Zetta Np1.5 is live. It is the P line's second engine, and it exists for one reason: the meshes it builds can come back as quads.
Why that is the whole announcement
Every generator here has produced triangles. For a mesh going straight into a game engine that is fine — the engine triangulates anyway, so quads would be a courtesy it immediately undoes.
It is not fine for a mesh somebody is going to open and keep working on. Subdivision needs quads. Edge loops need quads. Retopology tools, deformation for a rig, a modeller selecting a loop and sliding it — all of it assumes four-sided faces. Hand a modeller a triangulated mesh and the first thing they do is rebuild it.
So a generated asset has usually been a finished thing you accept or reject. Np1.5 makes it a starting point.
And a higher ceiling
Np1 stops at 20,000 faces. Np1.5 goes to 50,000 as triangles, or 25,000 as quads.
The halving is the vendor's rule and it is worth understanding rather than working around: a quad carries more topological information than a triangle, so the same budget buys fewer of them. Twenty-five thousand quads is not a lesser mesh than fifty thousand triangles — for anything you intend to edit, it is the better one.
The polygon field narrows on its own when you switch topology. You cannot ask for a count the engine will refuse.
What has not changed
Np1.5 is a sibling of Np1, not a replacement. Both stay available, at the same price for everything they share.
The rest of the P line carries over intact — the 8K texture tier, the four multi-view slots, the polygon budget you name rather than a band you pick, and no geometry quality settings to reason about. If you know Np1, you know this engine; there is one new control.
Np1 is still the right answer when you want the tightest possible mesh and are not going to edit it. Np1.5 is the right answer the moment the file is going to be opened.
Two things worth knowing before you spend credits on it
A quad mesh comes back as FBX, not GLB. That is the vendor's output format for quad topology rather than a preference of ours, and it is the one place the choice changes something downstream — check your importer expects it.
The two P engines share a queue. They are metered as one series, so running Np1 and Np1.5 jobs at once does not get you twice the throughput.
Where to find it
In the 3D Studio, on the Generate Smart 3D tab. That tab used to name its engine and offer no choice; it now has a picker with both. It opens on Np1, so nothing changes unless you reach for it.
Available on every paid plan, exactly as Np1 is.