fermatresearch/sdxl-weighting-prompts 🖼️🔢📝❓✓ → 🖼️

▶️ 3.5K runs 📅 Aug 2023 ⚙️ Cog 0.8.6 🔗 GitHub

image-inpainting image-to-image text-to-image

Performance

8.4sTypical run time

3.5KTotal runs

About

SDXL with prompt weighting available using Compel's syntax. Check the Github link for the docs.

Example Output

Prompt:

"a legendary bird is flying under the sea water staring at a multicolor sky, wallpaper, (corals and fish are swimming around)0.4"

Output

Performance Metrics

8.36s Prediction Time

8.32s Total Time

All Input Parameters

{
  "seed": 12345,
  "width": 1024,
  "height": 1024,
  "prompt": "a legendary bird is flying under the sea water staring at a multicolor sky, wallpaper, (corals and fish are swimming around)0.4",
  "refine": "expert_ensemble_refiner",
  "scheduler": "K_EULER",
  "lora_scale": 0.6,
  "num_outputs": 1,
  "guidance_scale": 7.5,
  "apply_watermark": true,
  "high_noise_frac": 0.8,
  "negative_prompt": "ugly, blurry, realistic",
  "prompt_strength": 0.8,
  "prompt_weighting": true,
  "num_inference_steps": 50
}

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 astronaut riding a rainbow unicorn: Input prompt
refine Default: expert_ensemble_refiner: Which refine style to use
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
prompt_weighting Type: booleanDefault: false: Use Compel for prompt weighting
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

Using seed: 12345
Prompt: a legendary bird is flying under the sea water staring at a multicolor sky, wallpaper, (corals and fish are swimming around)0.4
Using Compel for prompt embeddings
txt2img mode
Prompt embeddings calculated by Compel
  0%|          | 0/40 [00:00<?, ?it/s]
  2%|▎         | 1/40 [00:00<00:05,  6.87it/s]
  5%|▌         | 2/40 [00:00<00:04,  7.66it/s]
  8%|▊         | 3/40 [00:00<00:04,  8.07it/s]
 10%|█         | 4/40 [00:00<00:04,  8.27it/s]
 12%|█▎        | 5/40 [00:00<00:04,  8.38it/s]
 15%|█▌        | 6/40 [00:00<00:04,  8.45it/s]
 18%|█▊        | 7/40 [00:00<00:03,  8.50it/s]
 20%|██        | 8/40 [00:00<00:03,  8.42it/s]
 22%|██▎       | 9/40 [00:01<00:03,  8.47it/s]
 25%|██▌       | 10/40 [00:01<00:03,  8.50it/s]
 28%|██▊       | 11/40 [00:01<00:03,  8.52it/s]
 30%|███       | 12/40 [00:01<00:03,  8.52it/s]
 32%|███▎      | 13/40 [00:01<00:03,  8.53it/s]
 35%|███▌      | 14/40 [00:01<00:03,  8.54it/s]
 38%|███▊      | 15/40 [00:01<00:02,  8.42it/s]
 40%|████      | 16/40 [00:01<00:02,  8.46it/s]
 42%|████▎     | 17/40 [00:02<00:02,  8.25it/s]
 45%|████▌     | 18/40 [00:02<00:02,  8.35it/s]
 48%|████▊     | 19/40 [00:02<00:02,  8.41it/s]
 50%|█████     | 20/40 [00:02<00:02,  8.45it/s]
 52%|█████▎    | 21/40 [00:02<00:02,  8.48it/s]
 55%|█████▌    | 22/40 [00:02<00:02,  8.51it/s]
 57%|█████▊    | 23/40 [00:02<00:02,  8.50it/s]
 60%|██████    | 24/40 [00:02<00:01,  8.52it/s]
 62%|██████▎   | 25/40 [00:02<00:01,  8.50it/s]
 65%|██████▌   | 26/40 [00:03<00:01,  8.51it/s]
 68%|██████▊   | 27/40 [00:03<00:01,  8.53it/s]
 70%|███████   | 28/40 [00:03<00:01,  8.54it/s]
 72%|███████▎  | 29/40 [00:03<00:01,  8.55it/s]
 75%|███████▌  | 30/40 [00:03<00:01,  8.55it/s]
 78%|███████▊  | 31/40 [00:03<00:01,  8.55it/s]
 80%|████████  | 32/40 [00:03<00:00,  8.53it/s]
 82%|████████▎ | 33/40 [00:03<00:00,  8.53it/s]
 85%|████████▌ | 34/40 [00:04<00:00,  8.50it/s]
 88%|████████▊ | 35/40 [00:04<00:00,  8.34it/s]
 90%|█████████ | 36/40 [00:04<00:00,  8.40it/s]
 92%|█████████▎| 37/40 [00:04<00:00,  8.44it/s]
 95%|█████████▌| 38/40 [00:04<00:00,  8.47it/s]
 98%|█████████▊| 39/40 [00:04<00:00,  8.49it/s]
100%|██████████| 40/40 [00:04<00:00,  8.44it/s]
100%|██████████| 40/40 [00:04<00:00,  8.43it/s]
  0%|          | 0/10 [00:00<?, ?it/s]
 10%|█         | 1/10 [00:00<00:01,  7.72it/s]
 20%|██        | 2/10 [00:00<00:01,  7.67it/s]
 30%|███       | 3/10 [00:00<00:00,  7.67it/s]
 40%|████      | 4/10 [00:00<00:00,  7.67it/s]
 50%|█████     | 5/10 [00:00<00:00,  7.67it/s]
 60%|██████    | 6/10 [00:00<00:00,  7.68it/s]
 70%|███████   | 7/10 [00:00<00:00,  7.69it/s]
 80%|████████  | 8/10 [00:01<00:00,  7.68it/s]
 90%|█████████ | 9/10 [00:01<00:00,  7.67it/s]
100%|██████████| 10/10 [00:01<00:00,  7.68it/s]
100%|██████████| 10/10 [00:01<00:00,  7.68it/s]

Version Details

Version ID: 66175a2993706e1721076d5c7f92f0c81ec6d065ec20717527f05dd8528a1fc7
Version Created: August 17, 2023

Run on Replicate →