Make the reference, then make the model

Image Studio

Generate, edit and compose images in the same product that turns them into meshes — so a reference you are happy with is one click from a generation instead of one export away.

What it does

Generate

A picture from a description, in ten frame shapes from 9:16 to 21:9 and at three sizes.

Edit

Change a picture you already have. The same mechanism behind AI Texturing’s preview.

Compose

Several images into one — a subject, a style reference and a layout, resolved together.

Straight into 3D

A generated image is already in your library, so it can be the input to an image-to-3D job without leaving the browser.

How it works

  1. Describe the pictureFor a 3D reference, aim for one subject, evenly lit, on a plain background — the qualities that reconstruct well are not the ones that look best framed.
  2. Pick a shape and a sizeSquare by default, and 1K by default — it is the cheapest, which is what iteration should be.
  3. IterateDrafts at 1K, then one final at the size you actually need. Size is a pixel count, not a quality setting.
  4. Generate a model from itOr download it and use it elsewhere.

What people use it for

  • Producing a clean reference for image-to-3D from a description
  • Building a set of consistent views for multi-view reconstruction
  • Concepting a look before committing to a mesh
  • Making texture references for retexturing

What it cannot do

  • A picture that looks good is not automatically a picture that reconstructs well — dramatic lighting and tight crops both hurt
  • Consistency across separately generated views is not guaranteed, and views that disagree produce a confused mesh
  • Larger sizes cost proportionally more and are not a better picture of the same thing

The details

Modes
Generate, edit, compose
Frames
Ten, from 9:16 to 21:9
Sizes
1K, 2K, 4K
Default
Square, 1K