Multi-Model 3D Generator: Same Input, Several Engines
Send one prompt or image to the local engine and every configured third-party engine, keep the results side by side in one viewport, and pay only for the engines you actually run.
Why one engine is a single point of failure
Every engine has jobs it handles badly. One under-segments hard-surface props, another flattens relief on statues, a third produces a clean mesh but weak UVs. When you only have one engine, those weaknesses become your problem and the only remedy is to re-roll and hope.
Multi-model mode turns that variance into a controlled choice. The input is written once; the engines run in parallel; the results land in one comparison view with identical viewport settings, which is the only way a side-by-side means anything.
How a comparison run is charged
Base cost per engine
a standard generation is about 10 credits on the base engine. Each additional engine in the same comparison adds that base cost again.
Texturing is included in the total quote
texturing adds 4 credits and HD texture another 4 to the full comparison quote; that total is allocated across tasks.
You keep one, the rest stay in your assets
keeping a result does not charge anything extra. The non-kept meshes remain downloadable from the task, so a failed comparison is still a comparison.
One pool, one ledger
every charge appears in the credit ledger with its task reference, so you can see which engine is costing you money.
A comparison workflow that saves credits
Run the first pass in Speed mode with texturing off. Compare silhouettes in wireframe and X-ray, decide which engine understood the brief, then re-run only the winner in Quality with texture on. The extra credits spent on the losing engines are typically less than the cost of discovering the problem after a full-quality run.
Batch comparisons extend this to a product line: upload a folder, and each image goes through the same engine set with independent results.
Which engine wins which job
| Job | First engine to try | Why |
|---|---|---|
| Rigid, geometric, printable part | Local engine | parametric construction gives clean walls and predictable watertightness |
| Stylised character or creature | Tripo3D | strong silhouette handling on organic subjects |
| Statue, bust, portrait | Hitem3D | built for fine surface relief that survives printing |
| Textured standalone prop | Meshy | texture quality on flat surfaces |
| Realistic organic subject | Rodin | geometry accuracy on smooth, scanned-looking surfaces |
What comparison does not fix
Running five engines will not rescue a bad reference image or a vague prompt — it multiplies the same mistake five times. It also cannot compare engines whose API keys are not configured on your deployment; those are simply absent from the run.
Third-party engines need an API key configured by the operator. The built-in local engine is always available and produces real, downloadable meshes.
Frequently Asked Questions
What exactly is a multi-model generator?
One input routed to several generation engines, with the results shown together in a single comparison view. You are not aggregating files after the fact — the engines run from the same prompt or image set and the same settings.
How many engines can run on one input?
Up to five: the local engine plus Tripo3D, Meshy, Rodin and Hitem3D, minus any whose API key is not configured on the deployment.
Do I pay for all of them?
Yes, per engine. Each engine in a comparison charges the base generation cost, and any engine that ran with texturing charges the texture cost too. The credit ledger itemises every engine separately.
Are the results comparable if engines are different services?
The viewport, shading modes, camera and property readouts are identical, so topology, silhouette and surface detail can be judged directly. Textures differ by provider and are the one thing to judge by eye.
Can I run a batch through several engines?
Yes. Batch mode creates one independent task per uploaded image, and each of those tasks can use the same engine set.
What if an engine is not configured?
It is shown as not configured and cannot be selected. The local engine is always available, so a comparison run is never empty.