Add Baikal LoopSR v2 x2 EMA weights and updated examples

Publish EMA update 30000 with its architecture config and paired examples; retain LoopSR v1 and SwinFIR v31.

Baikal_LoopSR_x2_v2.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a9ec49141e58b9ba109f3dde7d408dfccabbae08c0ba8fb3545ba3017e95f643
+size 20954228

README.md

@@ -1,45 +1,46 @@
----
-license: mit
-pipeline_tag: image-to-image
-library_name: pytorch
-tags: [super-resolution, image-restoration, anime, illustration, comfyui, transformer, loopsr, swinfir]
----
-
-# Baikal Anime Upscaler
-
-**x2 upscaling for anime and illustrations**, using Baikal LoopSR or SwinFIR v31.
-
-[ComfyUI nodes](https://github.com/SnJake/ComfyUI-Baikal-Anime/tree/dev) · [Example workflow](https://github.com/SnJake/ComfyUI-Baikal-Anime/blob/dev/workflows/Baikal_LoopSR_x2.json)
-
-| Model | Download |
-|---|---|
-| Baikal LoopSR x2 | [Baikal_LoopSR_x2.safetensors](Baikal_LoopSR_x2.safetensors) |
-| Baikal SwinFIR v31 | [Baikal_SwinFIR_Anime_x2_v31.safetensors](Baikal_SwinFIR_Anime_x2_v31.safetensors) |
-
-## Examples
-
-**Bicubic x2 → Baikal LoopSR x2.** Same display size, no extra sharpening. LoopSR: EMA step 24000, loops 4.
-
-![Illustration before and after](examples/illustration-overview.webp)
-
-![Illustration details](examples/illustration-details.png)
-
-[Full input](examples/illustration-input.png) · [Full x2 result](examples/illustration-loopsr-x2.png)
-
-![Anime before and after](examples/anime-overview.webp)
-
-![Anime details](examples/anime-details.png)
-
-[Full input](examples/anime-input.png) · [Full x2 result](examples/anime-loopsr-x2.png)
-
-## ComfyUI
-
-Install [ComfyUI-Baikal-Anime](https://github.com/SnJake/ComfyUI-Baikal-Anime/tree/dev), restart, and place weights in **`ComfyUI/models/anime_upscale/`**. The loader can also download them.
-
-Category **`Baikal/Upscale`**: **Baikal Model Loader → Baikal Anime Upscale → Save Image**.
-
-LoopSR supports **loops 1–4**. Default LR tile/overlap/halo: **256 / 32 / 96**. The custom loader also provides `UPSCALE_MODEL` for the standard upscaling node.
-
-LoopSR adapts shared loops, XSA and deep supervision from [Looped Diffusion Transformer](https://arxiv.org/abs/2609.40305). [Technical notes](https://github.com/SnJake/ComfyUI-Baikal-Anime/blob/dev/docs/TRAINING.md).
-
-MIT · [License](LICENSE.md)
+---
+license: mit
+pipeline_tag: image-to-image
+library_name: pytorch
+tags: [super-resolution, image-restoration, anime, illustration, comfyui, transformer, loopsr, swinfir]
+---
+
+# Baikal Anime Upscaler
+
+**x2 upscaling for anime and illustrations**, using Baikal LoopSR v2, LoopSR v1 or SwinFIR v31.
+
+[ComfyUI nodes](https://github.com/SnJake/ComfyUI-Baikal-Anime/tree/dev) · [Example workflow](https://github.com/SnJake/ComfyUI-Baikal-Anime/blob/dev/workflows/Baikal_LoopSR_x2_v2.json)
+
+| Model | Download |
+|---|---|
+| Baikal LoopSR v2 x2 | [Baikal_LoopSR_x2_v2.safetensors](https://huggingface.co/SnJake/Baikal-Anime-Upscaler/resolve/dev/Baikal_LoopSR_x2_v2.safetensors?download=true) |
+| Baikal LoopSR v1 x2 | [Baikal_LoopSR_x2.safetensors](Baikal_LoopSR_x2.safetensors) |
+| Baikal SwinFIR v31 | [Baikal_SwinFIR_Anime_x2_v31.safetensors](Baikal_SwinFIR_Anime_x2_v31.safetensors) |
+
+## Examples
+
+**Bicubic x2 → Baikal LoopSR v2 x2.** Same display size, no extra sharpening. V2: EMA step 30000, loops 4.
+
+![Illustration before and after](examples/illustration-overview.webp)
+
+![Illustration details](examples/illustration-details.png)
+
+[Full input](examples/illustration-input.png) · [Full x2 result](examples/illustration-loopsr-x2.png)
+
+![Anime before and after](examples/anime-overview.webp)
+
+![Anime details](examples/anime-details.png)
+
+[Full input](examples/anime-input.png) · [Full x2 result](examples/anime-loopsr-x2.png)
+
+## ComfyUI
+
+Install [ComfyUI-Baikal-Anime](https://github.com/SnJake/ComfyUI-Baikal-Anime/tree/dev), restart, and place weights in **`ComfyUI/models/anime_upscale/`**. The loader can also download them.
+
+Category **`Baikal/Upscale`**: **Baikal Model Loader → Baikal Anime Upscale → Save Image**.
+
+LoopSR supports **loops 1–4**. Default LR tile/overlap/halo: **256 / 32 / 96**. The custom loader also provides `UPSCALE_MODEL` for the standard upscaling node.
+
+LoopSR adapts shared loops, XSA and deep supervision from [Looped Diffusion Transformer](https://arxiv.org/abs/2609.40305). [Technical notes](https://github.com/SnJake/ComfyUI-Baikal-Anime/blob/dev/docs/TRAINING.md).
+
+MIT · [License](LICENSE.md)

examples/metadata.json

@@ -1,7 +1,7 @@
 {
-  "model": "Baikal_LoopSR_x2.safetensors",
+  "model": "Baikal_LoopSR_x2_v2.safetensors",
   "weights": "EMA",
-  "training_steps": 24000,
+  "training_steps": 30000,
   "baseline": "bicubic x2, not an HR reference",
   "examples": [
     {

loopsr_x2_v2_config.json

@@ -0,0 +1,19 @@
+{
+  "architecture": "Baikal_LoopSR",
+  "model_version": 2,
+  "model": {
+    "scale": 2,
+    "dim": 224,
+    "heads": 7,
+    "window": 8,
+    "pre_blocks": 3,
+    "loop_blocks": 4,
+    "post_blocks": 2,
+    "loops": 4,
+    "ffn_ratio": 2.0,
+    "attention": "xsa",
+    "checkpoint_blocks": false
+  },
+  "training_steps": 30000,
+  "weights": "EMA"
+}