"AI retopology in seconds" is a real feature. It does what the button says.
It is also not retopology. If you plan to bake a high-poly onto the result, that difference is the whole job.
I sell forty base mesh packs and have spent twenty years cleaning up geometry so a projection lands.
So when remesh buttons started appearing in every generator, I did not look at the wireframe. I looked at what happens when you bake a sculpt onto the mesh they hand back. Almost nobody publishes that test.
The vendors are not lying
Read the documentation instead of the landing page. Meshy's own retopology guide says remeshing rebuilds the polygon structure, which invalidates the old UV layout. Textures do not survive the trip.
3D AI Studio's remesh page says its texture option rebakes the original surface appearance onto the new topology. That is not projecting a sculpt through a cage you control — two operations wearing the same verb.
So both tools reduce polygons and rebuild a surface. Retopology, the way an artist uses the word, is deciding where the polygons go. Neither claims to do that.
The test is the bake, not the wireframe
Do this once, on an asset you have already shipped.
- Take a sculpt you have a known-good hand-built low-poly for.
- Remesh it at the vendor's own recommended budget, not one that flatters the result. 3D AI Studio suggests 5,000–15,000 faces for game assets; Meshy suggests around 50K for PC.
- Auto-unwrap and auto-repack the output. Fix nothing by hand. You are testing the tool.
- Bake normal, curvature, AO and world-space position from the same high-poly onto both low-polys. Same cage, same ray distance, same 4K output, same 4 px padding.
- Look at all four maps on their own, side by side, before you open a material.
If that step is unfamiliar, why mesh maps are not optional covers what each map is for.
The density is uniform. Your model is not.
A remesher spreads polygons by surface area. Your object needs them on the silhouette and wherever the form turns, and almost none in the middle of a flat plate.
At 10K, a hand-built low-poly spends nearly all of that on outline and transitions. A remesh spends it evenly, so the rim that draws the silhouette gets the same density as the flat back.
The normal map is asked to make up the difference and it cannot. A normal map fakes surface; it has never faked an outline. A faceted silhouette stays faceted forever.
Seams, smoothing and the cage
Three things break, in this order.
Hard edges without UV seams
Wherever the mesh has a smoothing split there must be a UV seam under it, or the bake writes a gradient across the break.
Auto-repack cannot know which edges are meant to be hard. You get back everything smooth or everything split.
Island count
Auto-unwrap optimises for packing efficiency, not for where a seam can be seen.
You get many small islands and seams running across surfaces the camera reads. At 4K with 4 px padding, every extra island is padding paid for out of texel density.
The cage
A remesh approximates volume. It is not projected to sit just outside the high-poly, so it floats outside the sculpt in one place and cuts inside it in another. No single ray distance captures both.
That is the smeared normal map students send me asking what went wrong. Nothing went wrong. The low-poly was never built against that high-poly.
"Quad-dominant" is my word too
I should be fair — my own base mesh listings say quad-dominant.
Edge loops matter for exactly two things: deformation, and subdivision or displacement. Both need loops that run along the form, not loops laid out to cover the surface evenly.
A solver optimising for even coverage gives you regular, beautiful, completely uninformed topology. It photographs better in a wireframe than a good low-poly does, which is why this argument keeps getting won by the wrong side.
Where I use it anyway
They are sold hardest for the one job they are worst at. Four places they earn their keep:
- Decimating a scan for a background prop the camera never approaches. Silhouette error nobody can see is not error.
- Mid and far LODs. Build LOD0 by hand and remesh everything under it. Uniform density is exactly correct there.
- Blockouts and kitbash filler, where the mesh is a placeholder for a decision, not for a bake.
- 3D printing, where topology carries no shading at all.
And one real workflow rather than a shortcut: pulling a shape out of a generator and modelling over it.
That is the subject of the AI 3D modeling and game-ready workflow course.
The part I put first is choosing which outputs to keep. Most of the time this workflow saves comes from throwing a bad result away early.
What would change my mind
One change would fix most of this: a remesh that takes the high-poly as an input and projects to it. Cage-aware, silhouette-aware, spending faces where the curvature is.
Meshy already notes that its Smart Topology only works at generation time. That tells you they know where the hard part is.
Until then, the wireframe screenshot is not the deliverable. The four mesh maps are. Bake it before you believe it.
