Seamless Blend (subjects)

The Seamless Blend (subjects) mode is built for realistic composites from multiple photos, with a key difference vs Compose multiple images:

• in Compose you describe the scene and let the model merge;
• in Seamless Blend you decide which subjects to use and where to place them, thanks to a canvas with placement tools (scale, rotation, opacity) and a final harmonization phase.

In other words: Seamless Blend is the studio workflow when you want control over the layout and then want the AI to do the hard part: clean edges, consistent light, believable shadows, natural integration.

Use Seamless Blend if:

• you want to insert people/objects into a scene with control over position and size;
• you want to “move” a subject from one photo to another;
• you want a multi-subject composition while keeping manual control;
• you want to fix a composite that Compose did well but needs detail cleanup.

• If you just need an automatic merge described in words: use Compose multiple images.
• If you need precision edits on a single photo: use Single image edit.
• If you only need to change the background while keeping the subject: use Background edit (fast).

• Main image: optional (you can work on an empty canvas).
• Subject photos: up to 5 (app limit).
• For each subject you can use:
  • Full (use the whole subject photo), or
  • Mask (cut out the subject with a cleaner extraction).

Seamless Blend works in two phases:

1. Composite preparation (manual): upload, cut out (if needed), position, and arrange subjects on the canvas.
2. Harmonization (AI): when you generate, the AI:
   • cleans edges and cutout artifacts,
   • matches light and color,
   • adds shadows and integration,
   • rebuilds any missing background coherently,
   • makes the scene look like a single photo.

If you do not write a prompt, a default harmonization prompt is used (built for this phase).

In this mode you typically see:

• Canvas / scene: preview where you place subjects.
• Main upload: optional, sets the base background.
• Add subjects: upload up to 5 subject photos.
• Mask / Full for each subject:
  • Mask -> open the mask editor and paint the subject to extract;
  • Full -> use the whole image (faster, less precise).
• Placement (when selecting a subject on the canvas):
  • Size (scale),
  • Rotation,
  • Opacity,
  • Remove (delete the subject).
• Edit composite (optional): opens a markup phase to refine the composite.
• Seamless Blend: starts final harmonization.

1. Upload a main image if you want a specific context (otherwise use empty canvas).
2. Add 1-5 subjects.
3. For each subject:
   • use Mask if you want a clean cutout,
   • use Full only if it is ok to take the whole image (e.g. object on a uniform background).
4. Place subjects on the canvas and adjust scale/rotation.
5. (Optional) Open Edit composite to refine with markup.
6. Generate the output (harmonization).
7. If needed, generate variants or use “Use as input” in Edit for final micro retouching.

In the “Mode” selector, choose Seamless Blend (subjects).

Upload an image if you want it as the base of the scene. If you do not upload it, you work on an empty canvas: useful for creating scenes from scratch with cutout subjects.

Upload the subject images you want to insert. Tip: if possible use photos:

• with light similar to the final scene;
• with clear edges (easier cutouts).

For each subject choose:

• Mask:
  • open the mask editor,
  • paint the subject area,
  • apply to extract the cutout.
• Full:
  • use the whole subject photo (fast).

When to choose Mask:

• people (hair, shoulders, silhouette);
• objects on complex backgrounds;
• when you want clean edges.

Click Place and then:

• drag the subject to move it;
• select it to adjust:
  • Size,
  • Rotation,
  • Opacity.

Practical tip:

• use opacity < 1 temporarily to align the subject, then set it back to 1.

If you notice:

• awkward joins,
• messy overlaps,
• areas to rebuild, open Edit composite and use markup to indicate what to clean or how to harmonize.

Press Seamless Blend to harmonize the scene. If you do not write a prompt, the AI applies a default harmonization prompt.

• Generate 2-4 variants if you want to pick the best render (shadows, integration, colors).
• If only refinement is missing, use Use as input and do a pass in Single image edit.

You can often leave the prompt empty. Write it when you want to:

• define the background (if canvas is empty) or the style of the scene;
• request a specific atmosphere (studio, outdoor, cinematic);
• enforce constraints (realistic, no artifacts, no text).

Useful template: “Harmonize the composite realistically: clean edges, coherent light, believable shadows, same color temperature. No text or watermark.”

• Maximum limit: 5 subjects.
• If you place subjects with impossible scale/perspective, the AI may struggle to make them believable: help by specifying “realistic scale” in the prompt and choosing compatible inputs.
• If you want to change local details after harmonization, the cleanest path is: Use as input -> Single image edit.

• Use Mask to extract the subject better.
• Generate variants: often one handles edges better.
• After output, use “Use as input” and do Edit with markup on the edges.

• In the prompt write: “add shadows consistent with light direction”.
• Specify “light from left/right, soft”.

• Ask: “harmonize color temperature and exposure between subjects”.
• Choose inputs with similar balance when possible.

• Reduce overly strong contrast between subject and background.
• Specify: “natural integration, slight depth of field, coherent grain”.