wty-ustc/hairclip 🖼️❓📝 → 🖼️

▶️ 299.6K runs 📅 Mar 2022 ⚙️ Cog 0.1.0-alpha 🔗 GitHub 📄 Paper ⚖️ License
hair-editing image-editing image-to-image

Performance

12.5sTypical run time
299.6KTotal runs

About

Design Your Hair by Text and Reference Image

Example Output

Output

Example output

Performance Metrics

12.45s Prediction Time
12.63s Total Time
All Input Parameters 
{
  "image": "https://replicate.delivery/mgxm/b8be17a7-abcb-4421-80f2-e6a1e3fe38c7/MarkZuckerberg.jpg",
  "editing_type": "both",
  "color_description": "blond",
  "hairstyle_description": "hi-top fade hairstyle"
}
Input Parameters
image (required) Type: string
Input image. Image will be aligned and resized. Output will be the concatenation of the inverted input and the image with edited hair.
editing_type Default: hairstyle
Edit hairstyle or color or both.
color_description Type: string
Color text prompt, eg: purple, red, orange. Valid if editing_type is color or both.
hairstyle_description
Hairstyle text prompt. Valid if input_type is text or text_image.
Output Schema

Output

Type: string • Format: uri

Example Execution Logs 
Loading from checkpoint: pretrained_models/hairclip.pt
Aligned image has shape: (256, 256)
Latent code calculated!
Loading UNet for AvgLabLoss

  0%|          | 0/1 [00:00<?, ?it/s]
100%|██████████| 1/1 [00:02<00:00,  2.00s/it]
100%|██████████| 1/1 [00:02<00:00,  2.00s/it]
Version Details
Version ID
b95cb2a16763bea87ed7ed851d5a3ab2f4655e94bcfb871edba029d4814fa587
Version Created
March 27, 2022
Run on Replicate →