🧵 Untitled Thread
Anonymous at Tue, 19 Sep 2023 12:02:06 UTC No. 958812
my work-flow? non-destructive!
Anonymous at Tue, 19 Sep 2023 13:03:06 UTC No. 958819
>30 fps on 12 teraflops
Anonymous at Tue, 19 Sep 2023 13:08:51 UTC No. 958821
laughing at boomers who spent years learning topology when we can't just brute force anything with powerful machines
Anonymous at Tue, 19 Sep 2023 13:13:01 UTC No. 958823
>>958820
>>958821
It makes me feel better now I can simply stop worrying and make my fucking waifu waifu pantsu summer days volleyball sexy waifu trash, and will be much better than AAA goyslop.
Anonymous at Wed, 20 Sep 2023 07:30:58 UTC No. 958938
>>958812
> if you're worried about being good at your job here's an example from someone bad at theirs
Anonymous at Wed, 20 Sep 2023 11:46:45 UTC No. 958957
>>958823
>my fucking waifu waifu pantsu summer days volleyball sexy waifu trash
What did he mean by this?
Anonymous at Wed, 20 Sep 2023 13:22:40 UTC No. 958966
>>958820
Those days have been over for a long time anon. The rule is more 'use as many polygons as you need' than trying to conserve every where.
But also realize that the more you conserve somewhere the more you can waste somewhere more important so being an absolute retard about it isn't good practice either.
My recommendation to artists targeting AAA for the last decade has been to be poly conscious but not poly concerned.
You going out of your way to optimize a surface shaving of the last few polygons is a waste of your time, the number of polygons on a individual model is not what will
tank performance on PC or game console hardware, it's batching, drawcalls, overdraw, occlusion culling, shader complexity and stuff like that which will tank performance now days. You're not gonna brush up against mesh geometry limits unless you go absolutely ham with your polycount.
Anonymous at Thu, 21 Sep 2023 02:22:50 UTC No. 959034
itt anons forget about gpu topology tesselation and in scene game rips
Anonymous at Thu, 21 Sep 2023 06:06:28 UTC No. 959047
>>959033
That's strange. Why isn't the triangulated game model not clean? Lmao
I know that game engines like to fuck around when triangulating, but I've never seen it this bad.
Anonymous at Thu, 21 Sep 2023 09:09:02 UTC No. 959063
>>958812
I mean it looks pretty ugly but if it doesn't need to deform in anyway I surpose it doesn't really matter
Anonymous at Thu, 21 Sep 2023 09:18:19 UTC No. 959065
They are probably using some tech like nanite already, why is everyone crying about it
Shipped an asset smaller than a human with 700k poly this week for a game, such is life
Anonymous at Thu, 21 Sep 2023 13:10:45 UTC No. 959095
>>958863
that's still abysmal topology
Anonymous at Thu, 21 Sep 2023 13:36:29 UTC No. 959098
>>958821
>oh boy guess I have to buy another 4090 again to run this year's Ding-Dong Bing Bing Wahoo Adventure
Anonymous at Sat, 23 Sep 2023 13:07:14 UTC No. 959293
>>959095
No it isn't. Mesh isn't going to be deformed so it doesn't need to be all quads. Nice job telling us you started using blender last month retard.
Anonymous at Sun, 24 Sep 2023 16:42:58 UTC No. 959435
>>959095
Nah, it's pretty much perfect. Even the lettuce is smartly modeled instead of being a transparent texture.
Anonymous at Sun, 24 Sep 2023 19:49:12 UTC No. 959444
While this isn't a good model I'm tired of model fags obsess about muh topology. It all gets broken down to triangles anyways. GLTF literally doesn't support anything but triangles. If your model has something other than triangles it requires an extra preprocessing step where you must split and interpolate all vertex attributes. Adding extra triangles because muh topology is bad.
Anonymous at Sun, 24 Sep 2023 20:34:14 UTC No. 959457
>>959047
>game engine
>fuck around when triangulating
You realize this is the same engine that Oblivion uses, just with some extra patches added to enable extra shit right? There's no way this shitty engine is performing any kind of optimization, that's the raw model.
Anonymous at Mon, 25 Sep 2023 16:55:12 UTC No. 959551
>>959444
>It all gets broken down to triangles anyways. GLTF literally doesn't support anything but triangles. If your model has something other than triangles it requires an extra preprocessing step where you must split and interpolate all vertex attributes.
It's not that the concept of triangles is bad, it's that too many of them wastes resources in rendering, and historically there wasn't enough spare processing power for a game to handle that. In addition, if you are animating a mesh, the deformation will be determined by the topology, and it will look like shit if it isn't optimized for the expected motions - for example, there are special techniques used to model things like elbows and knees in character models so that they look correct when bent.
As mentioned by a few posters above, static models don't need to be concerned with topo for deform, and modern computers are often powerful enough to render high poly counts without too much concern - although having multiple models optimized this poorly will stack the effect quickly.
Anonymous at Mon, 25 Sep 2023 22:21:11 UTC No. 959571
>>959033
wtf its just a cylinder
Anonymous at Wed, 27 Sep 2023 19:53:03 UTC No. 959789
>>959457
You don't understand how realtime rendering works, they definitely use LODs and algorithms to automatically generate them, no engine is drawing 100,000 triangles for a sandwich on the other end of the map
Anonymous at Sun, 1 Oct 2023 18:48:48 UTC No. 960126
>>959293
>>959435
NTA but I thought mesh density should be uniform? The top tries are so much larger.
Anonymous at Sun, 1 Oct 2023 22:02:22 UTC No. 960137
>>960126
>NTA but I thought mesh density should be uniform?
No.
Anonymous at Thu, 5 Oct 2023 01:31:56 UTC No. 960388
>>960126
Why would mesh density need to be uniform?
Anonymous at Thu, 5 Oct 2023 18:04:00 UTC No. 960483
>>959033
even i could do better
fuck
Anonymous at Fri, 6 Oct 2023 05:24:58 UTC No. 960548
>>960547
topology is important for animation rigging and bending, for solid and static things not so much though you never want to waste processing power on needless polygons like the pic on OP
Anonymous at Fri, 6 Oct 2023 05:36:37 UTC No. 960551
>>960550
you might try averaging the faces, but if that doesn't work it means you'll need to add more topology
Anonymous at Fri, 6 Oct 2023 06:14:12 UTC No. 960552
>>960551
But wouldn't it make a lot more unnecessary triangles/quads?
Anonymous at Fri, 6 Oct 2023 06:50:37 UTC No. 960559
>>958823
I hate how you zoomers call everything a waifu. Stop using our words
Anonymous at Sat, 14 Oct 2023 08:08:15 UTC No. 961217
>>960126
Only when it's a high poly mesh where the topology is mostly quads. If it's a triangulated video game prop that doesn't bend, twist or get its surface displaced, it can have very ugly topology as long as it's the correct shape.
Anonymous at Mon, 16 Oct 2023 07:20:07 UTC No. 961334
what many fools here don't understand, the time of a 3d designer is more expensive than time on a renderfarm
Anonymous at Tue, 17 Oct 2023 00:30:48 UTC No. 961440
>>960547
Worry about topology a lot when you intend to animate something, and the more you need to animate the more you need to worry. You also need to worry a bit for deforming stuff.
For everything else, you just need as many polygons as is required to define something's shape. More exactly, to define an object's silhouette. If that is too many polygons for the application, make compromises to the shape or change the design.
For your example, the problem is the shading and maybe how the sub-d is applied. There's ways to work around that and get what you want, I would probably start with a weighted normal modifier and start messing around with there. There's probably a way to get that crease out with some kind of bevel setting too.
IDK exactly because I don't do sub-d modelling and that kind of shading issue rarely comes up for me. But there'll be a way around it.
Anonymous at Tue, 17 Oct 2023 10:36:26 UTC No. 961474
>>958812
correct me if i'm wrong but this model is only used when you drop it on the floor or look at it in your inventory, so you wouldn't see it in regular play anyway
Anonymous at Tue, 17 Oct 2023 16:07:27 UTC No. 961504
>>960552
They aren't unnecessary if they are holding or smoothing the shape
Anonymous at Thu, 19 Oct 2023 08:43:46 UTC No. 961656
>>959033
How do you even reach this state? It doesn't seem sculpted and even if you crank the subdiv to the maximum, it will not fuck with geometry this much
Anonymous at Fri, 20 Oct 2023 11:57:08 UTC No. 961731
>>959033
>Feminist cup is horrible underneath
Anonymous at Fri, 20 Oct 2023 12:04:54 UTC No. 961732
>I like to collect all the sandwiches
how fast is her computer bros
Anonymous at Sat, 21 Oct 2023 01:50:57 UTC No. 961767
>>961656
Most likely a scanned mug they didn't bother cleaning up.
Anonymous at Sun, 22 Oct 2023 14:17:01 UTC No. 961895
I've inspected a number of bethesda 3d models before. They hold a better average level of quality than 95% of gamedevs these days.
A larger portion of their budget goes into art creation than almost any studio.
Anonymous at Sat, 28 Oct 2023 11:27:25 UTC No. 962413
>>961895
same, i've looked at a lot of their models over the years in fallout games and they're really good. usually in games, only hero assets are really clean and nice looking and the rest of it is outsourced to india or whatever, but in bethesda games, everything seems to have consistently good quality.
Anonymous at Wed, 1 Nov 2023 05:58:20 UTC No. 962796
>>958812
>all those fucking edges coming from what looks like a single vert.
What in the-
>>959033
Are they even trying?
I'm a novice who's not buggered about with 3d programs for about 4-5 years being busy with other shit and even in my rusty balls state I could do better than this shit.
Anonymous at Mon, 6 Nov 2023 07:46:46 UTC No. 963366
>>958812
fake btw
Anonymous at Mon, 6 Nov 2023 08:19:28 UTC No. 963369
Retard here: what’s wrong with this sandwich? It looks like a sandwich to me.
Anonymous at Mon, 6 Nov 2023 10:42:54 UTC No. 963382
>>962796
>Are they even trying?
The 3d credits for Starfield is like 80% indian with a few chinese supervisors.
Unless Bethesda explicitly rejects shit models and tells them exactly what to do this is about what you'd expect.
Anonymous at Wed, 8 Nov 2023 19:46:42 UTC No. 963639
>>958812
The real issue here isn't the topology, it's the triangle count
Anonymous at Wed, 8 Nov 2023 22:16:52 UTC No. 963661
>>962796
>>963382
>>963639
It's fake, guys. Stop believing everything you see on the internet.
Anonymous at Wed, 8 Nov 2023 23:40:43 UTC No. 963671
>>963661
good marning saar
Anonymous at Fri, 1 Dec 2023 10:56:21 UTC No. 965787
Anonymous at Fri, 1 Dec 2023 13:59:15 UTC No. 965797
>>959444
That. Quadrangle have one angle more than triangle. That's 1/4 more angles. Everything you can do with triangle you can do with quadrangles, and you can do better with quadrangles.
Anonymous at Fri, 1 Dec 2023 18:21:27 UTC No. 965817
>>965797
I always found the industry obsession with UV wrappings when infrared wrappings were never tried absolutely retarded.
Anonymous at Fri, 1 Dec 2023 21:58:10 UTC No. 965831
>>959034
whoa there you're gonna educate them