Multi-Reference Image Generation: The Feature Your Workflow Needs

Watermark notice — Every image model on NovAI is served by a Chinese provider (Zhipu AI, ByteDance, Tencent) and burns a visible AI-generated label such as AI生成 into the corner of the output under China’s labelling rules — NovAI cannot switch it off. Video output varies by model: in our 4 August 2026 tests Seedance 2.0 and CogVideoX-Flash showed no visible label, while Hunyuan Video 1.5 is labelled. Text and chat models are not affected. Which models, and what it rules out →

Combine product photos, style guides, and portraits in a single API call. This changes everything for e-commerce and marketing automation.

Beyond Simple Text-to-Image

Standard AI image generators accept one prompt -> one image. Advanced models add one reference image for style transfer. Seedream 4.5 accepts up to 5 reference images and intelligently synthesizes them into a coherent output.

Use Case 1: E-Commerce Product Photography

Product photo on white background + lifestyle reference image = product naturally placed in scene. No Photoshop required:

{"model":"doubao-seedream-4.5",
 "prompt":"Natural lifestyle placement of the product",
 "reference_images":["https://cdn.example.com/product.jpg",
                     "https://cdn.example.com/lifestyle-room.jpg"],
 "n":4}

Use Case 2: Brand-Consistent Social Media

Style guide + product + setting = on-brand social posts, all programmatic:

{"model":"doubao-seedream-4.5",
 "prompt":"Summer sale announcement, beach theme",
 "reference_images":["https://brand.example.com/style-guide.jpg",
                     "https://brand.example.com/product-hero.jpg"]}

Use Case 3: Group Photo Composition

AI has historically struggled with group photos. Seedream 4.5 handles it by processing individual portraits and composing a natural group shot:

{"model":"doubao-seedream-4.5",
 "prompt":"Team photo in modern office, natural lighting",
 "reference_images":["person1.jpg","person2.jpg","person3.jpg"]}

Performance Notes

Frequently Asked Questions

What is multi-reference image generation?

Multi-reference image generation allows you to provide up to 5 reference images (product photos, style guides, portraits) and the AI generates a new image combining elements from all references. Seedream 4.5 supports this feature.

How many reference images can Seedream 4.5 accept?

Seedream 4.5 accepts up to 5 reference images in a single API call. You can combine product photos, style guides, and portrait images to generate customized output.

Can I use multi-reference generation via API?

Yes, Seedream 4.5's multi-reference feature is available through NovAI's API at /v1/images/generations. Pass reference images as base64-encoded data or URLs in the request body.

Related Articles

Related Articles

Start Building with NovAI

20 Chinese AI models -- Chat, Image Gen, Video Gen -- one API, $2 free credit.

Get Free $2Browse Models