Ideogram ghost mannequin API views vs Sume input references

Ideogram's POST /v2/tool/ghost-mannequin is a purpose-built tool: you upload garment photos in fields named for the camera direction and choose the output view. Sume has no tool with that name in the image docs; you pass garment photos as a flat input_references list on POST /v1/images and describe the ghost-mannequin result in the prompt.

Vendor details are from the Ideogram reference page; Sume details from the Image API docs, both read 2026-10-01.

What does the Ideogram tool take?

It accepts front_image, back_image, left_image, right_image, top_image and bottom_image, each optional, plus garment_images for photos when the direction is unknown. view is required. It is multipart only, returns a generation_id, and the output always has a clean white studio background.

What does Sume take?

ChatGPT Image 2.5 supports text-to-image and up to 16 image references. Reference URLs must be public HTTPS; localhost, private-network and non-HTTPS URLs are rejected. Models whose input_references range is 0 to 0 are text-to-image only and reject references, so read the catalog before sending photos.

How do I tell the model which photo is which?

Sume has no per-reference label field in the docs, so say it in the prompt: which reference is the front and which is the back, and the output you want. Keep the output description concrete, such as a single garment on a plain white background. Sume's catalog gates parameters, so confirm the model lists input_references first. For a fuller walkthrough of the use case, see AI ghost mannequin.

Can I get several views in one call?

The docs say you can request up to 10 images per call with n, and that per-model ceilings are lower, so read the n range for your model. Whether those images come out as different views depends on your prompt; the docs do not describe a view parameter.