https://github.com/ChenDarYen/ComfyUI-TimestepShiftModel
ComfyUI implemtation for timestep shift used in NitroSD
https://github.com/ChenDarYen/ComfyUI-TimestepShiftModel
Last synced: 3 months ago
JSON representation
ComfyUI implemtation for timestep shift used in NitroSD
- Host: GitHub
- URL: https://github.com/ChenDarYen/ComfyUI-TimestepShiftModel
- Owner: ChenDarYen
- Created: 2025-01-06T21:18:05.000Z (3 months ago)
- Default Branch: main
- Last Pushed: 2025-01-06T23:08:00.000Z (3 months ago)
- Last Synced: 2025-01-06T23:36:52.520Z (3 months ago)
- Language: Python
- Size: 1.95 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
- awesome-comfyui - **ComfyUI-TimestepShiftModel** - Fidelity Single-Step Diffusion through Dynamic Adversarial Training.](https://arxiv.org/abs/2412.02030) (All Workflows Sorted by GitHub Stars)
README
# ComfyUI-TimestepShiftModel
This is a ComfyUI implementation of the timestep shift technique used in [NitroFusion: High-Fidelity Single-Step Diffusion through Dynamic Adversarial Training](https://arxiv.org/abs/2412.02030).For more details, visit the [official NitroFusion GitHub repository](https://github.com/ChenDarYen/NitroFusion).
## Usage
Clone this repository into the `ComfyUI/custom_nodes` directory:
```bash
git clone https://github.com/ChenDarYen/ComfyUI-TimestepShiftModel.git
```Download the NitroSD models from [Hugging Face](https://huggingface.co/ChenDY/NitroFusion) or use the following commands:
```bash
wget https://huggingface.co/ChenDY/NitroFusion/resolve/main/nitrosd-realism_comfyui.safetensors
wget https://huggingface.co/ChenDY/NitroFusion/resolve/main/nitrosd-vibrant_comfyui.safetensors
```The `Timestep Shift Model` node takes a model and a `shifted_timestep` value as input, producing a timestep-shifted model.
Apart from this step, the workflow is the same as a standard text-to-image generation workflow.
Have fun with the example workflow [ComfyUI_NitroSD_workflow.json](./ComfyUI_NitroSD_workflow.json)!
