计划执行与检查节点
executing-plans
按既定实施计划完成工作,保持过程检查和反馈。
sdxl-ad-inpaint【效率与协作】SDXL Ad Inpaint | Product Ad Image Generation. Generate product advertisement images with inpainting using SDXL. Triggers: sdxl ad inpaint, product ad, ad generation, product photography, ad inpainting
独立 GitHub 快照 · 自动分类 · 以原作者文档为准
Generate product advertisement images with inpainting using SDXL. Place products in professional ad settings with configurable composition, refinement steps, and product fill options.
Requires an each::labs API key. Get one at eachlabs.ai.
curl -X POST https://api.eachlabs.ai/v1/prediction \
-H "Content-Type: application/json" \
-H "X-API-Key: $EACHLABS_API_KEY" \
-d '{
"model": "sdxl-ad-inpaint",
"version": "0.0.1",
"input": {
"prompt": "a premium skincare product on a marble surface with rose petals, luxury beauty advertisement, studio lighting",
"image": "https://example.com/product-cutout.jpg",
"guidance_scale": 7.5,
"num_inference_steps": 40,
"img_size": "1024, 1024"
}
}'
| Parameter | Type | Default | Description |
|---|---|---|---|
| apply_img | boolean | true | Apply image as input to guide the model |
| condition_scale | number | 0.9 | Condition scale for ControlNet strength |
| guidance_scale | number | 7.5 | Guidance scale for generation |
| image | string | - | Input image to guide the output |
| img_size | string | 1024, 1024 | Image size dimensions |
| negative_prompt | string | low quality, out of frame, illustration, 3d, sepia, painting, cartoons, sketch, watermark, text, Logo, advertisement | Negative prompt |
| num_inference_steps | integer | 40 | Number of inference steps |
| num_refine_steps | integer | 10 | Additional refinement passes |
| product_fill | string | Original | Product fill mode. enum: Original, 80, 70, 60, 50, 40, 30, 20 |
| prompt | string | - | Text prompt for ad generation |
| scheduler | string | K_EULER | Scheduler. enum: DDIM, DPMSolverMultistep, HeunDiscrete, KarrasDPM, K_EULER_ANCESTRAL, K_EULER, PNDM |
| seed | integer | - | Random seed for reproducibility |
Luxury product ad:
curl -X POST https://api.eachlabs.ai/v1/prediction \
-H "Content-Type: application/json" \
-H "X-API-Key: $EACHLABS_API_KEY" \
-d '{
"model": "sdxl-ad-inpaint",
"version": "0.0.1",
"input": {
"prompt": "a premium perfume bottle on a reflective gold surface, dramatic studio lighting, luxury advertisement, bokeh background",
"image": "https://example.com/perfume-bottle.jpg",
"guidance_scale": 8,
"num_inference_steps": 50,
"num_refine_steps": 15,
"product_fill": "70",
"scheduler": "KarrasDPM"
}
}'
Food product photography:
curl -X POST https://api.eachlabs.ai/v1/prediction \
-H "Content-Type: application/json" \
-H "X-API-Key: $EACHLABS_API_KEY" \
-d '{
"model": "sdxl-ad-inpaint",
"version": "0.0.1",
"input": {
"prompt": "a fresh juice bottle on a rustic wooden table with fresh fruits and herbs, natural daylight, food photography",
"image": "https://example.com/juice-bottle.jpg",
"guidance_scale": 7.5,
"num_inference_steps": 40,
"condition_scale": 0.85,
"seed": 42
}
}'