You can already do this today. Simply refer to your uploaded images as Image 1, Image 2, etc., in your prompt. The model understands which uploaded image you’re referring to.
For example:
Take the pose from Image 1 and use the character from Image 2.
This approach is already supported and works with uploaded reference images.
notsoangrypotato
•
Sep 1
In image editor it doesn’t work.
Arina
Team•
Sep 2
Dear @notsoangrypotato ,
The Seedream 5 Pro model works great for this task. Simply upload your image to the Image Editor and use a prompt like this:
“Enhance and restore the image quality. Improve sharpness, clarity, resolution, and fine details. Remove blur, noise, compression artifacts, grain, and other visible imperfections. Restore damaged or unclear areas naturally while preserving the original composition, colors, lighting, textures, facial features, and identity. Do not change the content, objects, proportions, pose, or overall appearance of the image. Keep the result realistic and faithful to the original, with clean, natural details and high resolution.”
Comments4
user847299
Aug 10
This would be useful for video generation also.
Arina
Aug 7
Dear Client,
Thank you for your suggestion!
You can already do this today. Simply refer to your uploaded images as Image 1, Image 2, etc., in your prompt. The model understands which uploaded image you’re referring to.
For example:
Take the pose from Image 1 and use the character from Image 2.
This approach is already supported and works with uploaded reference images.
notsoangrypotato
Sep 1
In image editor it doesn’t work.
Arina
Sep 2
Dear @notsoangrypotato ,
The Seedream 5 Pro model works great for this task. Simply upload your image to the Image Editor and use a prompt like this:
“Enhance and restore the image quality. Improve sharpness, clarity, resolution, and fine details. Remove blur, noise, compression artifacts, grain, and other visible imperfections. Restore damaged or unclear areas naturally while preserving the original composition, colors, lighting, textures, facial features, and identity. Do not change the content, objects, proportions, pose, or overall appearance of the image. Keep the result realistic and faithful to the original, with clean, natural details and high resolution.”