Input guidelines

Better inputs produce better geometry and textures. Use the clearest source material you have before spending credits on higher quality modes.

Supported image uploads

• File types: JPEG, JPG, PNG, and WebP.
• Maximum file size: 10MB per uploaded image.
• Best subjects: one main object, fully visible, sharp edges, and minimal background clutter.
• Avoid: heavy blur, extreme shadows, watermarks, cropped objects, duplicate objects, and busy scenes.

Single Image inputs

Use one clear image where the object fills the frame but is not cropped. Product photos, toys, furniture, props, and isolated objects usually work better than people, transparent glass, reflective metal, or thin wires.

Multiple Images inputs

Multiple Images works best when every upload shows the same object from a different angle. Do not mix different objects, styles, or scales in one task. Keep lighting and background consistent when possible.

Text prompts

Good text prompts describe the subject, shape, style, and important visual details.

Example:

A low-poly medieval treasure chest with brass corners, dark oak planks, a rounded lid, and subtle worn edges.

Avoid prompts that are too short, contradictory, or mostly about camera settings. The model should describe the object, not the scene around it.

Content safety

Inputs are checked by content safety systems before generation. If an image, image set, or prompt is blocked, revise the input and submit again. Credits are not deducted when content safety blocks the request before generation.