I'm real tired of seeing 3d amateurs ripping models from games to make fun of the topology. you still think ngons are bad because your Intro to 3d Modeling instructor told you that's how it was in 1997, be real.
Comments
Log in with your Bluesky account to leave a comment
It also just seems like a really great way to broadcast that you're a terrible person to work with? Like who wants to work with someone who will just rip on peers' work
gonna do a Twitter Clarification and state that 3d working professionals doing this isn't the same, because they have experience in this area and know what they're talking about. also doesn't apply to people ripping apart that shit genAI 3d garbage we've been seeing.
anyway, that Pyramid Head model from the Silent Hill 2 remake that was going around a while back is using Nanite, the game is made in Unreal Engine 5. if you don't know what that is, consider that you look stupid when trying to make fun of it.
I mean, not necessarily. It's just decimated, using few polygons on the flat surfaces, and more polygons on the jagged edges for the best silhouette. It's a rigid mesh, so they can get away with it, Nanite or not.
I don't even think Nanite has support for character meshes yet.
if the head is a seperate object and part of the rig, it can be nanite. this was showcased as a feature when they originally debuted nanite for unreal 5 years ago. but yeah it's possible that it's not nanite for project specific reasons.
We made some characters from skeletal meshes plus attached static meshes for Unreal 2, forever ago. I’m not sure we gained much, maybe no more than using rigid weights on a skeletal mesh version of the same geo, but I can see constructing characters with Nanite parts in the same spirit.
it's really powerful but absolutely breaks to pieces if you don't use it in the narrow band of assets it's good for, and even then you have to be careful how you author them. but it's great for rocks, and I fucking LOVE rocks, so I love nanite by extension
Relentless wank over avoiding n-gons and triangles made my first learning experience with 3D in 2010 complete shit since I was focussed on making everything a quad, and made me hate my teacher for his hard enforcement. 3D is hard enough to learn and easy enough to screw up without realising.
I totally agree with what you're saying and what Makkon is saying. But it is good to learn with quads first. you gotta know the rules to break them. Your teacher may have come from film and not video games where quads tend to matter more for subdivision.
there are some occasions where ngons or tris are placed strategically to take advantage of subdivions to create details like dimples or creases. I think another issue is being stuck in a vacuum. Most beginners don’t get to interact with professionals or explore beyond a donut tutorial.
not to mention many tools for exporting models from engine formats are not going to retain the original face composition as it was originally authored anyway...
I've been doing 3D modelling for a while, wouldn't consider myself an expert even now. I've come to stop myself from refusing to use a cheeky 🔺️ here and there but I still don't use ngons because I remember being told that the engine may have "calculation issues," whatever that means.
What they probably meant is the possibility that whatever you're importing or rendering with calculates triangulation for the ngon weird, causing normal/silhouette issues, or trouble with normal bakes, but that's easily fixed by manually triangulating the ngon whenever one causes trouble.
Though this thread does remind me of a conversation me and a friend once had where we wondered if we are only being taught one way to 3D model and if there are more wild techniques for get similar or better results.
everything is virtually tris under the hood, and that's how they import into engines. that information might have meant something in the early 2000s, but not anymore. but anyway, ngons are used in high-poly hard surface modeling pretty much exclusively, look at @oddenough.art 's work for examples
Also regarding "wondered if we are only being taught one way to 3D model", you absolutely were. There is never only one way to model. There are countless ways to model and the workflows to get the same result are all over the map.
Just had a quick thumb through your post history, saw you used ngons on making a balloon and I am wondering, I am primarily a 3D character/creature artist by the way, so do you only use ngons on smooth or organic shapes that you know are not gonna be rigged to deform?
The same people also have no idea what UDIMs are and only have experience making furry avatars for vr chat. Nothing wrong with that but it doesn't make you an industry vet.
UDIMs aren't commonly used in games (if at all?). specifically for people interested in video game art, I can understand someone not knowing what those are. the only reason I know is because I had to convert them to material IDs for realtime while I was working at ILMxLab, it was very messy
My instructors told me this in 2014 too, my skills tremendously improved when I allowed myself to do ngons. Then again, my instructors didn't want to teach too much because they say we will become competition when we get the degree. Great mindset, fking hate this school
people who dont do work in a medium deciding to become an expert when its hip to make fun of it has always been a thing, and i hate it; happens a lot with game development and indie animation, the 'armchair dev' sorts of gamers
I feel the same about a lot of people doing ‘better’ versions of the pixel art from 8-bit and 16-bit games, when then tools we had back then were so primitive compared to modern apps. No layers, often creating art on the actual target machine with the same limited palette, using a homemade editor.
I don't mind this because it's almost always coming from a place of admiration and love of the original work. I've done this, and yes I've seen people misinterpret my intentions and insult me over it too
I made my own low-poly models earlier this year and oopsed the UVs backwards on the bottom beak, lol. I look at games like Majora's Mask and Banjo Tooie and I weep at the current state of the industry. Incredible talent was once easier to come by, easier to hold onto. But the vice of greed squeezed.
I think its silly to say that the 'talent' in the industry is gone, rose tinted glasses and all that - incredible talent still exists nowadays its just people are overworked and given time constraints, time constraints on low poly models back then are the same given to 100k+ poly models in games now
Yeah... A lot of the talent today is ground away by impossible schedules and diminishing returns instead of being allowed to flourish and plant seeds of their own.
I was kind of the opposite way starting out. While I was learning about 3D modeling, I looked at ripped game assets as study material, because I was in awe of how few polygons you needed for certain things.
I have definitely run into the kind of person you're talking about, though.
Unfortunately ripping isn't great either, generally it's bad practice for learning. Engines do a lot, triangulation, removal of split per vertex etc when importing.
So you're actually witnessing the mesh after it's gone through the meat production pipeline and ended on your plate.
At the fidelity I like to model at, that doesn't matter. Most of the ripped models I looked at were from PS1, N64, Dreamcast, Nintendo DS, GameCube, PS2... Low-fidelity stuff.
Most of the time, I'm looking at the wireframe and the textures. Both of which tend to be intact.
Pyramid Head from the Silent Hill 2 remake comes to mind specifically, 3d amateurs were so sure that they had something, but the game was built in Unreal 5 and has specific features for this kind of thing
Honestly, this looks pretty rad to me. Would it perhaps even add to the crude vibe of the head?
I'm a baby in Blender and I'm no 3D artist. I know that a complex topology can lead to editing / texturing errors sometimes. But if it works, then isn't that fine? Or can it lead to cumulative errors? 🤔
this kind of model is definitely not beginner stuff. but regardless of it looking messed up, it's actually perfectly clean and has no errors. it's just poly dense
Its also probably worth noting why this is done this way over other techniques. This was done to preserve and accentuate the silhouette at glancing angles. Actually super cool and I wish more would do it.
I'd rather a few thousand extra tris over additional samples and higher res textures anyway.
Comments
I don't even think Nanite has support for character meshes yet.
Like, I can't think of a SINGLE other way to get that jagged rusty metal silhouette.
What DOES upset me is the absurdly high draw distance. THE FOG HAS A PURPOSE, WHY IS THE DRAW DISTANCE SO HIGH-
That and the right plant for foliage on the good rock... it never ends!
Also regarding "wondered if we are only being taught one way to 3D model", you absolutely were. There is never only one way to model. There are countless ways to model and the workflows to get the same result are all over the map.
That's honestly the beauty of 3D.
And tris are bad.
And boolean is bad.
And texel density matters.
Could be learning something but nope you gotta be enforcing rules you heard second hand and don’t even fully understand.
I have definitely run into the kind of person you're talking about, though.
So you're actually witnessing the mesh after it's gone through the meat production pipeline and ended on your plate.
Most of the time, I'm looking at the wireframe and the textures. Both of which tend to be intact.
I'm a baby in Blender and I'm no 3D artist. I know that a complex topology can lead to editing / texturing errors sometimes. But if it works, then isn't that fine? Or can it lead to cumulative errors? 🤔
I'd rather a few thousand extra tris over additional samples and higher res textures anyway.