Seedream 5.0 Pro AI Image Generator with References

Seedream 5.0 Pro fits campaign frames, editorial scenes, product compositions, and other briefs where several reference images help define the subject, materials, setting, or visual direction. It supports both prompt-only generation and reference-led image generation.

You can add up to 10 reference images. Accepted formats are JPEG, PNG, and WebP, with a 10 MB limit for each image. Use only the references that communicate a distinct subject, style, or composition requirement.

Basic produces a 1K result and starts at 7 credits. High produces a 2K result and starts at 14 credits. Reference-led jobs can cost more when several input images are used, so stabilize the direction before moving to High.

Seedream 5.0 Pro supports 1:1, 4:3, 3:4, 16:9, 9:16, 2:3, 3:2, and 21:9. Choose the delivery frame before writing fine composition details so the subject and negative space match the intended crop.

Choose a lighter model when you only need a quick low-cost concept, or use the separate Seedream layer decomposition workflow when the task is to split one existing image into reusable visual layers rather than create a new composition.

You can write the prompt, choose the frame and quality tier, select PNG or JPEG, and add up to 10 references before submitting a generation.

Yes. Use it as text-to-image with a prompt only, or add one or more references when the subject, material, composition, or visual direction needs stronger guidance.

Choose the aspect ratio first, then Basic 1K or High 2K, and finally review the reference set. Those decisions establish the delivery frame, credit tier, and amount of visual guidance before generation.

The model page is public, but uploading reference images and submitting a generation require sign-in so Rivya can manage credits and task history.

Use this page for the first direction test. Move into Studio when you need saved iterations, repeated variants, asset organization, or a longer campaign workflow around the same references.