wehub-resource-sync
40 lines
1.6 KiB
Markdown
40 lines
1.6 KiB
Markdown
# Sana Cartoon-Style Image Edit (MNN)
|
|
|
|
Sana is integrated into MNN for cartoon-style photo editing and face-style conversion on mobile/edge.
|
|
|
|
[中文文档](./README_ZH.md)
|
|
|
|
## Showcase
|
|
|
|
<p align="center">
|
|
<img width="40%" alt="Sana Showcase" src="https://meta.alicdn.com/data/mnn/assets/sana_showcase_android.jpg" style="margin: 0 10px;">
|
|
<img width="40%" alt="Sana Home Screenshot" src="https://meta.alicdn.com/data/mnn/assets/sana_showcase_ios.jpg" style="margin: 0 10px;">
|
|
</p>
|
|
|
|
## App Links
|
|
- [Android MNN LLM Chat](../../apps/Android/MnnLlmChat/README.md)
|
|
- [iOS MNN LLM Chat](../../apps/iOS/MNNLLMChat/README.md)
|
|
|
|
## Model Links
|
|
Cartoon-style edit model
|
|
|
|
- HuggingFace: [https://huggingface.co/taobao-mnn/MNN-Sana-Edit-V2](https://huggingface.co/taobao-mnn/MNN-Sana-Edit-V2)
|
|
- ModelScope: [https://modelscope.cn/models/MNN/MNN-Sana-Edit-V2](https://modelscope.cn/models/MNN/MNN-Sana-Edit-V2)
|
|
|
|
## Recommended Settings
|
|
- Input: use square images when possible; non-square input often gives worse results.
|
|
- Output: fixed at `512x512`.
|
|
- Prompt: fixed for this model pipeline. No extra prompt is needed; changing it may reduce quality.
|
|
- Step: `10` steps are recommended. Fewer steps usually reduce visual quality.
|
|
|
|
## Usage in MNN Chat App
|
|
1. Open `apps/Android/MnnLlmChat`.
|
|
2. Download model `MNN-Sana-Edit-V2` from model market.
|
|
3. Switch to Sana model in app.
|
|
4. Upload an input face image (square is preferred).
|
|
5. Keep prompt/settings as default and set step to `10`.
|
|
6. Run generation and review `512x512` output.
|
|
|
|
|
|
## Reference
|
|
- Sana paper: [https://arxiv.org/abs/2410.10629](https://arxiv.org/abs/2410.10629) |