yuanrui-mdt-info/sd-1.5-architect-fastest 🔢🖼️📝❓ → 🖼️
Performance
1.2sTypical run time
314Total runs
About
Fast and cost-efficient SD 1.5 (LCM) model for interior and architectural image generation.
Example Output
Prompt:
"modern living room"
Output
Performance Metrics
1.21s
Prediction Time
1.22s
Total Time
All Input Parameters
{
"image": "https://replicate.delivery/pbxt/OHFJtXu9Sc3BW03AknWn05bf8Eu5kjTyJ3AWoRZUsn0F3yTK/ds001.jpg",
"prompt": "modern living room",
"strength": 0.78,
"num_images": 1,
"output_format": "png",
"guidance_scale": 1.5,
"output_quality": 90,
"negative_prompt": "messy, dirty, cartoon",
"num_inference_steps": 6
}
Input Parameters
- seed
- Random seed. Leave blank to randomize the seed
- image (required)
- Input image for img2img
- prompt
- Input prompt
- strength
- Strength of the img2img transformation (0.0 to 1.0)
- num_images
- Number of images to generate (1-4 recommended)
- output_format
- Format of the output image
- guidance_scale
- Guidance scale (CFG). For LCM, keep this low (1.0 - 2.0)
- output_quality
- Quality of the output image (0-100)
- negative_prompt
- Negative prompt
- num_inference_steps
- Number of inference steps. For LCM, 4-8 is usually sufficient.
Output Schema
Output
Example Execution Logs
Using seed: 10585 Input image resized to: 768x560 Active adapters: ['lcm'] Processing semantic segmentation... Processing MLSD... 0%| | 0/4 [00:00<?, ?it/s] 50%|█████ | 2/4 [00:00<00:00, 12.08it/s] 100%|██████████| 4/4 [00:00<00:00, 13.34it/s] 100%|██████████| 4/4 [00:00<00:00, 13.12it/s] Saved image 1/1: /tmp/out.png
Version Details
- Version ID
cb56fbb8eba885a772adc9087dd5559715260b7d1349d79ffa0a25335d5cbbca- Version Created
- January 19, 2026