cjwbw/diffae ❓🖼️🔢 → ❓

▶️ 17.1K runs 📅 Aug 2022 ⚙️ Cog 0.3.13 🔗 GitHub 📄 Paper ⚖️ License

face-editing image-editing image-to-image

About

Image Manipulatinon with Diffusion Autoencoders

Example Output

Output

[object Object][object Object]

Performance Metrics

31.80s Prediction Time

109.72s Total Time

All Input Parameters

{
  "T_inv": 200,
  "image": "https://replicate.delivery/mgxm/c4d3d37d-8545-4941-b6aa-fce61d7d0769/download.png",
  "T_step": 100,
  "target_class": "Bangs",
  "manipulation_amplitude": 0.3
}

Input Parameters

T_inv Default: 200
image (required) Type: string: Input image for face manipulation. Image will be aligned and cropped, output aligned and manipulated images.
T_step Default: 100: Number of step for generation.
target_class Default: Bangs: Choose manipulation direction.
manipulation_amplitude Type: numberDefault: 0.3Range: -0.5 - 0.5: When set too strong it would result in artifact as it could dominate the original image information.

Output Schema

Output

Type: array

Example Execution Logs

Aligning image...
Encoding and Manipulating the aligned image...
predict.py:125: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).
  convert_img = torch.tensor(img)

Version Details

Version ID: 5d917b91659e117aa8b0c5d6213077e9132083e4a8a272f344cc52c3ba2f6e98
Version Created: August 3, 2022

Run on Replicate →