Image-to-Video
Diffusers
Diffusion Single File
English
Chinese
i2v
video generation
comfyui
distillation
LoRA
quantization
nvfp4
Instructions to use InsecureErasure/Wan2.1-I2V-14B-480P-StepDistill-CfgDistill-Lightx2v-NVFP4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use InsecureErasure/Wan2.1-I2V-14B-480P-StepDistill-CfgDistill-Lightx2v-NVFP4 with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline from diffusers.utils import load_image, export_to_video # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("InsecureErasure/Wan2.1-I2V-14B-480P-StepDistill-CfgDistill-Lightx2v-NVFP4", dtype=torch.bfloat16, device_map="cuda") pipe.to("cuda") prompt = "A man with short gray hair plays a red electric guitar." image = load_image( "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/guitar-man.png" ) output = pipe(image=image, prompt=prompt).frames[0] export_to_video(output, "output.mp4") - Diffusion Single File
How to use InsecureErasure/Wan2.1-I2V-14B-480P-StepDistill-CfgDistill-Lightx2v-NVFP4 with Diffusion Single File:
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js
- Notebooks
- Google Colab
- Kaggle
Update README.md
Browse filesAdd inference section
README.md
CHANGED
|
@@ -60,6 +60,23 @@ The table below details the quantization format applied per layer type across bl
|
|
| 60 |
| 16–22 | NVFP4 | FP8 | NVFP4 | FP8/NVFP4 (29/71) | NVFP4 | FP8 | NVFP4 | NVFP4 | NVFP4 | NVFP4 | FP8/NVFP4 (57/43) | FP8/NVFP4 (43/57) |
|
| 61 |
| 23–39 | NVFP4 | FP8 | NVFP4 | FP8/NVFP4 (12/88) | NVFP4 | FP8 | NVFP4 | NVFP4 | NVFP4 | NVFP4 | FP8/NVFP4 (35/65) | NVFP4 |
|
| 62 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 63 |
## License Agreement
|
| 64 |
This model is licensed under the [Apache 2.0 License](LICENSE.txt). You retain full ownership of your generated content, but are solely responsible for its use in compliance with the license terms and applicable laws.
|
| 65 |
|
|
|
|
| 60 |
| 16–22 | NVFP4 | FP8 | NVFP4 | FP8/NVFP4 (29/71) | NVFP4 | FP8 | NVFP4 | NVFP4 | NVFP4 | NVFP4 | FP8/NVFP4 (57/43) | FP8/NVFP4 (43/57) |
|
| 61 |
| 23–39 | NVFP4 | FP8 | NVFP4 | FP8/NVFP4 (12/88) | NVFP4 | FP8 | NVFP4 | NVFP4 | NVFP4 | NVFP4 | FP8/NVFP4 (35/65) | NVFP4 |
|
| 62 |
|
| 63 |
+
## Inference
|
| 64 |
+
The model can be used in ComfyUI with the following parameters, based on the distilled model's own recommendations:
|
| 65 |
+
|
| 66 |
+
| Parameter | Value |
|
| 67 |
+
|-----------|-------|
|
| 68 |
+
| Shift | 5.0 |
|
| 69 |
+
| Sampler | LCM |
|
| 70 |
+
| Scheduler | normal |
|
| 71 |
+
| CFG | 1.0 |
|
| 72 |
+
| Steps | 4 |
|
| 73 |
+
|
| 74 |
+
The combinations euler/simple and heun/linear_quadratic (sampler/scheduler) are also known to produce good results.
|
| 75 |
+
|
| 76 |
+
The model is designed to generate 81 frames and is not compatible with LoRAs. Sampling completes in under 60 seconds on an RTX 5060, making it possible to produce a full 81-frame video in under two minutes; with RIFE, those 81 frames convert to a 10-second video.
|
| 77 |
+
|
| 78 |
+
Abrupt camera movements or fast subject motion may produce artifacts. This is an inherent limitation of applying aggressive quantization to an already distilled model.
|
| 79 |
+
|
| 80 |
## License Agreement
|
| 81 |
This model is licensed under the [Apache 2.0 License](LICENSE.txt). You retain full ownership of your generated content, but are solely responsible for its use in compliance with the license terms and applicable laws.
|
| 82 |
|