A generated mesh is raw material. Shipping it means the same finishing work any outsourced or store-bought asset needs — the generator changes where the block of marble comes from, not the carving.
The minimum path
- Retopologise. Reconstruction density is arbitrary; a shipping asset needs a polygon count chosen for its role. See polygon budgets.
- Unwrap. Generated UV layouts are usually fragmented and unevenly scaled, which caps how good the texture can ever look.
- Set material channels. A colour map alone reads as flat under real lighting. Roughness and normal do most of the work of making a surface look like a material.
- Set scale and pivot. Generators have no unit system. A prop that imports at forty times its intended size, rotating around a point outside itself, is the most common first complaint.
- Add collision. Almost always a separate simplified shape, not the render mesh.
What you can often skip
- LODs, for props that are never seen at distance, or in projects where the engine generates them acceptably.
- Hand-painted detail, if the generated texture is close and the asset is not a hero.
- Custom rigs, for static props — most of a game's asset count never deforms.
Where generated assets are strongest
Volume work with low individual stakes: set dressing, background props, kitbash pieces, prototype stand-ins, variation on a theme. Forty crates that each need to be plausible and none of which need to be exactly right is precisely the shape of problem this solves.
Where they are weakest
Anything with an exact specification. A weapon that must match concept art line for line, a character with a defined silhouette, a modular kit whose pieces have to tile to the centimetre. Generators approximate; specifications do not tolerate approximation.
The honest accounting
The saving is real but it is not the whole asset. Expect a generated prop to reach shipping quality in a fraction of the time modelling it from scratch would take — but not zero, and the remaining work is skilled work. Teams that budget for "free assets" are the ones disappointed; teams that budget for a faster blockout are not.