zsxkib/flash-face 🔢❓📝🖼️ → 🖼️

▶️ 4.8K runs 📅 Apr 2024 ⚙️ Cog v0.9.5+dev 🔗 GitHub 📄 Paper ⚖️ License
image-consistent-character-generation text-to-image

About

FlashFace: Human Image Personalization with High-fidelity Identity Preservation

Example Output

Output

Example outputExample output

Performance Metrics

22.48s Prediction Time
22.58s Total Time
All Input Parameters
{
  "seed": 0,
  "steps": 50,
  "num_sample": 2,
  "face_guidance": 2.5,
  "lamda_feature": 0.9,
  "output_format": "webp",
  "output_quality": 80,
  "negative_prompt": "nsfw",
  "positive_prompt": "A handsome young man with long brown hair is sitting in the desert",
  "reference_face_1": "https://replicate.delivery/pbxt/KnE00y2FMOW0170eWcdyDUe09FZLCKiJrDwYYCUp42yWdni0/avatar.png",
  "face_bounding_box": "[0., 0., 0., 0.]",
  "text_control_scale": 7.5,
  "default_negative_prompt": "blurry, ugly, tiling, poorly drawn hands, poorly drawn feet, poorly drawn face, out of frame, extra limbs, disfigured, deformed, body out of frame, bad anatomy, watermark, signature, cut off, low contrast, underexposed, overexposed, bad art, beginner, amateur, distorted face",
  "default_position_prompt": "best quality, masterpiece,ultra-detailed, UHD 4K, photographic",
  "step_to_launch_face_guidance": 700
}
Input Parameters
seed Type: integer
Random seed. Leave blank to randomize the seed
steps Type: integerDefault: 35
Number of steps
num_sample Type: integerDefault: 1
Number of generated images
face_guidance Type: numberDefault: 2.2
Reference guidance strength
lamda_feature Type: numberDefault: 0.9
Reference feature strength
output_format Default: webp
Format of the output images
output_quality Type: integerDefault: 80Range: 1 - 100
Quality of the output images, from 0 to 100. 100 is best quality, 1 is lowest quality.
negative_prompt Type: stringDefault: nsfw
Negative prompt
positive_prompt (required) Type: string
Positive prompt
reference_face_1 (required) Type: string
Reference face image 1
reference_face_2 Type: string
Reference face image 2
reference_face_3 Type: string
Reference face image 3
reference_face_4 Type: string
Reference face image 4
face_bounding_box Type: stringDefault: [0., 0., 0., 0.]
Face position
text_control_scale Type: numberDefault: 7.5
Text guidance strength
default_negative_prompt Type: stringDefault: blurry, ugly, tiling, poorly drawn hands, poorly drawn feet, poorly drawn face, out of frame, extra limbs, disfigured, deformed, body out of frame, bad anatomy, watermark, signature, cut off, low contrast, underexposed, overexposed, bad art, beginner, amateur, distorted face
Default negative prompt postfix
default_position_prompt Type: stringDefault: best quality, masterpiece,ultra-detailed, UHD 4K, photographic
Default positive prompt postfix
step_to_launch_face_guidance Type: integerDefault: 600
Step index to launch reference guidance
Output Schema

Output

Type: arrayItems Type: stringItems Format: uri

Example Execution Logs
[!] (<class 'str'>) pos_prompt=A handsome young man with long brown hair is sitting in the desert
[!] (<class 'str'>) neg_prompt=nsfw
[!] (<class 'int'>) steps=50
[!] (<class 'str'>) face_bbox=[0., 0., 0., 0.]
[!] (<class 'float'>) lamda_feat=0.9
[!] (<class 'float'>) face_guidence=2.5
[!] (<class 'int'>) num_sample=2
[!] (<class 'float'>) text_control_scale=7.5
[!] (<class 'int'>) seed=0
[!] (<class 'int'>) step_to_launch_face_guidence=700
[!] (<class 'PIL.PngImagePlugin.PngImageFile'>) reference_face_1=<PIL.PngImagePlugin.PngImageFile image mode=RGBA size=1132x1120 at 0x7F4BAE7D10F0>
[!] (<class 'NoneType'>) reference_face_2=None
[!] (<class 'NoneType'>) reference_face_3=None
[!] (<class 'NoneType'>) reference_face_4=None
[!] (<class 'str'>) default_pos_prompt=best quality, masterpiece,ultra-detailed, UHD 4K, photographic
[!] (<class 'str'>) default_neg_prompt=blurry, ugly, tiling, poorly drawn hands, poorly drawn feet, poorly drawn face, out of frame, extra limbs, disfigured, deformed, body out of frame, bad anatomy, watermark, signature, cut off, low contrast, underexposed, overexposed, bad art, beginner, amateur, distorted face
0
final pos_prompt:  A handsome young man with long brown hair is sitting in the desert, best quality, masterpiece,ultra-detailed, UHD 4K, photographic
final neg_prompt:  nsfw, blurry, ugly, tiling, poorly drawn hands, poorly drawn feet, poorly drawn face, out of frame, extra limbs, disfigured, deformed, body out of frame, bad anatomy, watermark, signature, cut off, low contrast, underexposed, overexposed, bad art, beginner, amateur, distorted face
detected 1 faces
[0.0, 0.0, 0.0, 0.0]
  0%|          | 0/50 [00:00<?, ?it/s]
  2%|▏         | 1/50 [00:00<00:14,  3.30it/s]
  4%|▍         | 2/50 [00:00<00:14,  3.31it/s]
  6%|▌         | 3/50 [00:00<00:14,  3.30it/s]
  8%|▊         | 4/50 [00:01<00:13,  3.31it/s]
 10%|█         | 5/50 [00:01<00:13,  3.30it/s]
 12%|█▏        | 6/50 [00:01<00:13,  3.29it/s]
 14%|█▍        | 7/50 [00:02<00:13,  3.28it/s]
 16%|█▌        | 8/50 [00:02<00:12,  3.29it/s]
 18%|█▊        | 9/50 [00:02<00:12,  3.28it/s]
 20%|██        | 10/50 [00:03<00:12,  3.28it/s]
 22%|██▏       | 11/50 [00:03<00:11,  3.29it/s]
 24%|██▍       | 12/50 [00:03<00:11,  3.29it/s]
 26%|██▌       | 13/50 [00:03<00:11,  3.29it/s]
 28%|██▊       | 14/50 [00:04<00:10,  3.29it/s]
 30%|███       | 15/50 [00:04<00:10,  3.29it/s]
 32%|███▏      | 16/50 [00:04<00:11,  2.96it/s]
 34%|███▍      | 17/50 [00:05<00:11,  2.76it/s]
 36%|███▌      | 18/50 [00:05<00:12,  2.64it/s]
 38%|███▊      | 19/50 [00:06<00:12,  2.54it/s]
 40%|████      | 20/50 [00:06<00:12,  2.49it/s]
 42%|████▏     | 21/50 [00:07<00:11,  2.46it/s]
 44%|████▍     | 22/50 [00:07<00:11,  2.44it/s]
 46%|████▌     | 23/50 [00:07<00:11,  2.42it/s]
 48%|████▊     | 24/50 [00:08<00:10,  2.42it/s]
 50%|█████     | 25/50 [00:08<00:10,  2.41it/s]
 52%|█████▏    | 26/50 [00:09<00:09,  2.40it/s]
 54%|█████▍    | 27/50 [00:09<00:09,  2.40it/s]
 56%|█████▌    | 28/50 [00:10<00:09,  2.39it/s]
 58%|█████▊    | 29/50 [00:10<00:08,  2.39it/s]
 60%|██████    | 30/50 [00:10<00:08,  2.39it/s]
 62%|██████▏   | 31/50 [00:11<00:07,  2.39it/s]
 64%|██████▍   | 32/50 [00:11<00:07,  2.39it/s]
 66%|██████▌   | 33/50 [00:12<00:07,  2.39it/s]
 68%|██████▊   | 34/50 [00:12<00:06,  2.40it/s]
 70%|███████   | 35/50 [00:12<00:06,  2.39it/s]
 72%|███████▏  | 36/50 [00:13<00:05,  2.39it/s]
 74%|███████▍  | 37/50 [00:13<00:05,  2.39it/s]
 76%|███████▌  | 38/50 [00:14<00:05,  2.39it/s]
 78%|███████▊  | 39/50 [00:14<00:04,  2.39it/s]
 80%|████████  | 40/50 [00:15<00:04,  2.39it/s]
 82%|████████▏ | 41/50 [00:15<00:03,  2.39it/s]
 84%|████████▍ | 42/50 [00:15<00:03,  2.39it/s]
 86%|████████▌ | 43/50 [00:16<00:02,  2.39it/s]
 88%|████████▊ | 44/50 [00:16<00:02,  2.39it/s]
 90%|█████████ | 45/50 [00:17<00:02,  2.39it/s]
 92%|█████████▏| 46/50 [00:17<00:01,  2.39it/s]
 94%|█████████▍| 47/50 [00:17<00:01,  2.38it/s]
 96%|█████████▌| 48/50 [00:18<00:00,  2.39it/s]
 98%|█████████▊| 49/50 [00:18<00:00,  2.39it/s]
100%|██████████| 50/50 [00:19<00:00,  2.39it/s]
100%|██████████| 50/50 [00:19<00:00,  2.60it/s]
[~] Saving to /tmp/image_0.webp...
[~] Output format: WEBP
[~] Output quality: 80
Saved image 0 at /tmp/image_0.webp
[~] Saving to /tmp/image_1.webp...
[~] Output format: WEBP
[~] Output quality: 80
Saved image 1 at /tmp/image_1.webp
Version Details
Version ID
edb17f54faec253ee86e58e0b5f18f24a89c4e31fe7fcefa970e13d8ad934117
Version Created
April 22, 2024
Run on Replicate →