zylim0702/sdxl-lora-customize-model 🖼️🔢📝❓✓ → 🖼️

▶️ 100.8K runs 📅 Aug 2023 ⚙️ Cog 0.7.2

About

Introducing a text-to-image AI that crafts stunning 1024x1024 visuals. Load LoRa models via URLs for instant outputs. Train using this link: https://replicate.com/zylim0702/sdxl-lora-customize-training.

Example Output

Prompt:

"a photo of TOK , white background, pink blazer, american hairstyle"

Output

Performance Metrics

35.84s Prediction Time

35.85s Total Time

All Input Parameters

{
  "width": 1024,
  "height": 1024,
  "prompt": "a photo of TOK , white background, pink blazer, american hairstyle",
  "refine": "no_refiner",
  "Lora_url": "https://replicate.delivery/pbxt/aEt6InPOlTpZJZ9Y1664pLuWBf5ZdIKdjDK4FmHVB2EA47tIA/trained_model.tar",
  "scheduler": "K_EULER",
  "lora_scale": 0.6,
  "num_outputs": 4,
  "guidance_scale": 7.5,
  "apply_watermark": true,
  "high_noise_frac": 0.8,
  "prompt_strength": 0.8,
  "num_inference_steps": 20
}

Input Parameters

mask Type: string: Input mask for inpaint mode. Black areas will be preserved, white areas will be inpainted.
seed Type: integer: Random seed. Leave blank to randomize the seed
image Type: string: Input image for img2img or inpaint mode
width Type: integerDefault: 1024: Width of output image
height Type: integerDefault: 1024: Height of output image
prompt Type: stringDefault: An TOK riding a rainbow unicorn: Input prompt
refine Default: no_refiner: Which refine style to use
Lora_url (required) Type: string: Load Lora model
scheduler Default: K_EULER: scheduler
lora_scale Type: numberDefault: 0.6Range: 0 - 1: LoRA additive scale. Only applicable on trained models.
num_outputs Type: integerDefault: 1Range: 1 - 4: Number of images to output.
refine_steps Type: integer: For base_image_refiner, the number of steps to refine, defaults to num_inference_steps
guidance_scale Type: numberDefault: 7.5Range: 1 - 50: Scale for classifier-free guidance
apply_watermark Type: booleanDefault: true: Applies a watermark to enable determining if an image is generated in downstream applications. If you have other provisions for generating or deploying images safely, you can use this to disable watermarking.
high_noise_frac Type: numberDefault: 0.8Range: 0 - 1: For expert_ensemble_refiner, the fraction of noise to use
negative_prompt Type: stringDefault:: Input Negative Prompt
prompt_strength Type: numberDefault: 0.8Range: 0 - 1: Prompt strength when using img2img / inpaint. 1.0 corresponds to full destruction of information in image
num_inference_steps Type: integerDefault: 50Range: 1 - 500: Number of denoising steps

Output Schema

Output

Type: array • Items Type: string • Items Format: uri

Example Execution Logs

Loading sdxl txt2img pipeline...
Loading pipeline components...:   0%|          | 0/7 [00:00<?, ?it/s]
Loading pipeline components...:  14%|█▍        | 1/7 [00:00<00:02,  2.77it/s]
Loading pipeline components...:  29%|██▊       | 2/7 [00:01<00:03,  1.48it/s]
Loading pipeline components...:  43%|████▎     | 3/7 [00:01<00:02,  1.90it/s]
Loading pipeline components...: 100%|██████████| 7/7 [00:01<00:00,  5.68it/s]
Loading pipeline components...: 100%|██████████| 7/7 [00:01<00:00,  3.95it/s]
Loading fine-tuned model
Does not have Unet. Assume we are using LoRA
Loading Unet LoRA
Loading SDXL img2img pipeline...
Loading SDXL inpaint pipeline...
Loading SDXL refiner pipeline...
Loading refiner pipeline...
Loading pipeline components...:   0%|          | 0/4 [00:00<?, ?it/s]
Loading pipeline components...:  25%|██▌       | 1/4 [00:00<00:01,  1.52it/s]
Loading pipeline components...: 100%|██████████| 4/4 [00:00<00:00,  5.71it/s]
Using seed: 8931
Prompt: a photo of <s0><s1> , white background, pink blazer, american hairstyle
txt2img mode
  0%|          | 0/20 [00:00<?, ?it/s]
  5%|▌         | 1/20 [00:00<00:18,  1.00it/s]
 10%|█         | 2/20 [00:01<00:17,  1.00it/s]
 15%|█▌        | 3/20 [00:02<00:16,  1.00it/s]
 20%|██        | 4/20 [00:03<00:15,  1.00it/s]
 25%|██▌       | 5/20 [00:04<00:14,  1.00it/s]
 30%|███       | 6/20 [00:05<00:13,  1.00it/s]
 35%|███▌      | 7/20 [00:06<00:12,  1.00it/s]
 40%|████      | 8/20 [00:07<00:12,  1.00s/it]
 45%|████▌     | 9/20 [00:08<00:11,  1.00s/it]
 50%|█████     | 10/20 [00:09<00:09,  1.00it/s]
 55%|█████▌    | 11/20 [00:10<00:08,  1.00it/s]
 60%|██████    | 12/20 [00:11<00:07,  1.00it/s]
 65%|██████▌   | 13/20 [00:12<00:06,  1.00it/s]
 70%|███████   | 14/20 [00:13<00:05,  1.00it/s]
 75%|███████▌  | 15/20 [00:14<00:04,  1.00it/s]
 80%|████████  | 16/20 [00:15<00:03,  1.00it/s]
 85%|████████▌ | 17/20 [00:16<00:02,  1.00it/s]
 90%|█████████ | 18/20 [00:17<00:01,  1.00it/s]
 95%|█████████▌| 19/20 [00:18<00:01,  1.00s/it]
100%|██████████| 20/20 [00:19<00:00,  1.00s/it]
100%|██████████| 20/20 [00:19<00:00,  1.00it/s]

Version Details

Version ID: 5a2b1cff79a2cf60d2a498b424795a90e26b7a3992fbd13b340f73ff4942b81e
Version Created: August 20, 2023

Run on Replicate →