cjwbw/prompt-to-prompt 🔢📝❓ → ❓

▶️ 2.4K runs 📅 Oct 2022 ⚙️ Cog 0.4.4 🔗 GitHub 📄 Paper ⚖️ License
prompt-to-prompt text-to-image

About

Prompt-to-prompt image editing with cross-attention control

Example Output

Output

Performance Metrics

95.01s Prediction Time
276.18s Total Time
All Input Parameters
{
  "seed": 8888,
  "edited_prompt": "A painting of a lion eating a burger",
  "original_prompt": "A painting of a squirrel eating a burger",
  "prompt_edit_type": "Replacement",
  "self_replace_steps": 0.4,
  "cross_replace_steps": 0.8
}
Input Parameters
seed Type: integerDefault: 8888
Random seed. Leave blank to randomize the seed for original output. But make sure to use the same seed for original-updated prompt pair.
local_edit Type: string
Enable local editing. Provide the in the format of 'words_in_original_prompt | words_in_edited_prompt', and the rest content will be preserved.
edited_prompt Type: string
Prompted used for editing the original sd output image. If prompt_edit_type above is not set, then this field will be ignored. See below for more information for how to edit the prompt from the original prompt. For Re-weight, just provided words in proginal_prompt with new weights.
original_prompt Type: stringDefault: pink bear riding a bicycle
Input prompt used for the orinigal image
prompt_edit_type
Choose the type of the prompt editing. See below for more information. If you are generating the original output, leave this empty.
self_replace_steps Type: numberDefault: 0.4Range: 0 - 1
Self attention replace steps
cross_replace_steps Type: numberDefault: 0.8Range: 0 - 1
Cross attention replace steps
Output Schema

Output

Example Execution Logs
Using seed: 8888
  0%|          | 0/51 [00:00<?, ?it/s]
  0%|          | 0/51 [00:00<?, ?it/s]
Version Details
Version ID
77f9e56f3c0eb7e635d0197e192980173a48f414499ed07bbc80d5807bdb6191
Version Created
October 29, 2022
Run on Replicate →