96,
+"real photo, against its own input",photo 30,"Deblur anything, 8 steps",,,0.6,0.35,
+"real photo, against its own input",photo 30,Restore LoRA at 0.5,,,0.94,1.38,
+"real photo, against its own input",photo 30,Restore LoRA at 0.75,,,0.72,0.56,
+"real photo, against its own input",photo 30,Restore LoRA at 1.0,,,0.72,0.27,
+"real photo, against its own input",photo 31,"Qwen Image 2.1 alone, 8 steps",,0.875,3.7,1.44,
+"real photo, against its own input",photo 31,SeedVR2 3B,,0.917,0.4,0.24,
+"real photo, against its own input",photo 31,"Detail Enhancer, 8 steps",,0.585,6.55,17.92,
+"real photo, against its own input",photo 31,"Deblur anything, 8 steps",,0.956,0.51,0.2,
+"real photo, against its own input",photo 31,Restore LoRA at 0.5,,0.929,3.62,0.59,
+"real photo, against its own input",photo 31,Restore LoRA at 0.75,,0.948,1.51,0.32,
+"real photo, against its own input",photo 31,Restore LoRA at 1.0,,0.948,1.37,0.17,

qwen-image-2.1-restore-1250.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e48f68c7772a04ca160396aeb1e98f77e9d7a7b6353ae6fc3fb3e69746508b42
+size 159436584

qwen-image-2.1-restore.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:65de4f6c2468071d5aa6ffbd82c76bfee7cc8ac28fe960e91bbca4a58f43c08c
+size 159436584

workflows/qwen-image-2.1-restore.json

@@ -0,0 +1,2548 @@
+{
+ "id": "ea98369b-a6f3-4714-b34f-3b877fbe9017",
+ "revision": 0,
+ "last_node_id": 33,
+ "last_link_id": 31,
+ "nodes": [
+  {
+   "id": 1,
+   "type": "AUSBOSS_NODES_WorkflowNote",
+   "pos": [
+    0,
+    30
+   ],
+   "size": [
+    440,
+    1050
+   ],
+   "flags": {},
+   "order": 0,
+   "mode": 0,
+   "inputs": [
+    {
+     "localized_name": "note",
+     "name": "note",
+     "type": "STRING",
+     "widget": {
+      "name": "note"
+     },
+     "link": null
+    }
+   ],
+   "outputs": [],
+   "title": "🆎 AusBoss",
+   "properties": {
+    "Node name for S&R": "AUSBOSS_NODES_WorkflowNote",
+    "cnr_id": "ausboss-nodes",
+    "ver": "2.8.0"
+   },
+   "widgets_values": [
+    "{\"v\": 1, \"title\": \"Qwen Image 2.1 Restore\", \"subtitle\": \"Load a small, soft or worn picture. Qwen Image 2.1 redraws it sharp at about 2 megapixels with my Restore LoRA.\", \"author\": \"AusBoss\", \"layout\": \"card\", \"accent\": \"\", \"body\": \"1. Load your image. You can crop it right there with the blue handles.\\n2. Press **Run**.\\n3. Slide **Before and after** to compare. The result is saved as a PNG.\\n\\n### Switches\\n\\n- **Fast** on: 8 steps with the Viggle Turbo LoRA. Off: 25 steps without it, closer to your picture and softer, about twice as long.\\n\\n### Tips\\n\\n- **Blurry photo (out of focus, shaky, a frame from a video)? Lower the Restore LoRA to 0.5.** At 1.0 it stays true to the blur and the result comes back soft.\\n- Strength is a dial: 1.0 is the softest and stays truest to your picture. Lower it for more of Qwen's own texture.\\n- It is a redraw. Fine detail is made up to fit, so a face stays the same person, not the same pixels.\\n- Small faces are redrawn, so a face far from the camera can come back as a slightly different person.\\n- Small lettering on signs and labels can come back misspelled.\\n- For a result over 2 megapixels, use my tiled workflow, Qwen Image 2.1 Tiled Upscale.\\n- If the run stops at a red **LoRA Loader**, a LoRA that is switched on is missing. Download it from the list below.\\n\\nNeeds ComfyUI 0.38 or newer and AusBoss nodes 2.8.0 or newer (update first).\", \"models\": [{\"name\": \"qwen_image_2.1_int8_convrot.safetensors\", \"dir\": \"diffusion_models\", \"size\": \"7.26 GB\", \"url\": \"https://huggingface.co/Comfy-Org/Qwen-Image-2.1/resolve/main/diffusion_models/qwen_image_2.1_int8_convrot.safetensors\", \"note\": \"\"}, {\"name\": \"qwen3vl_8b_int8_convrot.safetensors\", \"dir\": \"text_encoders\", \"size\": \"9.35 GB\", \"url\": \"https://huggingface.co/Comfy-Org/Qwen-Image-2.1/resolve/main/text_encoders/qwen3vl_8b_int8_convrot.safetensors\", \"note\": \"\"}, {\"name\": \"qwen_image_2.1_vae_bf16.safetensors\", \"dir\": \"vae\", \"size\": \"676 MB\", \"url\": \"https://huggingface.co/Comfy-Org/Qwen-Image-2.1/resolve/main/vae/qwen_image_2.1_vae_bf16.safetensors\", \"note\": \"\"}, {\"name\": \"qwen-image-2.1-restore.safetensors\", \"dir\": \"loras\", \"size\": \"159 MB\", \"url\": \"https://huggingface.co/ausboss/Qwen-Image-2.1-Restore-LoRA/resolve/main/qwen-image-2.1-restore.safetensors\", \"note\": \"keeps the result close to your picture\"}, {\"name\": \"Qwen-Image-2.1-viggle-turbo-v0.3-6step-lora-r128.safetensors\", \"dir\": \"loras\", \"size\": \"680 MB\", \"url\": \"https://huggingface.co/Viggle/Qwen-Image-2.1-viggle-turbo/resolve/main/Qwen-Image-2.1-viggle-turbo-v0.3-6step-lora-r128.safetensors\", \"note\": \"for Fast on · Qwen Research License, research use\"}], \"packs\": [{\"name\": \"ComfyUI-AusBoss\", \"url\": \"https://github.com/ausboss/ComfyUI-AusBoss\", \"node\": \"AUSBOSS_NODES_Callout\"}], \"links\": [{\"label\": \"GitHub · ausboss\", \"url\": \"https://github.com/ausboss\"}, {\"label\": \"Civitai · AusBoss\", \"url\": \"https://civitai.com/user/AusBoss\"}, {\"label\": \"X · Zanzibased\", \"url\": \"https://x.com/Zanzibased\"}]}"
+   ],
+   "widgets_values_named": {
+    "note": "{\"v\": 1, \"title\": \"Qwen Image 2.1 Restore\", \"subtitle\": \"Load a small, soft or worn picture. Qwen Image 2.1 redraws it sharp at about 2 megapixels with my Restore LoRA.\", \"author\": \"AusBoss\", \"layout\": \"card\", \"accent\": \"\", \"body\": \"1. Load your image. You can crop it right there with the blue handles.\\n2. Press **Run**.\\n3. Slide **Before and after** to compare. The result is saved as a PNG.\\n\\n### Switches\\n\\n- **Fast** on: 8 steps with the Viggle Turbo LoRA. Off: 25 steps without it, closer to your picture and softer, about twice as long.\\n\\n### Tips\\n\\n- **Blurry photo (out of focus, shaky, a frame from a video)? Lower the Restore LoRA to 0.5.** At 1.0 it stays true to the blur and the result comes back soft.\\n- Strength is a dial: 1.0 is the softest and stays truest to your picture. Lower it for more of Qwen's own texture.\\n- It is a redraw. Fine detail is made up to fit, so a face stays the same person, not the same pixels.\\n- Small faces are redrawn, so a face far from the camera can come back as a slightly different person.\\n- Small lettering on signs and labels can come back misspelled.\\n- For a result over 2 megapixels, use my tiled workflow, Qwen Image 2.1 Tiled Upscale.\\n- If the run stops at a red **LoRA Loader**, a LoRA that is switched on is missing. Download it from the list below.\\n\\nNeeds ComfyUI 0.38 or newer and AusBoss nodes 2.8.0 or newer (update first).\", \"models\": [{\"name\": \"qwen_image_2.1_int8_convrot.safetensors\", \"dir\": \"diffusion_models\", \"size\": \"7.26 GB\", \"url\": \"https://huggingface.co/Comfy-Org/Qwen-Image-2.1/resolve/main/diffusion_models/qwen_image_2.1_int8_convrot.safetensors\", \"note\": \"\"}, {\"name\": \"qwen3vl_8b_int8_convrot.safetensors\", \"dir\": \"text_encoders\", \"size\": \"9.35 GB\", \"url\": \"https://huggingface.co/Comfy-Org/Qwen-Image-2.1/resolve/main/text_encoders/qwen3vl_8b_int8_convrot.safetensors\", \"note\": \"\"}, {\"name\": \"qwen_image_2.1_vae_bf16.safetensors\", \"dir\": \"vae\", \"size\": \"676 MB\", \"url\": \"https://huggingface.co/Comfy-Org/Qwen-Image-2.1/resolve/main/vae/qwen_image_2.1_vae_bf16.safetensors\", \"note\": \"\"}, {\"name\": \"qwen-image-2.1-restore.safetensors\", \"dir\": \"loras\", \"size\": \"159 MB\", \"url\": \"https://huggingface.co/ausboss/Qwen-Image-2.1-Restore-LoRA/resolve/main/qwen-image-2.1-restore.safetensors\", \"note\": \"keeps the result close to your picture\"}, {\"name\": \"Qwen-Image-2.1-viggle-turbo-v0.3-6step-lora-r128.safetensors\", \"dir\": \"loras\", \"size\": \"680 MB\", \"url\": \"https://huggingface.co/Viggle/Qwen-Image-2.1-viggle-turbo/resolve/main/Qwen-Image-2.1-viggle-turbo-v0.3-6step-lora-r128.safetensors\", \"note\": \"for Fast on · Qwen Research License, research use\"}], \"packs\": [{\"name\": \"ComfyUI-AusBoss\", \"url\": \"https://github.com/ausboss/ComfyUI-AusBoss\", \"node\": \"AUSBOSS_NODES_Callout\"}], \"links\": [{\"label\": \"GitHub · ausboss\", \"url\": \"https://github.com/ausboss\"}, {\"label\": \"Civitai · AusBoss\", \"url\": \"https://civitai.com/user/AusBoss\"}, {\"label\": \"X · Zanzibased\", \"url\": \"https://x.com/Zanzibased\"}]}"
+   },
+   "color": "#14424d",
+   "bgcolor": "#161f21"
+  },
+  {
+   "id": 2,
+   "type": "AUSBOSS_NODES_ImageCropRotatePad",
+   "pos": [
+    500,
+    286.0
+   ],
+   "size": [
+    424,
+    1054
+   ],
+   "flags": {},
+   "order": 2,
+   "mode": 0,
+   "inputs": [
+    {
+     "localized_name": "image",
+     "name": "image",
+     "type": "COMBO",
+     "widget": {
+      "name": "image"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "rotation_degrees",
+     "name": "rotation_degrees",
+     "type": "FLOAT",
+     "widget": {
+      "name": "rotation_degrees"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "crop_aspect_ratio",
+     "name": "crop_aspect_ratio",
+     "type": "COMBO",
+     "widget": {
+      "name": "crop_aspect_ratio"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "crop_x",
+     "name": "crop_x",
+     "type": "INT",
+     "widget": {
+      "name": "crop_x"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "crop_y",
+     "name": "crop_y",
+     "type": "INT",
+     "widget": {
+      "name": "crop_y"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "crop_width",
+     "name": "crop_width",
+     "type": "INT",
+     "widget": {
+      "name": "crop_width"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "crop_height",
+     "name": "crop_height",
+     "type": "INT",
+     "widget": {
+      "name": "crop_height"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "pad_left",
+     "name": "pad_left",
+     "type": "INT",
+     "widget": {
+      "name": "pad_left"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "pad_top",
+     "name": "pad_top",
+     "type": "INT",
+     "widget": {
+      "name": "pad_top"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "pad_right",
+     "name": "pad_right",
+     "type": "INT",
+     "widget": {
+      "name": "pad_right"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "pad_bottom",
+     "name": "pad_bottom",
+     "type": "INT",
+     "widget": {
+      "name": "pad_bottom"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "feather",
+     "name": "feather",
+     "type": "INT",
+     "widget": {
+      "name": "feather"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "canvas_multiple",
+     "name": "canvas_multiple",
+     "type": "INT",
+     "widget": {
+      "name": "canvas_multiple"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "fill_color",
+     "name": "fill_color",
+     "type": "STRING",
+     "widget": {
+      "name": "fill_color"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "resize_to_megapixels",
+     "name": "resize_to_megapixels",
+     "shape": 7,
+     "type": "BOOLEAN",
+     "widget": {
+      "name": "resize_to_megapixels"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "megapixels",
+     "name": "megapixels",
+     "shape": 7,
+     "type": "FLOAT",
+     "widget": {
+      "name": "megapixels"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "resize_method",
+     "name": "resize_method",
+     "shape": 7,
+     "type": "COMBO",
+     "widget": {
+      "name": "resize_method"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "resolution_steps",
+     "name": "resolution_steps",
+     "shape": 7,
+     "type": "INT",
+     "widget": {
+      "name": "resolution_steps"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "stitch_blend",
+     "name": "stitch_blend",
+     "shape": 7,
+     "type": "INT",
+     "widget": {
+      "name": "stitch_blend"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "stitch_grow",
+     "name": "stitch_grow",
+     "shape": 7,
+     "type": "INT",
+     "widget": {
+      "name": "stitch_grow"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "upload",
+     "name": "upload",
+     "type": "IMAGEUPLOAD",
+     "widget": {
+      "name": "upload"
+     },
+     "link": null
+    }
+   ],
+   "outputs": [
+    {
+     "localized_name": "image",
+     "name": "image",
+     "type": "IMAGE",
+     "links": [
+      21,
+      26
+     ]
+    },
+    {
+     "localized_name": "mask",
+     "name": "mask",
+     "type": "MASK",
+     "links": null
+    },
+    {
+     "localized_name": "stitcher",
+     "name": "stitcher",
+     "type": "AUSBOSS_STITCHER",
+     "links": null
+    },
+    {
+     "localized_name": "original",
+     "name": "original",
+     "type": "IMAGE",
+     "links": null
+    },
+    {
+     "localized_name": "width",
+     "name": "width",
+     "type": "INT",
+     "links": null
+    },
+    {
+     "localized_name": "height",
+     "name": "height",
+     "type": "INT",
+     "links": null
+    },
+    {
+     "localized_name": "prompt_image",
+     "name": "prompt_image",
+     "type": "IMAGE",
+     "links": null
+    }
+   ],
+   "title": "Image Crop + Rotate + Pad - Your picture",
+   "properties": {
+    "Node name for S&R": "AUSBOSS_NODES_ImageCropRotatePad",
+    "cnr_id": "ausboss-nodes",
+    "ver": "2.8.0"
+   },
+   "widgets_values": {
+    "image": "",
+    "rotation_degrees": 0,
+    "crop_aspect_ratio": "free",
+    "crop_x": 0,
+    "crop_y": 0,
+    "crop_width": 0,
+    "crop_height": 0,
+    "pad_left": 0,
+    "pad_top": 0,
+    "pad_right": 0,
+    "pad_bottom": 0,
+    "feather": 24,
+    "canvas_multiple": 1,
+    "fill_color": "#808080",
+    "resize_to_megapixels": true,
+    "megapixels": 2,
+    "resize_method": "lanczos",
+    "resolution_steps": 32,
+    "stitch_blend": 32,
+    "stitch_grow": 0
+   },
+   "widgets_values_named": {
+    "image": "",
+    "rotation_degrees": 0,
+    "crop_aspect_ratio": "free",
+    "crop_x": 0,
+    "crop_y": 0,
+    "crop_width": 0,
+    "crop_height": 0,
+    "pad_left": 0,
+    "pad_top": 0,
+    "pad_right": 0,
+    "pad_bottom": 0,
+    "feather": 24,
+    "canvas_multiple": 1,
+    "fill_color": "#808080",
+    "resize_to_megapixels": true,
+    "megapixels": 2,
+    "resize_method": "lanczos",
+    "resolution_steps": 32,
+    "stitch_blend": 32,
+    "stitch_grow": 0
+   },
+   "color": "#14424d",
+   "bgcolor": "#161f21"
+  },
+  {
+   "id": 3,
+   "type": "AUSBOSS_NODES_Text",
+   "pos": [
+    964,
+    286.0
+   ],
+   "size": [
+    424,
+    236
+   ],
+   "flags": {},
+   "order": 3,
+   "mode": 0,
+   "inputs": [
+    {
+     "localized_name": "text",
+     "name": "text",
+     "type": "STRING",
+     "widget": {
+      "name": "text"
+     },
+     "link": null
+    }
+   ],
+   "outputs": [
+    {
+     "localized_name": "text",
+     "name": "text",
+     "type": "STRING",
+     "links": [
+      24
+     ]
+    }
+   ],
+   "title": "Text - The request",
+   "properties": {
+    "Node name for S&R": "AUSBOSS_NODES_Text",
+    "cnr_id": "ausboss-nodes",
+    "ver": "2.8.0"
+   },
+   "widgets_values": [
+    "Restore the photo: remove blur, noise and compression artifacts and recover sharp, natural detail, keeping the same composition, people and colors."
+   ],
+   "widgets_values_named": {
+    "text": "Restore the photo: remove blur, noise and compression artifacts and recover sharp, natural detail, keeping the same composition, people and colors."
+   },
+   "color": "#14424d",
+   "bgcolor": "#161f21"
+  },
+  {
+   "id": 4,
+   "type": "AUSBOSS_NODES_Seed",
+   "pos": [
+    964,
+    1118.0
+   ],
+   "size": [
+    424,
+    216
+   ],
+   "flags": {},
+   "order": 7,
+   "mode": 0,
+   "inputs": [
+    {
+     "localized_name": "seed",
+     "name": "seed",
+     "type": "INT",
+     "widget": {
+      "name": "seed"
+     },
+     "link": null
+    }
+   ],
+   "outputs": [
+    {
+     "localized_name": "seed",
+     "name": "seed",
+     "type": "INT",
+     "links": [
+      29
+     ]
+    }
+   ],
+   "title": "Seed",
+   "properties": {
+    "Node name for S&R": "AUSBOSS_NODES_Seed",
+    "cnr_id": "ausboss-nodes",
+    "ver": "2.8.0"
+   },
+   "widgets_values": [
+    20261006,
+    "randomize"
+   ],
+   "widgets_values_named": {
+    "seed": 20261006,
+    "control_after_generate": "randomize"
+   },
+   "color": "#14424d",
+   "bgcolor": "#161f21"
+  },
+  {
+   "id": 5,
+   "type": "AUSBOSS_NODES_Integer",
+   "pos": [
+    964,
+    944.0
+   ],
+   "size": [
+    424,
+    94
+   ],
+   "flags": {},
+   "order": 6,
+   "mode": 0,
+   "inputs": [
+    {
+     "localized_name": "value",
+     "name": "value",
+     "type": "INT",
+     "widget": {
+      "name": "value"
+     },
+     "link": null
+    }
+   ],
+   "outputs": [
+    {
+     "localized_name": "int",
+     "name": "int",
+     "type": "INT",
+     "links": [
+      30
+     ]
+    }
+   ],
+   "title": "Integer - Steps",
+   "properties": {
+    "Node name for S&R": "AUSBOSS_NODES_Integer",
+    "cnr_id": "ausboss-nodes",
+    "ver": "2.8.0"
+   },
+   "widgets_values": [
+    8
+   ],
+   "widgets_values_named": {
+    "value": 8
+   },
+   "color": "#14424d",
+   "bgcolor": "#161f21"
+  },
+  {
+   "id": 6,
+   "type": "CLIPLoader",
+   "pos": [
+    1468,
+    614.0
+   ],
+   "size": [
+    300,
+    106
+   ],
+   "flags": {},
+   "order": 10,
+   "mode": 0,
+   "inputs": [
+    {
+     "localized_name": "clip_name",
+     "name": "clip_name",
+     "type": "COMBO",
+     "widget": {
+      "name": "clip_name"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "type",
+     "name": "type",
+     "type": "COMBO",
+     "widget": {
+      "name": "type"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "device",
+     "name": "device",
+     "shape": 7,
+     "type": "COMBO",
+     "widget": {
+      "name": "device"
+     },
+     "link": null
+    }
+   ],
+   "outputs": [
+    {
+     "localized_name": "CLIP",
+     "name": "CLIP",
+     "type": "CLIP",
+     "links": [
+      25
+     ]
+    }
+   ],
+   "title": "Text encoder",
+   "properties": {
+    "Node name for S&R": "CLIPLoader",
+    "models": [
+     {
+      "name": "qwen3vl_8b_int8_convrot.safetensors",
+      "url": "https://huggingface.co/Comfy-Org/Qwen-Image-2.1/resolve/main/text_encoders/qwen3vl_8b_int8_convrot.safetensors",
+      "directory": "text_encoders"
+     }
+    ],
+    "cnr_id": "comfy-core",
+    "ver": "0.38.0"
+   },
+   "widgets_values": [
+    "qwen3vl_8b_int8_convrot.safetensors",
+    "qwen_image",
+    "default"
+   ],
+   "widgets_values_named": {
+    "clip_name": "qwen3vl_8b_int8_convrot.safetensors",
+    "type": "qwen_image",
+    "device": "default"
+   },
+   "color": "#432",
+   "bgcolor": "#653"
+  },
+  {
+   "id": 7,
+   "type": "VAELoader",
+   "pos": [
+    1468,
+    794.0
+   ],
+   "size": [
+    300,
+    58
+   ],
+   "flags": {},
+   "order": 11,
+   "mode": 0,
+   "inputs": [
+    {
+     "localized_name": "vae_name",
+     "name": "vae_name",
+     "type": "COMBO",
+     "widget": {
+      "name": "vae_name"
+     },
+     "link": null
+    }
+   ],
+   "outputs": [
+    {
+     "localized_name": "VAE",
+     "name": "VAE",
+     "type": "VAE",
+     "links": [
+      27
+     ]
+    }
+   ],
+   "title": "VAE",
+   "properties": {
+    "Node name for S&R": "VAELoader",
+    "models": [
+     {
+      "name": "qwen_image_2.1_vae_bf16.safetensors",
+      "url": "https://huggingface.co/Comfy-Org/Qwen-Image-2.1/resolve/main/vae/qwen_image_2.1_vae_bf16.safetensors",
+      "directory": "vae"
+     }
+    ],
+    "cnr_id": "comfy-core",
+    "ver": "0.38.0"
+   },
+   "widgets_values": [
+    "qwen_image_2.1_vae_bf16.safetensors"
+   ],
+   "widgets_values_named": {
+    "vae_name": "qwen_image_2.1_vae_bf16.safetensors"
+   },
+   "color": "#322",
+   "bgcolor": "#533"
+  },
+  {
+   "id": 8,
+   "type": "UNETLoader",
+   "pos": [
+    1468,
+    926.0
+   ],
+   "size": [
+    300,
+    82
+   ],
+   "flags": {},
+   "order": 12,
+   "mode": 0,
+   "inputs": [
+    {
+     "localized_name": "unet_name",
+     "name": "unet_name",
+     "type": "COMBO",
+     "widget": {
+      "name": "unet_name"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "weight_dtype",
+     "name": "weight_dtype",
+     "type": "COMBO",
+     "widget": {
+      "name": "weight_dtype"
+     },
+     "link": null
+    }
+   ],
+   "outputs": [
+    {
+     "localized_name": "MODEL",
+     "name": "MODEL",
+     "type": "MODEL",
+     "links": [
+      6
+     ]
+    }
+   ],
+   "title": "Diffusion model",
+   "properties": {
+    "Node name for S&R": "UNETLoader",
+    "models": [
+     {
+      "name": "qwen_image_2.1_int8_convrot.safetensors",
+      "url": "https://huggingface.co/Comfy-Org/Qwen-Image-2.1/resolve/main/diffusion_models/qwen_image_2.1_int8_convrot.safetensors",
+      "directory": "diffusion_models"
+     }
+    ],
+    "cnr_id": "comfy-core",
+    "ver": "0.38.0"
+   },
+   "widgets_values": [
+    "qwen_image_2.1_int8_convrot.safetensors",
+    "default"
+   ],
+   "widgets_values_named": {
+    "unet_name": "qwen_image_2.1_int8_convrot.safetensors",
+    "weight_dtype": "default"
+   },
+   "color": "#323",
+   "bgcolor": "#535"
+  },
+  {
+   "id": 9,
+   "type": "AUSBOSS_NODES_LoraLoader",
+   "pos": [
+    1808,
+    806.0
+   ],
+   "size": [
+    400,
+    224
+   ],
+   "flags": {},
+   "order": 14,
+   "mode": 0,
+   "inputs": [
+    {
+     "localized_name": "model",
+     "name": "model",
+     "type": "MODEL",
+     "link": 6
+    },
+    {
+     "localized_name": "clip",
+     "name": "clip",
+     "shape": 7,
+     "type": "CLIP",
+     "link": null
+    },
+    {
+     "localized_name": "loras",
+     "name": "loras",
+     "type": "STRING",
+     "widget": {
+      "name": "loras"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "trigger_separator",
+     "name": "trigger_separator",
+     "shape": 7,
+     "type": "STRING",
+     "widget": {
+      "name": "trigger_separator"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "on_missing",
+     "name": "on_missing",
+     "shape": 7,
+     "type": "COMBO",
+     "widget": {
+      "name": "on_missing"
+     },
+     "link": null
+    }
+   ],
+   "outputs": [
+    {
+     "localized_name": "model",
+     "name": "model",
+     "type": "MODEL",
+     "links": [
+      28
+     ]
+    },
+    {
+     "localized_name": "clip",
+     "name": "clip",
+     "type": "CLIP",
+     "links": null
+    },
+    {
+     "localized_name": "triggers",
+     "name": "triggers",
+     "type": "STRING",
+     "links": null
+    }
+   ],
+   "title": "LoRA Loader - Restore LoRA",
+   "properties": {
+    "Node name for S&R": "AUSBOSS_NODES_LoraLoader",
+    "ausbossLoraLinked": true,
+    "models": [
+     {
+      "name": "qwen-image-2.1-restore.safetensors",
+      "url": "https://huggingface.co/ausboss/Qwen-Image-2.1-Restore-LoRA/resolve/main/qwen-image-2.1-restore.safetensors",
+      "directory": "loras"
+     },
+     {
+      "name": "Qwen-Image-2.1-viggle-turbo-v0.3-6step-lora-r128.safetensors",
+      "url": "https://huggingface.co/Viggle/Qwen-Image-2.1-viggle-turbo/resolve/main/Qwen-Image-2.1-viggle-turbo-v0.3-6step-lora-r128.safetensors",
+      "directory": "loras"
+     }
+    ],
+    "cnr_id": "ausboss-nodes",
+    "ver": "2.8.0"
+   },
+   "widgets_values": [
+    "[{\"name\": \"qwen-image-2.1-restore.safetensors\", \"strength\": 1.0, \"enabled\": true, \"triggers\": \"\"}, {\"name\": \"Qwen-Image-2.1-viggle-turbo-v0.3-6step-lora-r128.safetensors\", \"strength\": 1.0, \"enabled\": true, \"triggers\": \"\"}]",
+    ", ",
+    "error"
+   ],
+   "widgets_values_named": {
+    "loras": "[{\"name\": \"qwen-image-2.1-restore.safetensors\", \"strength\": 1.0, \"enabled\": true, \"triggers\": \"\"}, {\"name\": \"Qwen-Image-2.1-viggle-turbo-v0.3-6step-lora-r128.safetensors\", \"strength\": 1.0, \"enabled\": true, \"triggers\": \"\"}]",
+    "trigger_separator": ", ",
+    "on_missing": "error"
+   },
+   "color": "#14424d",
+   "bgcolor": "#161f21"
+  },
+  {
+   "id": 11,
+   "type": "AUSBOSS_NODES_Compare",
+   "pos": [
+    2288,
+    80.0
+   ],
+   "size": [
+    460,
+    460
+   ],
+   "flags": {},
+   "order": 15,
+   "mode": 0,
+   "inputs": [
+    {
+     "localized_name": "image_a",
+     "name": "image_a",
+     "type": "IMAGE",
+     "link": 21
+    },
+    {
+     "localized_name": "image_b",
+     "name": "image_b",
+     "type": "IMAGE",
+     "link": 22
+    }
+   ],
+   "outputs": [
+    {
+     "localized_name": "image_a",
+     "name": "image_a",
+     "type": "IMAGE",
+     "links": null
+    }
+   ],
+   "title": "Image Compare A/B - Before and after",
+   "properties": {
+    "Node name for S&R": "AUSBOSS_NODES_Compare",
+    "ausboss_compare_mode": "slide",
+    "cnr_id": "ausboss-nodes",
+    "ver": "2.8.0"
+   },
+   "widgets_values": [],
+   "widgets_values_named": {},
+   "color": "#14424d",
+   "bgcolor": "#161f21"
+  },
+  {
+   "id": 12,
+   "type": "AUSBOSS_NODES_SaveImage",
+   "pos": [
+    2288,
+    610.0
+   ],
+   "size": [
+    460,
+    566
+   ],
+   "flags": {},
+   "order": 16,
+   "mode": 0,
+   "inputs": [
+    {
+     "localized_name": "images",
+     "name": "images",
+     "type": "IMAGE",
+     "link": 23
+    },
+    {
+     "localized_name": "filename",
+     "name": "filename",
+     "shape": 7,
+     "type": "STRING",
+     "link": null
+    },
+    {
+     "localized_name": "caption_text",
+     "name": "caption_text",
+     "shape": 7,
+     "type": "STRING",
+     "link": null
+    },
+    {
+     "localized_name": "filename_prefix",
+     "name": "filename_prefix",
+     "type": "STRING",
+     "widget": {
+      "name": "filename_prefix"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "format",
+     "name": "format",
+     "type": "COMBO",
+     "widget": {
+      "name": "format"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "save_metadata",
+     "name": "save_metadata",
+     "type": "BOOLEAN",
+     "widget": {
+      "name": "save_metadata"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "exact_name",
+     "name": "exact_name",
+     "shape": 7,
+     "type": "STRING",
+     "widget": {
+      "name": "exact_name"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "on_existing",
+     "name": "on_existing",
+     "shape": 7,
+     "type": "COMBO",
+     "widget": {
+      "name": "on_existing"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "output_dir",
+     "name": "output_dir",
+     "shape": 7,
+     "type": "STRING",
+     "widget": {
+      "name": "output_dir"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "caption",
+     "name": "caption",
+     "shape": 7,
+     "type": "STRING",
+     "widget": {
+      "name": "caption"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "name_counter",
+     "name": "name_counter",
+     "shape": 7,
+     "type": "BOOLEAN",
+     "widget": {
+      "name": "name_counter"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "name_date",
+     "name": "name_date",
+     "shape": 7,
+     "type": "BOOLEAN",
+     "widget": {
+      "name": "name_date"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "name_time",
+     "name": "name_time",
+     "shape": 7,
+     "type": "BOOLEAN",
+     "widget": {
+      "name": "name_time"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "name_size",
+     "name": "name_size",
+     "shape": 7,
+     "type": "BOOLEAN",
+     "widget": {
+      "name": "name_size"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "name_batch",
+     "name": "name_batch",
+     "shape": 7,
+     "type": "BOOLEAN",
+     "widget": {
+      "name": "name_batch"
+     },
+     "link": null
+    },
+    {
+     "localized_name": "preview",
+     "name": "preview",
+     "shape": 7,
+     "type": "BOOLEAN",
+     "widget": {
+      "name": "preview"
+     },
+     "link": null
+    }
+   ],
+   "outputs": [
+    {
+     "localized_name": "file_path",
+     "name": "file_path",
+     "type": "STRING",
+     "links": null
+    },
+    {
+     "localized_name": "images",
+     "name": "images",
+     "type": "IMAGE",
+     "links": null
+    }
+   ],
+   "title": "Save Image",
+   "properties": {
+    "Node name for S&R": "AUSBOSS_NODES_SaveImage",
+    "cnr_id": "ausboss-nodes",
+    "ver": "2.8.0"
+   },
+   "widgets_values": {
+    "filename_prefix": "qwen21_restore",
+    "format": "png",
+    "save_metadata": true,
+    "exact_name": "",
+    "on_existing": "overwrite",
+    "output_dir": "",
+    "caption": "",
+    "name_counter": true,
+    "name_date": false,
+    "name_time": false,
+    "name_size": false,
+    "name_batch": false,
+    "preview": true
+   },
+   "widgets_values_named": {
+    "filename_prefix": "qwen21_restore",
+    "format": "png",
+    "save_metadata": true,
+    "exact_name": "",
+    "on_existing": "overwrite",
+    "output_dir": "",
+    "caption": "",
+    "name_counter": true,
+    "name_date": false,
+    "name_time": false,
+    "name_size": false,
+    "name_batch": false,
+    "preview": true
+   },
+   "color": "#14424d",
+   "bgcolor": "#161f21"
+  },
+  {
+   "id": 13,
+   "type": "AUSBOSS_NODES_RunTimer",
+   "pos": [
+    2413,
+    1246.0
+   ],
+   "size": [
+    210,
+    67
+   ],
+   "flags": {
+    "no_title": true
+   },
+   "order": 17,
+   "mode": 0,
+   "inputs": [],
+   "outputs": [],
+   "title": "Run Timer",
+   "properties": {
+    "Node name for S&R": "AUSBOSS_NODES_RunTimer",
+    "cnr_id": "ausboss-nodes",
+    "ver": "2.8.0"
+   },
+   "color": "#07090a",
+   "bgcolor": "#07090a"
+  },
+  {
+   "id": 14,
+   "type": "AUSBOSS_NODES_WorkflowSwitches",
+   "pos": [
+    964,
+    790.0
+   ],
+   "size": [
+    424,
+    66
+   ],
+   "flags": {},
+   "order": 5,
+   "mode": 0,
+   "inputs": [],
+   "outputs": [],
+   "title": "Workflow Switches - Speed",
+   "properties": {
+    "Node name for S&R": "AUSBOSS_NODES_WorkflowSwitches",
+    "ausboss_workflow_switches": {
+     "off": "bypass",
+     "groups": "none",
+     "match": "",
+     "order": "canvas",
+     "exclusive": false,
+     "switches": [
+      {
+       "id": 1,
+       "title": "Fast: Viggle Turbo, 8 steps",
+       "nodes": [],
+       "groups": [],
+       "on": {
+        "5": {
+         "value": 8
+        },
+        "9": {
+         "loras": "[{\"name\": \"Qwen-Image-2.1-viggle-turbo-v0.3-6step-lora-r128.safetensors\", \"strength\": 1.0, \"enabled\": true, \"triggers\": \"\"}]"
+        }
+       },
+       "off": {
+        "5": {
+         "value": 25
+        },
+        "9": {
+         "loras": "[{\"name\": \"Qwen-Image-2.1-viggle-turbo-v0.3-6step-lora-r128.safetensors\", \"strength\": 1.0, \"enabled\": false, \"triggers\": \"\"}]"
+        }
+       }
+      }
+     ]
+    },
+    "ausboss_show_edit": false,
+    "cnr_id": "ausboss-nodes",
+    "ver": "2.8.0"
+   },
+   "widgets_values": [],
+   "widgets_values_named": {},
+   "color": "#14424d",
+   "bgcolor": "#161f21"
+  },
+  {
+   "id": 23,
+   "type": "3e3084e1-f840-441b-989b-2ab0f904b43e",
+   "pos": [
+    1468,
+    286.0
+   ],
+   "size": [
+    740,
+    188
+   ],
+   "flags": {},
+   "order": 9,
+   "mode": 0,
+   "inputs": [
+    {
+     "name": "request",
+     "type": "*",
+     "link": 24
+    },
+    {
+     "name": "your picture",
+     "shape": 7,
+     "type": "IMAGE",
+     "link": 26
+    },
+    {
+     "name": "steps",
+     "type": "*",
+     "link": 30
+    },
+    {
+     "name": "seed",
+     "type": "*",
+     "link": 29
+    },
+    {
+     "name": "text encoder",
+     "type": "CLIP",
+     "link": 25
+    },
+    {
+     "name": "vae",
+     "shape": 7,
+     "type": "VAE",
+     "link": 27
+    },
+    {
+     "name": "diffusion model",
+     "type": "MODEL",
+     "link": 28
+    }
+   ],
+   "outputs": [
+    {
+     "name": "your picture, redrawn",
+     "type": "IMAGE",
+     "links": [
+      22,
+      23
+     ]
+    }
+   ],
+   "properties": {
+    "previewExposures": [
+     {
+      "name": "$$canvas-image-preview",
+      "sourceNodeId": "20",
+      "sourcePreviewName": "$$canvas-image-preview"
+     }
+    ]
+   },
+   "color": "#14424d",
+   "bgcolor": "#161f21"
+  },
+  {
+   "id": 24,
+   "flags": {},
+   "order": 1,
+   "mode": 0,
+   "type": "AUSBOSS_NODES_Callout",
+   "pos": [
+    500,
+    80.0
+   ],
+   "size": [
+    888,
+    136
+   ],
+   "inputs": [
+    {
+     "localized_name": "text",
+     "name": "text",
+     "type": "STRING",
+     "widget": {
+      "name": "text"
+     },
+     "link": null
+    }
+   ],
+   "outputs": [],
+   "title": "Callout",
+   "properties": {
+    "Node name for S&R": "AUSBOSS_NODES_Callout",
+    "cnr_id": "ausboss-nodes",
+    "ver": "2.8.0"
+   },
+   "widgets_values": [
+    "⬇️#2 Load your picture here.\n\nOptional: drag the blue handles to crop it first."
+   ],
+   "widgets_values_named": {
+    "text": "⬇️#2 Load your picture here.\n\nOptional: drag the blue handles to crop it first."
+   },
+   "color": "#1f2c38",
+   "bgcolor": "#2a3b4a"
+  },
+  {
+   "id": 25,
+   "flags": {},
+   "order": 4,
+   "mode": 0,
+   "type": "AUSBOSS_NODES_Callout",
+   "pos": [
+    964,
+    598.0
+   ],
+   "size": [
+    424,
+    122
+   ],
+   "inputs": [
+    {
+     "localized_name": "text",
+     "name": "text",
+     "type": "STRING",
+     "widget": {
+      "name": "text"
+     },
+     "link": null
+    }
+   ],
+   "outputs": [],
+   "title": "Callout",
+   "properties": {
+    "Node name for S&R": "AUSBOSS_NODES_Callout",
+    "cnr_id": "ausboss-nodes",
+    "ver": "2.8.0"
+   },
+   "widgets_values": [
+    "⬇️#14 Click **Fast** to toggle fast mode."
+   ],
+   "widgets_values_named": {
+    "text": "⬇️#14 Click **Fast** to toggle fast mode."
+   },
+   "color": "#1f2c38",
+   "bgcolor": "#2a3b4a"
+  },
+  {
+   "id": 27,
+   "flags": {},
+   "order": 8,
+   "mode": 0,
+   "type": "AUSBOSS_NODES_Callout",
+   "pos": [
+    1468,
+    80.0
+   ],
+   "size": [
+    740,
+    136
+   ],
+   "inputs": [
+    {
+     "localized_name": "text",
+     "name": "text",
+     "type": "STRING",
+     "widget": {
+      "name": "text"
+     },
+     "link": null
+    }
+   ],
+   "outputs": [],
+   "title": "Callout",
+   "properties": {
+    "Node name for S&R": "AUSBOSS_NODES_Callout",
+    "cnr_id": "ausboss-nodes",
+    "ver": "2.8.0"
+   },
+   "widgets_values": [
+    "One box does the work: Qwen reads the request and redraws your picture.\n\nDouble-click it to see inside."
+   ],
+   "widgets_values_named": {
+    "text": "One box does the work: Qwen reads the request and redraws your picture.\n\nDouble-click it to see inside."
+   },
+   "color": "#1f2c38",
+   "bgcolor": "#2a3b4a"
+  },
+  {
+   "id": 28,
+   "flags": {},
+   "order": 13,
+   "mode": 0,
+   "type": "AUSBOSS_NODES_Callout",
+   "pos": [
+    1808,
+    614.0
+   ],
+   "size": [
+    400,
+    122
+   ],
+   "inputs": [
+    {
+     "localized_name": "text",
+     "name": "text",
+     "type": "STRING",
+     "widget": {
+      "name": "text"
+     },
+     "link": null
+    }
+   ],
+   "outputs": [],
+   "title": "Callout",
+   "properties": {
+    "Node name for S&R": "AUSBOSS_NODES_Callout",
+    "cnr_id": "ausboss-nodes",
+    "ver": "2.8.0"
+   },
+   "widgets_values": [
+    "⬇️#9 Blurry photo? Set the first row to **0.5**."
+   ],
+   "widgets_values_named": {
+    "text": "⬇️#9 Blurry photo? Set the first row to **0.5**."
+   },
+   "color": "#1f2c38",
+   "bgcolor": "#2a3b4a"
+  }
+ ],
+ "links": [
+  [
+   6,
+   8,
+   0,
+   9,
+   0,
+   "MODEL"
+  ],
+  [
+   21,
+   2,
+   0,
+   11,
+   0,
+   "IMAGE"
+  ],
+  [
+   22,
+   23,
+   0,
+   11,
+   1,
+   "IMAGE"
+  ],
+  [
+   23,
+   23,
+   0,
+   12,
+   0,
+   "IMAGE"
+  ],
+  [
+   24,
+   3,
+   0,
+   23,
+   0,
+   "STRING"
+  ],
+  [
+   25,
+   6,
+   0,
+   23,
+   4,
+   "CLIP"
+  ],
+  [
+   26,
+   2,
+   0,
+   23,
+   1,
+   "IMAGE"
+  ],
+  [
+   27,
+   7,
+   0,
+   23,
+   5,
+   "VAE"
+  ],
+  [
+   28,
+   9,
+   0,
+   23,
+   6,
+   "MODEL"
+  ],
+  [
+   29,
+   4,
+   0,
+   23,
+   3,
+   "INT"
+  ],
+  [
+   30,
+   5,
+   0,
+   23,
+   2,
+   "INT"
+  ]
+ ],
+ "groups": [
+  {
+   "id": 1,
+   "title": "1 · Your picture",
+   "bounding": [
+    480,
+    0.0,
+    928,
+    1360.0
+   ],
+   "color": "#1d5560",
+   "flags": {}
+  },
+  {
+   "id": 2,
+   "title": "2 · Behind the scenes",
+   "bounding": [
+    1448,
+    0.0,
+    780,
+    494.0
+   ],
+   "color": "#1d5560",
+   "flags": {}
+  },
+  {
+   "id": 3,
+   "title": "3 · Models",
+   "bounding": [
+    1448,
+    534.0,
+    780,
+    826.0
+   ],
+   "color": "#1d5560",
+   "flags": {}
+  },
+  {
+   "id": 4,
+   "title": "4 · Result",
+   "bounding": [
+    2268,
+    0.0,
+    500,
+    1360.0
+   ],
+   "color": "#1d5560",
+   "flags": {}
+  }
+ ],
+ "definitions": {
+  "subgraphs": [
+   {
+    "id": "3e3084e1-f840-441b-989b-2ab0f904b43e",
+    "version": 1,
+    "state": {
+     "lastGroupId": 7,
+     "lastNodeId": 33,
+     "lastLinkId": 31,
+     "lastRerouteId": 0
+    },
+    "revision": 0,
+    "config": {},
+    "name": "Redraw it with Qwen",
+    "inputNode": {
+     "id": -10,
+     "bounding": [
+      -260,
+      120,
+      150,
+      180
+     ]
+    },
+    "outputNode": {
+     "id": -20,
+     "bounding": [
+      1690,
+      120,
+      150,
+      60
+     ]
+    },
+    "inputs": [
+     {
+      "id": "b7d63873-f7a2-4a55-8d67-726cd64e8603",
+      "name": "request",
+      "type": "*",
+      "linkIds": [
+       2
+      ],
+      "pos": [
+       51,
+       24
+      ],
+      "localized_name": "request"
+     },
+     {
+      "id": "460d3ae3-4a76-4adc-8896-64c12c36f494",
+      "name": "your picture",
+      "type": "IMAGE",
+      "linkIds": [
+       1
+      ],
+      "localized_name": "your picture",
+      "shape": 7,
+      "pos": [
+       51,
+       44
+      ]
+     },
+     {
+      "id": "9bd884f9-9bf8-4d5a-9d94-6d0834242441",
+      "name": "steps",
+      "type": "*",
+      "linkIds": [
+       11
+      ],
+      "pos": [
+       51,
+       64
+      ],
+      "localized_name": "steps"
+     },
+     {
+      "id": "a18b12bd-0c48-4a18-9b7e-bb109d76fc98",
+      "name": "seed",
+      "type": "*",
+      "linkIds": [
+       9
+      ],
+      "pos": [
+       51,
+       84
+      ],
+      "localized_name": "seed"
+     },
+     {
+      "id": "c289c048-98ac-4a5e-af92-9bff20aca5b6",
+      "name": "text encoder",
+      "type": "CLIP",
+      "linkIds": [
+       4
+      ],
+      "localized_name": "text encoder",
+      "pos": [
+       51,
+       104
+      ]
+     },
+     {
+      "id": "3d56affa-af5c-4fc9-a07f-fb3de7e2a711",
+      "name": "vae",
+      "type": "VAE",
+      "linkIds": [
+       5,
+       17
+      ],
+      "localized_name": "vae",
+      "shape": 7,
+      "pos": [
+       51,
+       124
+      ]
+     },
+     {
+      "id": "217fb786-1813-43f2-a326-56ed4caf398f",
+      "name": "diffusion model",
+      "type": "MODEL",
+      "linkIds": [
+       7
+      ],
+      "localized_name": "diffusion model",
+      "pos": [
+       51,
+       144
+      ]
+     }
+    ],
+    "outputs": [
+     {
+      "id": "67263c15-bfaa-44bb-8c61-909c6a2bb06b",
+      "name": "your picture, redrawn",
+      "type": "IMAGE",
+      "linkIds": [
+       19
+      ],
+      "localized_name": "your picture, redrawn",
+      "pos": [
+       24,
+       24
+      ]
+     }
+    ],
+    "widgets": [],
+    "nodes": [
+     {
+      "id": 15,
+      "type": "Reroute",
+      "pos": [
+       20,
+       272.0
+      ],
+      "size": [
+       75,
+       26
+      ],
+      "flags": {},
+      "order": 0,
+      "mode": 0,
+      "inputs": [
+       {
+        "name": "",
+        "type": "*",
+        "widget": {
+         "name": "value"
+        },
+        "link": 2
+       }
+      ],
+      "outputs": [
+       {
+        "name": "",
+        "type": "STRING",
+        "links": [
+         3
+        ]
+       }
+      ],
+      "properties": {
+       "showOutputText": false,
+       "horizontal": false
+      }
+     },
+     {
+      "id": 16,
+      "type": "TextEncodeQwenImage21",
+      "pos": [
+       20,
+       358.0
+      ],
+      "size": [
+       460,
+       266
+      ],
+      "flags": {},
+      "order": 1,
+      "mode": 0,
+      "inputs": [
+       {
+        "localized_name": "clip",
+        "name": "clip",
+        "type": "CLIP",
+        "link": 4
+       },
+       {
+        "localized_name": "image_1",
+        "name": "images.image_1",
+        "shape": 7,
+        "type": "IMAGE",
+        "link": 1
+       },
+       {
+        "localized_name": "vae",
+        "name": "vae",
+        "shape": 7,
+        "type": "VAE",
+        "link": 5
+       },
+       {
+        "localized_name": "prompt",
+        "name": "prompt",
+        "type": "STRING",
+        "widget": {
+         "name": "prompt"
+        },
+        "link": 3
+       },
+       {
+        "localized_name": "negative_prompt",
+        "name": "negative_prompt",
+        "type": "STRING",
+        "widget": {
+         "name": "negative_prompt"
+        },
+        "link": null
+       },
+       {
+        "localized_name": "resolution",
+        "name": "resolution",
+        "type": "INT",
+        "widget": {
+         "name": "resolution"
+        },
+        "link": null
+       },
+       {
+        "localized_name": "image_2",
+        "name": "images.image_2",
+        "shape": 7,
+        "type": "IMAGE",
+        "link": null
+       }
+      ],
+      "outputs": [
+       {
+        "localized_name": "positive",
+        "name": "positive",
+        "type": "CONDITIONING",
+        "links": [
+         13
+        ]
+       },
+       {
+        "localized_name": "negative",
+        "name": "negative",
+        "type": "CONDITIONING",
+        "links": [
+         14
+        ]
+       },
+       {
+        "localized_name": "latent",
+        "name": "latent",
+        "type": "LATENT",
+        "links": [
+         15
+        ]
+       }
+      ],
+      "title": "Text Encode Qwen Image 2.1 - Your picture and the request",
+      "properties": {
+       "Node name for S&R": "TextEncodeQwenImage21",
+       "cnr_id": "comfy-core",
+       "ver": "0.38.0"
+      },
+      "widgets_values": [
+       "",
+       "",
+       0
+      ],
+      "widgets_values_named": {
+       "prompt": "",
+       "negative_prompt": "",
+       "resolution": 0
+      },
+      "color": "#14424d",
+      "bgcolor": "#161f21"
+     },
+     {
+      "id": 17,
+      "type": "QwenImage21Cache",
+      "pos": [
+       560,
+       292.0
+      ],
+      "size": [
+       460,
+       82
+      ],
+      "flags": {},
+      "order": 2,
+      "mode": 0,
+      "inputs": [
+       {
+        "localized_name": "model",
+        "name": "model",
+        "type": "MODEL",
+        "link": 7
+       },
+       {
+        "localized_name": "device",
+        "name": "device",
+        "type": "COMBO",
+        "widget": {
+         "name": "device"
+        },
+        "link": null
+       },
+       {
+        "localized_name": "dtype",
+        "name": "dtype",
+        "type": "COMBO",
+        "widget": {
+         "name": "dtype"
+        },
+        "link": null
+       }
+      ],
+      "outputs": [
+       {
+        "localized_name": "MODEL",
+        "name": "MODEL",
+        "type": "MODEL",
+        "links": [
+         8
+        ]
+       }
+      ],
+      "title": "Reference cache",
+      "properties": {
+       "Node name for S&R": "QwenImage21Cache",
+       "cnr_id": "comfy-core",
+       "ver": "0.38.0"
+      },
+      "widgets_values": [
+       "auto",
+       "default"
+      ],
+      "widgets_values_named": {
+       "device": "auto",
+       "dtype": "default"
+      },
+      "color": "#14424d",
+      "bgcolor": "#161f21"
+     },
+     {
+      "id": 18,
+      "type": "Reroute",
+      "pos": [
+       560,
+       448.0
+      ],
+      "size": [
+       75,
+       26
+      ],
+      "flags": {},
+      "order": 3,
+      "mode": 0,
+      "inputs": [
+       {
+        "name": "",
+        "type": "*",
+        "widget": {
+         "name": "value"
+        },
+        "link": 9
+       }
+      ],
+      "outputs": [
+       {
+        "name": "",
+        "type": "INT",
+        "links": [
+         10
+        ]
+       }
+      ],
+      "properties": {
+       "showOutputText": false,
+       "horizontal": false
+      }
+     },
+     {
+      "id": 19,
+      "type": "Reroute",
+      "pos": [
+       560,
+       514.0
+      ],
+      "size": [
+       75,
+       26
+      ],
+      "flags": {},
+      "order": 4,
+      "mode": 0,
+      "inputs": [
+       {
+        "name": "",
+        "type": "*",
+        "widget": {
+         "name": "value"
+        },
+        "link": 11
+       }
+      ],
+      "outputs": [
+       {
+        "name": "",
+        "type": "INT",
+        "links": [
+         12
+        ]
+       }
+      ],
+      "properties": {
+       "showOutputText": false,
+       "horizontal": false
+      }
+     },
+     {
+      "id": 20,
+      "type": "KSampler",
+      "pos": [
+       560,
+       600.0
+      ],
+      "size": [
+       460,
+       262
+      ],
+      "flags": {},
+      "order": 5,
+      "mode": 0,
+      "inputs": [
+       {
+        "localized_name": "model",
+        "name": "model",
+        "type": "MODEL",
+        "link": 8
+       },
+       {
+        "localized_name": "positive",
+        "name": "positive",
+        "type": "CONDITIONING",
+        "link": 13
+       },
+       {
+        "localized_name": "negative",
+        "name": "negative",
+        "type": "CONDITIONING",
+        "link": 14
+       },
+       {
+        "localized_name": "latent_image",
+        "name": "latent_image",
+        "type": "LATENT",
+        "link": 15
+       },
+       {
+        "localized_name": "seed",
+        "name": "seed",
+        "type": "INT",
+        "widget": {
+         "name": "seed"
+        },
+        "link": 10
+       },
+       {
+        "localized_name": "steps",
+        "name": "steps",
+        "type": "INT",
+        "widget": {
+         "name": "steps"
+        },
+        "link": 12
+       },
+       {
+        "localized_name": "cfg",
+        "name": "cfg",
+        "type": "FLOAT",
+        "widget": {
+         "name": "cfg"
+        },
+        "link": null
+       },
+       {
+        "localized_name": "sampler_name",
+        "name": "sampler_name",
+        "type": "COMBO",
+        "widget": {
+         "name": "sampler_name"
+        },
+        "link": null
+       },
+       {
+        "localized_name": "scheduler",
+        "name": "scheduler",
+        "type": "COMBO",
+        "widget": {
+         "name": "scheduler"
+        },
+        "link": null
+       },
+       {
+        "localized_name": "denoise",
+        "name": "denoise",
+        "type": "FLOAT",
+        "widget": {
+         "name": "denoise"
+        },
+        "link": null
+       }
+      ],
+      "outputs": [
+       {
+        "localized_name": "LATENT",
+        "name": "LATENT",
+        "type": "LATENT",
+        "links": [
+         16
+        ]
+       }
+      ],
+      "title": "Sampler - CFG 1, euler",
+      "properties": {
+       "Node name for S&R": "KSampler",
+       "cnr_id": "comfy-core",
+       "ver": "0.38.0"
+      },
+      "widgets_values": [
+       0,
+       "fixed",
+       25,
+       1,
+       "euler",
+       "simple",
+       1
+      ],
+      "widgets_values_named": {
+       "seed": 0,
+       "control_after_generate": "fixed",
+       "steps": 25,
+       "cfg": 1,
+       "sampler_name": "euler",
+       "scheduler": "simple",
+       "denoise": 1
+      },
+      "color": "#14424d",
+      "bgcolor": "#161f21"
+     },
+     {
+      "id": 21,
+      "type": "VAEDecode",
+      "pos": [
+       1100,
+       292.0
+      ],
+      "size": [
+       460,
+       46
+      ],
+      "flags": {},
+      "order": 6,
+      "mode": 0,
+      "inputs": [
+       {
+        "localized_name": "samples",
+        "name": "samples",
+        "type": "LATENT",
+        "link": 16
+       },
+       {
+        "localized_name": "vae",
+        "name": "vae",
+        "type": "VAE",
+        "link": 17
+       }
+      ],
+      "outputs": [
+       {
+        "localized_name": "IMAGE",
+        "name": "IMAGE",
+        "type": "IMAGE",
+        "links": [
+         18
+        ]
+       }
+      ],
+      "title": "Decode",
+      "properties": {
+       "Node name for S&R": "VAEDecode",
+       "cnr_id": "comfy-core",
+       "ver": "0.38.0"
+      },
+      "color": "#14424d",
+      "bgcolor": "#161f21"
+     },
+     {
+      "id": 22,
+      "type": "SplitImageWithAlpha",
+      "pos": [
+       1100,
+       408.0
+      ],
+      "size": [
+       460,
+       46
+      ],
+      "flags": {},
+      "order": 7,
+      "mode": 0,
+      "inputs": [
+       {
+        "localized_name": "image",
+        "name": "image",
+        "type": "IMAGE",
+        "link": 18
+       }
+      ],
+      "outputs": [
+       {
+        "localized_name": "IMAGE",
+        "name": "IMAGE",
+        "type": "IMAGE",
+        "links": [
+         19
+        ]
+       },
+       {
+        "localized_name": "MASK",
+        "name": "MASK",
+        "type": "MASK",
+        "links": null
+       }
+      ],
+      "title": "Split Image with Alpha - Drop the see-through channel",
+      "properties": {
+       "Node name for S&R": "SplitImageWithAlpha",
+       "cnr_id": "comfy-core",
+       "ver": "0.38.0"
+      },
+      "color": "#14424d",
+      "bgcolor": "#161f21"
+     },
+     {
+      "id": 31,
+      "flags": {},
+      "order": 0,
+      "mode": 0,
+      "type": "AUSBOSS_NODES_Callout",
+      "pos": [
+       20,
+       80.0
+      ],
+      "size": [
+       460,
+       142
+      ],
+      "inputs": [
+       {
+        "localized_name": "text",
+        "name": "text",
+        "type": "STRING",
+        "widget": {
+         "name": "text"
+        },
+        "link": null
+       }
+      ],
+      "outputs": [],
+      "title": "Callout",
+      "properties": {
+       "Node name for S&R": "AUSBOSS_NODES_Callout",
+       "cnr_id": "ausboss-nodes",
+       "ver": "2.8.0"
+      },
+      "widgets_values": [
+       "Qwen reads your picture at its own size, together with the request. The same node sets the size of the result."
+      ],
+      "widgets_values_named": {
+       "text": "Qwen reads your picture at its own size, together with the request. The same node sets the size of the result."
+      },
+      "color": "#1f2c38",
+      "bgcolor": "#2a3b4a"
+     },
+     {
+      "id": 32,
+      "flags": {},
+      "order": 0,
+      "mode": 0,
+      "type": "AUSBOSS_NODES_Callout",
+      "pos": [
+       560,
+       80.0
+      ],
+      "size": [
+       460,
+       142
+      ],
+      "inputs": [
+       {
+        "localized_name": "text",
+        "name": "text",
+        "type": "STRING",
+        "widget": {
+         "name": "text"
+        },
+        "link": null
+       }
+      ],
+      "outputs": [],
+      "title": "Callout",
+      "properties": {
+       "Node name for S&R": "AUSBOSS_NODES_Callout",
+       "cnr_id": "ausboss-nodes",
+       "ver": "2.8.0"
+      },
+      "widgets_values": [
+       "Steps and seed come from the top level, CFG 1. The Restore LoRA is on the model, so the result stays close to your picture."
+      ],
+      "widgets_values_named": {
+       "text": "Steps and seed come from the top level, CFG 1. The Restore LoRA is on the model, so the result stays close to your picture."
+      },
+      "color": "#1f2c38",
+      "bgcolor": "#2a3b4a"
+     },
+     {
+      "id": 33,
+      "flags": {},
+      "order": 0,
+      "mode": 0,
+      "type": "AUSBOSS_NODES_Callout",
+      "pos": [
+       1100,
+       80.0
+      ],
+      "size": [
+       460,
+       142
+      ],
+      "inputs": [
+       {
+        "localized_name": "text",
+        "name": "text",
+        "type": "STRING",
+        "widget": {
+         "name": "text"
+        },
+        "link": null
+       }
+      ],
+      "outputs": [],
+      "title": "Callout",
+      "properties": {
+       "Node name for S&R": "AUSBOSS_NODES_Callout",
+       "cnr_id": "ausboss-nodes",
+       "ver": "2.8.0"
+      },
+      "widgets_values": [
+       "Qwen's VAE returns a see-through channel too. It is dropped here, so the saved PNG is a plain picture."
+      ],
+      "widgets_values_named": {
+       "text": "Qwen's VAE returns a see-through channel too. It is dropped here, so the saved PNG is a plain picture."
+      },
+      "color": "#1f2c38",
+      "bgcolor": "#2a3b4a"
+     }
+    ],
+    "groups": [
+     {
+      "id": 5,
+      "title": "1 · Read",
+      "bounding": [
+       0,
+       0.0,
+       500,
+       676.0
+      ],
+      "color": "#1d5560",
+      "flags": {}
+     },
+     {
+      "id": 6,
+      "title": "2 · Redraw",
+      "bounding": [
+       540,
+       0.0,
+       500,
+       882.0
+      ],
+      "color": "#1d5560",
+      "flags": {}
+     },
+     {
+      "id": 7,
+      "title": "3 · Picture",
+      "bounding": [
+       1080,
+       0.0,
+       500,
+       482.0
+      ],
+      "color": "#1d5560",
+      "flags": {}
+     }
+    ],
+    "links": [
+     {
+      "id": 3,
+      "origin_id": 15,
+      "origin_slot": 0,
+      "target_id": 16,
+      "target_slot": 3,
+      "type": "STRING"
+     },
+     {
+      "id": 8,
+      "origin_id": 17,
+      "origin_slot": 0,
+      "target_id": 20,
+      "target_slot": 0,
+      "type": "MODEL"
+     },
+     {
+      "id": 13,
+      "origin_id": 16,
+      "origin_slot": 0,
+      "target_id": 20,
+      "target_slot": 1,
+      "type": "CONDITIONING"
+     },
+     {
+      "id": 14,
+      "origin_id": 16,
+      "origin_slot": 1,
+      "target_id": 20,
+      "target_slot": 2,
+      "type": "CONDITIONING"
+     },
+     {
+      "id": 15,
+      "origin_id": 16,
+      "origin_slot": 2,
+      "target_id": 20,
+      "target_slot": 3,
+      "type": "LATENT"
+     },
+     {
+      "id": 10,
+      "origin_id": 18,
+      "origin_slot": 0,
+      "target_id": 20,
+      "target_slot": 4,
+      "type": "INT"
+     },
+     {
+      "id": 12,
+      "origin_id": 19,
+      "origin_slot": 0,
+      "target_id": 20,
+      "target_slot": 5,
+      "type": "INT"
+     },
+     {
+      "id": 16,
+      "origin_id": 20,
+      "origin_slot": 0,
+      "target_id": 21,
+      "target_slot": 0,
+      "type": "LATENT"
+     },
+     {
+      "id": 18,
+      "origin_id": 21,
+      "origin_slot": 0,
+      "target_id": 22,
+      "target_slot": 0,
+      "type": "IMAGE"
+     },
+     {
+      "id": 2,
+      "origin_id": -10,
+      "origin_slot": 0,
+      "target_id": 15,
+      "target_slot": 0,
+      "type": "STRING"
+     },
+     {
+      "id": 4,
+      "origin_id": -10,
+      "origin_slot": 4,
+      "target_id": 16,
+      "target_slot": 0,
+      "type": "CLIP"
+     },
+     {
+      "id": 1,
+      "origin_id": -10,
+      "origin_slot": 1,
+      "target_id": 16,
+      "target_slot": 1,
+      "type": "IMAGE"
+     },
+     {
+      "id": 5,
+      "origin_id": -10,
+      "origin_slot": 5,
+      "target_id": 16,
+      "target_slot": 2,
+      "type": "VAE"
+     },
+     {
+      "id": 17,
+      "origin_id": -10,
+      "origin_slot": 5,
+      "target_id": 21,
+      "target_slot": 1,
+      "type": "VAE"
+     },
+     {
+      "id": 7,
+      "origin_id": -10,
+      "origin_slot": 6,
+      "target_id": 17,
+      "target_slot": 0,
+      "type": "MODEL"
+     },
+     {
+      "id": 9,
+      "origin_id": -10,
+      "origin_slot": 3,
+      "target_id": 18,
+      "target_slot": 0,
+      "type": "INT"
+     },
+     {
+      "id": 11,
+      "origin_id": -10,
+      "origin_slot": 2,
+      "target_id": 19,
+      "target_slot": 0,
+      "type": "INT"
+     },
+     {
+      "id": 19,
+      "origin_id": 22,
+      "origin_slot": 0,
+      "target_id": -20,
+      "target_slot": 0,
+      "type": "IMAGE"
+     }
+    ],
+    "extra": {}
+   }
+  ]
+ },
+ "config": {},
+ "extra": {
+  "ds": {
+   "scale": 0.6,
+   "offset": [
+    6,
+    60
+   ]
+  },
+  "frontendVersion": "1.53.6"
+ },
+ "version": 0.4
+}