qwen-edit-apps/qwen-image-edit-plus-lora-relight 🔢🖼️📝✓❓ → 🖼️

⭐ Official ▶️ 1.4K runs 📅 Nov 2025 ⚙️ Cog 0.16.9 📄 Paper ⚖️ License
image-editing image-re-lighting

About

Relight – Soft, curtain-filtered relighting that repaints the scene with golden-hour or moody tones using the Relight LoRA.

Example Output

Prompt:

"重新照明,使用窗帘透光(柔和漫射)的光线对图片进行重新照明, 带来温暖的日落氛围. Relight with soft window glow and golden-hour warmth."

Output

Example output

Performance Metrics

28.27s Prediction Time
28.28s Total Time
All Input Parameters 
{
  "image": "https://replicate.delivery/pbxt/O3fhvZI2Iz6iqfLvbqF6SFijFpV1q1M8VhhMoBv59EVLW25R/image.png",
  "prompt": "重新照明,使用窗帘透光(柔和漫射)的光线对图片进行重新照明, 带来温暖的日落氛围. Relight with soft window glow and golden-hour warmth.",
  "lora_scale": 1,
  "aspect_ratio": "match_input_image",
  "lora_weights": "",
  "output_format": "webp",
  "output_quality": 95,
  "true_guidance_scale": 1
}
Input Parameters
seed Type: integer
Random seed. Set for reproducible generation.
image (required) Type: string
Image file uploaded to Cog (jpeg, png, gif, or webp).
prompt Type: stringDefault: 重新照明,使用窗帘透光(柔和漫射)的光线对图片进行重新照明, 带来温暖的日落氛围. Relight with soft window glow and golden-hour warmth.
Prompt hint: paste `重新照明,使用窗帘透光(柔和漫射)的光线对图片进行重新照明` (meaning “relight the scene with soft curtain-filtered light”), then continue in English with the mood or color temperature you want.
go_fast Type: boolean
If num_inference_steps is omitted, true runs the 8-step Lightning preset and false runs the 40-step detailed preset.
lora_scale Type: numberRange: 0 - 4
Strength applied to the selected LoRA. Leave blank for the proxy default.
aspect_ratio Default: match_input_image
Aspect ratio for the generated image
lora_weights Type: stringDefault:
LoRA weights to apply. Leave blank to use the proxy's default adapter.
output_format Default: webp
Format of the output images.
output_quality Type: integerDefault: 95Range: 0 - 100
Quality when saving the output images, from 0 to 100. 100 is best quality, 0 is lowest quality. Not relevant for .png outputs.
num_inference_steps Type: integerRange: 1 - 40
Explicit denoising step count (1-40). Leave blank to use the go_fast presets (fast = 8 steps, slow = 40 steps).
true_guidance_scale Type: numberDefault: 1Range: 0 - 10
True classifier-free guidance scale passed to the pipeline.
disable_safety_checker Type: booleanDefault: false
Disable safety checker for generated images.
Output Schema

Output

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

Example Execution Logs 
Proxy call -> model=qwen/qwen-image-edit-plus-lora, adapter=dx8152/Qwen-Image-Edit-2509-Relight, scale=1.0
Version Details
Version ID
2603bdd352ad1082452cea8a059d4a1f58b609332e32c74fb7d10d3af8a4c016
Version Created
November 14, 2025
Run on Replicate →