ThalisAI commited on
Commit
a296974
·
verified ·
1 Parent(s): 5f575c6

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +61 -0
README.md ADDED
@@ -0,0 +1,61 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ language: en
4
+ tags:
5
+ - image-to-image
6
+ - upscale
7
+ - super-resolution
8
+ - swin2sr
9
+ - photography
10
+ library_name: other
11
+ ---
12
+
13
+ # Swin2SR Upscaler (x2 and x4) — Custom Trained
14
+
15
+ Sharper skin. Better textures. Trained on AI-generated photorealistic images.
16
+
17
+ **Swin2SR Upscaler** is a series of custom-trained super-resolution models based on the Swin2SR architecture. Unlike generic upscalers that smooth out fine detail, these were trained specifically on high-resolution photorealistic AI-generated images to preserve and enhance skin textures, hair detail, and fabric weave.
18
+
19
+ Three variants are provided — all in safetensors format for easy integration with ComfyUI and other pipelines.
20
+
21
+ ## Gallery
22
+
23
+ | |
24
+ |---|
25
+ | ![Test Pattern](images/swin2sr-test-pattern.jpeg) |
26
+
27
+ ## Available Versions
28
+
29
+ | File | Scale | Training Data | Size |
30
+ |------|-------|--------------|------|
31
+ | `swin2sr-custom-x2.safetensors` | 2x | Custom photorealistic dataset | 58 MB |
32
+ | `swin2sr-custom-x4.safetensors` | 4x | Custom photorealistic dataset | 58 MB |
33
+ | `swin2sr-div2k-x2.safetensors` | 2x | DIV2K + Custom blend | 58 MB |
34
+
35
+ ## Usage
36
+
37
+ These are standalone Swin2SR models, not LoRAs. Use them in any pipeline that supports Swin2SR upscale models.
38
+
39
+ ### In ComfyUI
40
+
41
+ Load via the **UpscaleModelLoader** node and connect to **ImageUpscaleWithModel**.
42
+
43
+ ### Tips
44
+
45
+ - **x2 custom** is the best general-purpose choice — sharp detail without artifacts
46
+ - **x4 custom** for maximum upscale in one pass, though quality is slightly softer
47
+ - **div2k + custom x2** blends academic training data with custom images — may produce different texture characteristics
48
+ - **Best for photorealistic images** — these models were optimized for skin, hair, and fabric
49
+
50
+ ## Links
51
+
52
+ - [Civitai](https://civitai.com/models/1330184)
53
+ - [All Models](https://thalis.dev/models/)
54
+
55
+ ## About
56
+
57
+ Created by [Thalis AI](https://thalis.dev).
58
+
59
+ ## License
60
+
61
+ Released under the Apache 2.0 license.