tgohblio/instant-id-albedobase-xl 🖼️🔢📝✓ → 🖼️

▶️ 180.0K runs 📅 Jan 2024 ⚙️ Cog 0.9.4 🔗 GitHub 📄 Paper ⚖️ License
image-consistent-character-generation image-to-image

About

InstantID : Zero-shot Identity-Preserving Generation in Seconds with ⚡️LCM-LoRA⚡️. Using AlbedoBase-XL v2.0 as base model.

Example Output

Prompt:

"A digital elven princess with vibrant anime art style shines in a palette of dark pink and teal, In the style of artists Shilin Huang and Guillaume Seignac. Exudes an aura of heavenly angel, perfectly blending the elements of sultriness and ethereal beauty. The eyes, brimming with life, sparkle with a mesmerizing shine, adding a touch of spectral fascination to her already captivating presence"

Output

Example output

Performance Metrics

16.78s Prediction Time
16.81s Total Time
All Input Parameters 
{
  "image": "https://replicate.delivery/pbxt/KIuaBkQJUNjzsAce8hZDbWmKAdOYIhoeCcyA4EBoguEnhjDU/bp_lisa.png",
  "width": 640,
  "height": 640,
  "prompt": "A digital elven princess with vibrant anime art style shines in a palette of dark pink and teal, In the style  of artists Shilin Huang and Guillaume Seignac. Exudes an aura of heavenly angel, perfectly blending the elements of sultriness and ethereal beauty. The eyes, brimming with life, sparkle with a mesmerizing shine, adding a touch of spectral fascination to her already captivating presence",
  "guidance_scale": 0,
  "safety_checker": true,
  "ip_adapter_scale": 0.2,
  "num_inference_steps": 6,
  "controlnet_conditioning_scale": 0.8
}
Input Parameters
image (required) Type: string
Input image
width Type: integerDefault: 640Range: 512 - 2048
Width of output image
height Type: integerDefault: 640Range: 512 - 2048
Height of output image
prompt Type: stringDefault: analog film photo of a man. faded film, desaturated, 35mm photo, grainy, vignette, vintage, Kodachrome, Lomography, stained, highly detailed, found footage, masterpiece, best quality
Input prompt
guidance_scale Type: numberDefault: 0Range: 0 - 10
Scale for classifier-free guidance. With LCM-LoRA, optimum is 0-5.
safety_checker Type: booleanDefault: true
Safety checker is enabled by default. Un-tick to expose unfiltered results.
negative_prompt Type: stringDefault:
Input Negative Prompt
ip_adapter_scale Type: numberDefault: 0.8Range: 0 - 1
Scale for IP adapter
num_inference_steps Type: integerDefault: 6Range: 1 - 30
Number of denoising steps. With LCM-LoRA, optimum is 6-8.
controlnet_conditioning_scale Type: numberDefault: 0.8Range: 0 - 1
Scale for ControlNet conditioning
Output Schema

Output

Type: string • Format: uri

Example Execution Logs 
The following part of your input was truncated because CLIP can only handle sequences up to 77 tokens: ['ation to her already captivating presence']
The following part of your input was truncated because CLIP can only handle sequences up to 77 tokens: ['ation to her already captivating presence']
  0%|          | 0/6 [00:00<?, ?it/s]
 17%|█▋        | 1/6 [00:00<00:00,  5.41it/s]
 33%|███▎      | 2/6 [00:00<00:00,  5.39it/s]
 50%|█████     | 3/6 [00:00<00:00,  5.37it/s]
 67%|██████▋   | 4/6 [00:00<00:00,  5.36it/s]
 83%|████████▎ | 5/6 [00:00<00:00,  5.35it/s]
100%|██████████| 6/6 [00:01<00:00,  5.34it/s]
100%|██████████| 6/6 [00:01<00:00,  5.36it/s]
Version Details
Version ID
2a2afbff09996b53247b0714577d4ff82d2c9da8e8b00c5499b5b34510bb8b5e
Version Created
January 28, 2024
Run on Replicate →