You can run it yourself here: https://huggingface.co/spaces/rerun/InstantMesh
That's InstantMesh, not MeshAnything.
The MeshAnything demo is at
MeshAnything – Converts 3D representations into efficient 3D meshes
31–40 of 71 posts
Re: MeshAnything – Converts 3D representations into efficient 3D meshes
#32Certainly a lot of scope for this kind of thing .. people who do lidar scans or photogrammetry of buildings tend to end up with very large meshes or very large point clouds.. which means they need souped up PCs and expensive software to wrangle into some usable CAD format. Its an area where things can be improved a lot imho - I did some work a while back fitting flat planes to pointclouds, and ended up with mesh mode…
Surely you heard about surfels, sounds like a very similar approach.
Looked up surfels .. I think they are different from fitting local planes - I was interested in the geometry, not the lighting.
But surfels seems like a clever idea, pre-dating view dependent NeRFs.
Re: MeshAnything – Converts 3D representations into efficient 3D meshes
#33I am all in for any development in this domain. Just to spread some sense of scale, We recently processed (manually) the point cloud scan of one of the ( So the scale shown in this paper feels like toys! Not undermining the effort at all. We need to start somewhere anyway. For the same reason, I feel puzzled looking at Industrial scenes in Video Games. They are like 3 order of magnitude simplified compared to a real…
Real life castles were designed to withstand a siege, video game castles are designed to give off a castle vibe. Once you've achieved that you stop adding stuff, as anything beyond that just creates problems - you start killing performance, visibility starts to suffer, it's not clear what's interactive and what is decoration, gameplay starts to take a hit as the AI and player start getting stuck in the clutter, etc, etc...
Most people don't care as they don't have deep knowledge of how a castle or a power plant really functions, you only notice oversimplifications in media in the field you work in.
It's also very likely the designers and artists didn't have time to do much research, and the whole thing is based off a Pinterest reference board.
Re: MeshAnything – Converts 3D representations into efficient 3D meshes
#34The converted meshes are not efficient. They are also full of n-gons so you need to retopo no matter what...
I can only see tris. It's quite rare to see .obj files with anything except triangles. Or do you mean something like "contains implied ngons because of the way coplanar tri faces are arranged"?
When working with meshes what you generally want is is quads, not triangles. The reason is that quads form nice closed loops.
Further more you would only allow quads to meet in 3, 4, or 5 edges per vertex. The 4 edges per vertex is the "normal" case that most of your mesh should have, it causes a regular grid of parabolic (euclidian) geometry with neutral curvature. Then patches of these meet in vertices with 3 edges to make it elliptic geometry with positive curvature or 5 edges to make it hyperbolic geometry with negative curvature.
You can ignore all of these and just randomly connect nearest neighbors to form triangles. But, then you still have only geometry, no useful topology, so not any better than a point cloud. A good topology is necessary for texturing, skinning, animation etc.
Re: MeshAnything – Converts 3D representations into efficient 3D meshes
#35Earlier quoted context omitted.
I can only see tris. It's quite rare to see .obj files with anything except triangles. Or do you mean something like "contains implied ngons because of the way coplanar tri faces are arranged"?
When working with meshes what you generally want is is quads, not triangles. The reason is that quads form nice closed loops. Further more you would only allow quads to meet in 3, 4, or 5 edges per vertex. The 4 edges per vertex is the "normal" case that most of your mesh should have, it causes a regular grid of parabolic (euclidian) geometry with neutral curvature. Then patches of these meet in vertices with 3 edges…
Sure. I (mostly) knew all that. I was specifically asking why you said "thy are full of n-gons" - my understanding of the terminology seems to be different to yours in that "n-gons" means "5 or more sides on a face". i.e. not a tri or a quad.
Re: MeshAnything – Converts 3D representations into efficient 3D meshes
#36Earlier quoted context omitted.
When working with meshes what you generally want is is quads, not triangles. The reason is that quads form nice closed loops. Further more you would only allow quads to meet in 3, 4, or 5 edges per vertex. The 4 edges per vertex is the "normal" case that most of your mesh should have, it causes a regular grid of parabolic (euclidian) geometry with neutral curvature. Then patches of these meet in vertices with 3 edges…
Sure. I (mostly) knew all that. I was specifically asking why you said "thy are full of n-gons" - my understanding of the terminology seems to be different to yours in that "n-gons" means "5 or more sides on a face". i.e. not a tri or a quad.
Your definition is right. Look at the produced models and not the images in the paper.
Re: MeshAnything – Converts 3D representations into efficient 3D meshes
#37Definitely the best result for low polygon creation I've seen. Great job!
Still triangles rather than polygons, but we are getting closer.
The end goal should be:
1) Polygons, mostly 4 sided, rather than triangles.
2) Edge smoothness/creases to separate hard coders from soft corners. (Which when combined with polygons enables SubD support: https://graphics.pixar.com/opensubdiv/docs/subdivision_surfa...)
3) UV for textures that are aligned with the natural flow of textures on those components.
4) Repeating textures (although sometimes not) that work with the UVs and combine to create PBR textures. (Getting closer all the time: https://gvecchio.com/stablematerials/)
After the above works, I think people should move on to inferring proper CAD models from an image. Basically infer all the constraints and the various construction steps.
Re: MeshAnything – Converts 3D representations into efficient 3D meshes
#38I am all in for any development in this domain. Just to spread some sense of scale, We recently processed (manually) the point cloud scan of one of the ( So the scale shown in this paper feels like toys! Not undermining the effort at all. We need to start somewhere anyway. For the same reason, I feel puzzled looking at Industrial scenes in Video Games. They are like 3 order of magnitude simplified compared to a real…
Could you go into some details here?
- sensors used
- postprocessing
- registration algorithm(s)
Are all things that would interest me greatly :)
Re: MeshAnything – Converts 3D representations into efficient 3D meshes
#39Definitely the best result for low polygon creation I've seen. Great job! Still triangles rather than polygons, but we are getting closer. The end goal should be: 1) Polygons, mostly 4 sided, rather than triangles. 2) Edge smoothness/creases to separate hard coders from soft corners. (Which when combined with polygons enables SubD support: https://graphics.pixar.com/opensubdiv/docs/subdivision_surfa... ) 3) UV for te…
why are polygons preferred over triangles?
Re: MeshAnything – Converts 3D representations into efficient 3D meshes
#40Definitely the best result for low polygon creation I've seen. Great job! Still triangles rather than polygons, but we are getting closer. The end goal should be: 1) Polygons, mostly 4 sided, rather than triangles. 2) Edge smoothness/creases to separate hard coders from soft corners. (Which when combined with polygons enables SubD support: https://graphics.pixar.com/opensubdiv/docs/subdivision_surfa... ) 3) UV for te…
why are polygons preferred over triangles?
It's much easier and cleaner to subdivide quads to refine shapes when modeling. For example, you can split the quads along an entire edge to get a new clean edge for manipulation (ex. to bevel it). If you try to do the same with triangles, you get a jagged mess.