adirik/inst-inpaint 🔢🖼️✓📝 → 🖼️

▶️ 595 runs 📅 Oct 2023 ⚙️ Cog 0.8.3 🔗 GitHub 📄 Paper ⚖️ License

image-editing image-object-removal

About

Inst-Inpaint: Instructing to Remove Objects with Diffusion Models

Example Output

Output

Performance Metrics

3.39s Prediction Time

3.38s Total Time

All Input Parameters

{
  "image": "https://replicate.delivery/pbxt/JdkrjeZNtrIY45NznFenZz0fXuRoZaYCLqTE18YGUwWHB3kE/bus-tree.jpg",
  "center_crop": false,
  "instruction": "Remove the red bus",
  "num_inference_steps": 30
}

Input Parameters

seed Type: integerRange: 0 - 100: Seed for the diffusion process. If not provided, it will be randomized.
image (required) Type: string: Input image
center_crop Type: booleanDefault: true: Whether to center crop the image
instruction (required) Type: string: Text instruction describing which object/s to remove
num_inference_steps Type: integerDefault: 30Range: 0 - 100: Number of diffusion steps

Output Schema

Output

Type: string • Format: uri

Example Execution Logs

Data shape for DDIM sampling is (1, 4, 32, 32), eta 0.0
Running DDIM Sampling with 31 timesteps
DDIM Sampler:   0%|          | 0/31 [00:00<?, ?it/s]
DDIM Sampler:  10%|▉         | 3/31 [00:00<00:01, 20.58it/s]
DDIM Sampler:  19%|█▉        | 6/31 [00:00<00:01, 20.55it/s]
DDIM Sampler:  29%|██▉       | 9/31 [00:00<00:01, 20.60it/s]
DDIM Sampler:  39%|███▊      | 12/31 [00:00<00:00, 20.37it/s]
DDIM Sampler:  48%|████▊     | 15/31 [00:00<00:00, 20.37it/s]
DDIM Sampler:  58%|█████▊    | 18/31 [00:00<00:00, 20.44it/s]
DDIM Sampler:  68%|██████▊   | 21/31 [00:01<00:00, 20.50it/s]
DDIM Sampler:  77%|███████▋  | 24/31 [00:01<00:00, 20.48it/s]
DDIM Sampler:  87%|████████▋ | 27/31 [00:01<00:00, 20.51it/s]
DDIM Sampler:  97%|█████████▋| 30/31 [00:01<00:00, 20.60it/s]
DDIM Sampler: 100%|██████████| 31/31 [00:01<00:00, 20.52it/s]

Version Details

Version ID: ee04e541de8045e9c38cd5b27da0387df8f653972f1aa2c89f029d572685a938
Version Created: October 4, 2023

Run on Replicate →