Comparing AI 3D generators side by side: what I learned building a multi-engine tool

I've been working on a problem that bugs anyone building with AI-generated 3D assets: every AI 3D generator gives you different results for the same prompt, and you have no idea which one is best until you've paid credits on all of them.

Tripo3D might nail a hard-surface prop but struggle with organic shapes. Meshy might produce cleaner topology on characters. Rodin might win on textures. Most of us end up with 3 subscriptions and 12 browser tabs, re-running the same prompt everywhere and eyeballing the results.

So I built Trify3D - a studio that runs the same prompt or reference image through multiple engines at once and puts the results side by side:

1. Same prompt, wildly different meshes. The variance across engines is much bigger than most people expect. A "stylized chibi archer" prompt gave us a cute low-poly figure, a realistic mini, and something cursed. Picking the right engine per asset type matters a lot.

2. Topology quality ≠ texture quality. The engine with the best-looking preview often had the worst wireframe. That's why the comparison viewer needed professional render modes, not just pretty previews.

3. STL export is the sleeper feature. We built it for game devs, but 3D printing folks became the loudest users. Going from photo → printable model is a workflow that barely existed a year ago.

The studio is a web app (no installs), engines are consumed via their APIs, and every generated model belongs to the user commercially. There's a free tier if you want to poke at it.

If you've tried AI 3D generation - or you're building something with three.js/Unity pipelines that consumes generated assets - I'd genuinely love feedback on the comparison workflow. What would make it more useful? 🙏