360 option #3 · AI image-to-3D test

Air fryer — spin it

One packshot → AI 3D mesh → true 360 · Higgsfield image_to_3d · 30 cr · 2026-06-15

This is a real, interactive 3D model generated from the single air-fryer packshot — drag to rotate, or let it auto-spin. Unlike an AI video orbit, every angle is the same consistent object (no morphing), and the file is a GLB — the exact format Shopify uses for its 3D / AR product viewer.

Drag on the model to rotate it yourself. Loads ~5.6 MB — give it a moment.

The one source packshot it was built from

What to judge

Spin it all the way around and check:

The front — body shape, proportions, stainless finish (should be strong, it's the source view).
The back & sides — AI invented these; look for plausible vs. mushy geometry.
Fine detail — the touchscreen, the viewing window, the handle, the feet. Detailed parts are where image-to-3D usually softens.
Texture — does the surface read as brushed steel, or like melted plastic?

The verdict to reach: if the geometry holds and the unseen sides are plausible, this is the cheap true-360 path — 30 cr per hero item, a real GLB that doubles as Shopify AR ("view in your room"). If it's too soft, the accurate 360 falls back to Jessar turntables or commissioned 3D on the hero 40. Either way, this is a consistent 360 — no morphing — which the AI video orbit never was.

Note on this being the hardest case: the air fryer is detail-heavy (screen, window, handle) on purpose — the stress test. A simpler form (a kettle, a bowl, a board) would 3D-reconstruct cleaner, so judge this as the floor, not the ceiling. Local file: work/higgsfield_test/airfryer_3d_20260615/airfryer.glb