Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

MeshAnything – Converts 3D representations into efficient 3D meshes

buaacyw.github.io

Post reply on HN
Print view View original

41–50 of 71 posts

Re: MeshAnything – Converts 3D representations into efficient 3D meshes

#41

Post by Bjartr »

nkrisc wrote: post #30
ramshanker wrote: post #25

I 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…

> 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 plant. Because they are games, not oil refinery simulators. They are typically intending to only convey a general sense of “industrial environment” and nothing more. Do your models of oil refineries include the correct grass and other plant species growing in cracks in th…

> they are games, not oil refinery simulators.

That's an excellent point. I do feel compelled to mention the exception of oil refinery simulator games. Maxis (of SimCity, The Sims fame) made SimRefinery way back when.

https://en.m.wikipedia.org/wiki/SimRefinery

Re: MeshAnything – Converts 3D representations into efficient 3D meshes

#42

Post by CyberDildonics »

ramshanker wrote: post #25

I 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…

I feel puzzled looking at Industrial scenes in Video Games. They are like 3 order of magnitude simplified compared to a real plant.

Really? You don't know why video games don't have 80 billion points and you don't know why a tool made to simplify meshes into video game objects isn't using your 80 billion point lidar scan?

For starters, these are meshes and you're talking about points. If anyone is meshing those points and they have any sense, they are working with "toy" sized chunks too so they avoid doing nearest neighbor calculations on terabytes of data.

Re: MeshAnything – Converts 3D representations into efficient 3D meshes

#43

Post by andybak »

BoredPositron wrote: post #36
andybak wrote: post #35

Earlier quoted context omitted.

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.

I have generated two models and they consisted of nothing but triangular faces.

Re: MeshAnything – Converts 3D representations into efficient 3D meshes

#44

Post by nkrisc »

Bjartr wrote: post #41
nkrisc wrote: post #30

Earlier quoted context omitted.

> 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 plant. Because they are games, not oil refinery simulators. They are typically intending to only convey a general sense of “industrial environment” and nothing more. Do your models of oil refineries include the correct grass and other plant species growing in cracks in th…

> they are games, not oil refinery simulators. That's an excellent point. I do feel compelled to mention the exception of oil refinery simulator games. Maxis (of SimCity, The Sims fame) made SimRefinery way back when. https://en.m.wikipedia.org/wiki/SimRefinery

Yes, if a game is in fact a refinery simulator I would expect it to have an accurate representations of oil refineries. But whatever the latest Call of Duty game is? It’s going to be a grey block environment designed for gameplay that then gets covered in industrial props and textures and called a refinery.

Re: MeshAnything – Converts 3D representations into efficient 3D meshes

#45

Post by dsign »

obsoletehippo wrote: post #29

I like how the Social Impact paragraph notes reduced labor costs, yay! Not e.g., reduced need for artists, so you're all out of a job.

It's a funny euphemism, in a dark sort of way. But if there is a domain where AI is not getting humans out of a job anytime soon, I think it's this one. I've read dozens of papers about remeshing, but for all of the research, very few algorithms make it to production pipelines. And those that do, still crash and fail in spectacular ways, even after a decade or more of refining and bug-fixing.

Re: MeshAnything – Converts 3D representations into efficient 3D meshes

#46

Post by bee_rider »

obsoletehippo wrote: post #29

I like how the Social Impact paragraph notes reduced labor costs, yay! Not e.g., reduced need for artists, so you're all out of a job.

One group finds a way to automate a job, and then our whole society agrees that the people who previously did that job should be tossed out into the street. But for some reason we blame the first group rather than the second.

Re: MeshAnything – Converts 3D representations into efficient 3D meshes

#47

Post by dagmx »

The topology is decent but no artist is creating meshes like this. The name feels mismatched. I’ve seen some better topology generation papers at siggraph last year which addressed quads better, though I’d need to dig through my archive to find it.

The triangle topologies in this paper made don’t follow the logical loops that an artist would work as. Generally it’s rare an artist would work directly in triangles, versus quads. But that aside, you’d place the loops in more logical places along the surface.

The face and toilet really stand out to me as examples of meshes that look really off.

Anyway, I think this is a good attempt at a reasonable topology generation, but the tag line is a miss.

Re: MeshAnything – Converts 3D representations into efficient 3D meshes

#48

Post by debugnik »

Calling these meshes "Artist-Created Meshes" is disgusting. I know researchers in this field want the word "artist" to follow the same fate as "computer" thanks to their work, but it's too soon to say the least. Can we get AI researchers? I bet RLHF can make their writing more humble than the current ones.

Sentiments aside, that's an impressive approach.

Re: MeshAnything – Converts 3D representations into efficient 3D meshes

#49

Post by BoredPositron »

andybak wrote: post #43
BoredPositron wrote: post #36

Earlier quoted context omitted.

Your definition is right. Look at the produced models and not the images in the paper.

I have generated two models and they consisted of nothing but triangular faces.

Mate I really don't know how to help you but even on the examples in the pdf there are clearly n-gons. In 5 of my 10 test there were n-gons. There are always starfishes with 5 or more connected verts. If you want to nitpick on the wording go ahead but these meshes are shite.
Post reply on HN