ussoewwin commited on
Commit
972aa40
ยท
verified ยท
1 Parent(s): a85c56e

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +9 -4
README.md CHANGED
@@ -7,6 +7,7 @@ tags:
7
  - inpainting
8
  - sdxl
9
  - stable-diffusion-xl
 
10
  - qwen-image
11
  - qwen-image-2512
12
  - flux
@@ -20,14 +21,15 @@ pipeline_tag: image-to-image
20
 
21
  # ControlNet Models (ConvRot INT8)
22
 
23
- High-fidelity **ConvRot INT8** quantized weights for multi-condition ControlNet models across diverse generative architectures (**SDXL 1.0**, **Qwen-Image**, and **FLUX.1-dev**).
24
 
25
  ---
26
 
27
  ## ๐ŸŒŸ Model Overview
28
 
29
- This repository hosts high-quality **ConvRot INT8** quantized weights for all-in-one ControlNet Union models. By applying orthogonal Hadamard rotation prior to per-channel INT8 quantization, these models effectively eliminate activation outlier distortion and drastically reduce VRAM / disk footprint while preserving precise structural control fidelity:
30
 
 
31
  - **controlnet-union-pro-max-sdxl-1.0**: The comprehensive all-in-one ControlNet Union model for **SDXL 1.0** by **xinsir**, supporting 10+ control conditions in a single compact file.
32
  - **Qwen-Image-2512-Fun-Controlnet-Union-2602**: Multi-condition ControlNet Union model (5 layer blocks) for the **Qwen-Image-2512** architecture.
33
  - **FLUX.1-dev-ControlNet-Union-Pro-2.0**: Next-generation unified 7-in-1 ControlNet for the **FLUX.1-dev** architecture by **Shakker Labs**.
@@ -38,6 +40,8 @@ This repository hosts high-quality **ConvRot INT8** quantized weights for all-in
38
 
39
  | Filename | Base Architecture | Base Model | Supported Conditions | Quantization | File Size | License |
40
  | :--- | :--- | :--- | :--- | :--- | :--- | :--- |
 
 
41
  | `controlnet-union-pro-max-sdxl-1.0_convrot_int8.safetensors` | SDXL 1.0 | [xinsir/controlnet-union-sdxl-1.0](https://huggingface.co/xinsir/controlnet-union-sdxl-1.0) | OpenPose, Depth, Canny, Lineart, Anime Lineart, Scribble, Soft Edge, Normal, Segment, Tile, Inpaint | ConvRot INT8 | ~1.30 GB | OpenRAIL++-M / Apache-2.0 |
42
  | `Qwen-Image-2512-Fun-Controlnet-Union-2602_convrot_int8.safetensors` | Qwen-Image-2512 | [alibaba-pai/Qwen-Image-2512-Fun-Controlnet-Union](https://huggingface.co/alibaba-pai/Qwen-Image-2512-Fun-Controlnet-Union) | Canny, HED, Depth, Pose, MLSD, Scribble, Gray, Inpaint | ConvRot INT8 | ~1.64 GB | Apache-2.0 |
43
  | `FLUX.1-dev-ControlNet-Union-Pro-2.0_convrot_int8.safetensors` | FLUX.1-dev | [Shakker-Labs/FLUX.1-dev-ControlNet-Union-Pro-2.0](https://huggingface.co/Shakker-Labs/FLUX.1-dev-ControlNet-Union-Pro-2.0) | Canny, Depth, Pose, Blur, Gray, Soft Edge, Low Quality | ConvRot INT8 | ~2.00 GB | Other / Non-Commercial |
@@ -46,7 +50,7 @@ This repository hosts high-quality **ConvRot INT8** quantized weights for all-in
46
 
47
  ## ๐Ÿ› ๏ธ Key Features
48
 
49
- - **All-in-One Multi-Condition Control**: Unified architectures enabling single or blended conditioning inputs (Pose, Depth, Canny, Tile, etc.) without switching heavy checkpoints during workflow execution.
50
  - **ConvRot INT8 Precision**: Leverages orthogonal Hadamard rotations to redistribute channel-wise outlier spikes uniformly across dimensions, preventing quantization error buildup in deep control layers.
51
  - **VRAM & Storage Optimization**: Slashes VRAM consumption and disk footprint by ~50% to 70% compared to unquantized FP16 checkpoints, allowing seamless multi-ControlNet workflows on consumer GPUs.
52
 
@@ -74,10 +78,11 @@ Place the downloaded `.safetensors` files into your ComfyUI `models/controlnet/`
74
  ## ๐Ÿ“œ Credits & License
75
 
76
  ### Base Models & Research
 
77
  - **SDXL ControlNet Union:** [xinsir/controlnet-union-sdxl-1.0](https://huggingface.co/xinsir/controlnet-union-sdxl-1.0) by **xinsir**
78
  - **Qwen-Image ControlNet Union:** [alibaba-pai/Qwen-Image-2512-Fun-Controlnet-Union](https://huggingface.co/alibaba-pai/Qwen-Image-2512-Fun-Controlnet-Union) & [aigc-apps/VideoX-Fun](https://github.com/aigc-apps/VideoX-Fun) (Apache-2.0)
79
  - **FLUX.1-dev ControlNet Union Pro 2.0:** [Shakker-Labs/FLUX.1-dev-ControlNet-Union-Pro-2.0](https://huggingface.co/Shakker-Labs/FLUX.1-dev-ControlNet-Union-Pro-2.0) & [InstantX Team](https://huggingface.co/InstantX) (FLUX.1-dev Non-Commercial License)
80
- - **Base Architectures:** [Stability AI SDXL 1.0](https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0) & [black-forest-labs/FLUX.1-dev](https://huggingface.co/black-forest-labs/FLUX.1-dev)
81
 
82
  ---
83
 
 
7
  - inpainting
8
  - sdxl
9
  - stable-diffusion-xl
10
+ - illustrious-xl
11
  - qwen-image
12
  - qwen-image-2512
13
  - flux
 
21
 
22
  # ControlNet Models (ConvRot INT8)
23
 
24
+ High-fidelity **ConvRot INT8** quantized weights for multi-condition ControlNet models across diverse generative architectures (**Illustrious-XL / SDXL**, **SDXL 1.0**, **Qwen-Image**, and **FLUX.1-dev**).
25
 
26
  ---
27
 
28
  ## ๐ŸŒŸ Model Overview
29
 
30
+ This repository hosts high-quality **ConvRot INT8** quantized weights for all-in-one ControlNet Union models. By applying orthogonal Hadamard rotation prior to per-channel INT8 quantization, these models effectively eliminate activation outlier distortion and drastically reduce VRAM and disk footprint while preserving precise structural control fidelity:
31
 
32
+ - **CN-anytest4_illustrious2 (Variants A & B)**: Multi-purpose all-in-one Anytest v4 ControlNet models fine-tuned for **Illustrious-XL / SDXL** by **2vXpSwA7**, offering high-precision anime and illustration structure guiding.
33
  - **controlnet-union-pro-max-sdxl-1.0**: The comprehensive all-in-one ControlNet Union model for **SDXL 1.0** by **xinsir**, supporting 10+ control conditions in a single compact file.
34
  - **Qwen-Image-2512-Fun-Controlnet-Union-2602**: Multi-condition ControlNet Union model (5 layer blocks) for the **Qwen-Image-2512** architecture.
35
  - **FLUX.1-dev-ControlNet-Union-Pro-2.0**: Next-generation unified 7-in-1 ControlNet for the **FLUX.1-dev** architecture by **Shakker Labs**.
 
40
 
41
  | Filename | Base Architecture | Base Model | Supported Conditions | Quantization | File Size | License |
42
  | :--- | :--- | :--- | :--- | :--- | :--- | :--- |
43
+ | `CN-anytest4_illustrious2_A_convrot_int8.safetensors` | Illustrious-XL / SDXL | [2vXpSwA7/iroiro-lora (Anytest v4 Variant A)](https://huggingface.co/2vXpSwA7/iroiro-lora) | Multi-condition (Canny, Lineart, Depth, Pose, Structure) | ConvRot INT8 | ~1.30 GB | Fair AI / OpenRAIL++-M |
44
+ | `CN-anytest4_illustrious2_B_convrot_int8.safetensors` | Illustrious-XL / SDXL | [2vXpSwA7/iroiro-lora (Anytest v4 Variant B)](https://huggingface.co/2vXpSwA7/iroiro-lora) | Multi-condition (Canny, Lineart, Depth, Pose, Structure) | ConvRot INT8 | ~1.30 GB | Fair AI / OpenRAIL++-M |
45
  | `controlnet-union-pro-max-sdxl-1.0_convrot_int8.safetensors` | SDXL 1.0 | [xinsir/controlnet-union-sdxl-1.0](https://huggingface.co/xinsir/controlnet-union-sdxl-1.0) | OpenPose, Depth, Canny, Lineart, Anime Lineart, Scribble, Soft Edge, Normal, Segment, Tile, Inpaint | ConvRot INT8 | ~1.30 GB | OpenRAIL++-M / Apache-2.0 |
46
  | `Qwen-Image-2512-Fun-Controlnet-Union-2602_convrot_int8.safetensors` | Qwen-Image-2512 | [alibaba-pai/Qwen-Image-2512-Fun-Controlnet-Union](https://huggingface.co/alibaba-pai/Qwen-Image-2512-Fun-Controlnet-Union) | Canny, HED, Depth, Pose, MLSD, Scribble, Gray, Inpaint | ConvRot INT8 | ~1.64 GB | Apache-2.0 |
47
  | `FLUX.1-dev-ControlNet-Union-Pro-2.0_convrot_int8.safetensors` | FLUX.1-dev | [Shakker-Labs/FLUX.1-dev-ControlNet-Union-Pro-2.0](https://huggingface.co/Shakker-Labs/FLUX.1-dev-ControlNet-Union-Pro-2.0) | Canny, Depth, Pose, Blur, Gray, Soft Edge, Low Quality | ConvRot INT8 | ~2.00 GB | Other / Non-Commercial |
 
50
 
51
  ## ๐Ÿ› ๏ธ Key Features
52
 
53
+ - **All-in-One Multi-Condition Control**: Unified architectures enabling single or blended conditioning inputs (Pose, Depth, Lineart, Canny, Tile, etc.) without switching heavy checkpoints during workflow execution.
54
  - **ConvRot INT8 Precision**: Leverages orthogonal Hadamard rotations to redistribute channel-wise outlier spikes uniformly across dimensions, preventing quantization error buildup in deep control layers.
55
  - **VRAM & Storage Optimization**: Slashes VRAM consumption and disk footprint by ~50% to 70% compared to unquantized FP16 checkpoints, allowing seamless multi-ControlNet workflows on consumer GPUs.
56
 
 
78
  ## ๐Ÿ“œ Credits & License
79
 
80
  ### Base Models & Research
81
+ - **Illustrious Anytest ControlNet:** [2vXpSwA7/iroiro-lora](https://huggingface.co/2vXpSwA7/iroiro-lora) by **2vXpSwA7**
82
  - **SDXL ControlNet Union:** [xinsir/controlnet-union-sdxl-1.0](https://huggingface.co/xinsir/controlnet-union-sdxl-1.0) by **xinsir**
83
  - **Qwen-Image ControlNet Union:** [alibaba-pai/Qwen-Image-2512-Fun-Controlnet-Union](https://huggingface.co/alibaba-pai/Qwen-Image-2512-Fun-Controlnet-Union) & [aigc-apps/VideoX-Fun](https://github.com/aigc-apps/VideoX-Fun) (Apache-2.0)
84
  - **FLUX.1-dev ControlNet Union Pro 2.0:** [Shakker-Labs/FLUX.1-dev-ControlNet-Union-Pro-2.0](https://huggingface.co/Shakker-Labs/FLUX.1-dev-ControlNet-Union-Pro-2.0) & [InstantX Team](https://huggingface.co/InstantX) (FLUX.1-dev Non-Commercial License)
85
+ - **Base Architectures:** [Illustrious-XL](https://civitai.red/models/675574/illustrious-xl), [Stability AI SDXL 1.0](https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0) & [black-forest-labs/FLUX.1-dev](https://huggingface.co/black-forest-labs/FLUX.1-dev)
86
 
87
  ---
88