Is it viable at all to model entirely with triangles in Blender?I gave it a go and it's pretty easy even though you have to create each triangle manually, but Catmull-Clark subdivision turns it into quads which causes some weird distortion at vertices that are part of numerous edges.The reason I'm trying this instead of box modelling is because I hate how the only easy way to add resolution with box modelling is to create edge loops, which affects the model as a whole rather than just a singular part.
This is not how you 3d. If you don't like dealing with topology try sculpting
>>560132I like dealing with topology, it's useful to easily get exactly the form you want.What I don't like is dealing with mesh resolution, which box modelling really overcomplicates, but triangle modelling simplifies.
you literally cant model with triangelsunless you like to extrude verticies instead of edges its 2 steps against one
>>560134>you literally can't model with triangles>OP pic is modelled with triangles
>>560135ok, what now?
>>560129You could, but it's more work, and you can't use subdiv (as you observed already).Look up some topology/subd modelling guides on how to add details. Also, you could rough out with triangles (still sounds tedious) then retopo if you like. But the whole point of subd/all quads modeling is to aim for a good structure that's easily extensible/ detailable. Whole triangles goes quite against that. Try to think about how a given surface would flow.
>>560136What now? Nothing. You're wrong. The end.
>>560143yeah but his result is awful. so the fact that you can theoretically model with triangles isn't some fascinating discovery. its just a shallow technical fact that adds nothing to an already dead thread in a dead general that is littered with lunaticsthis place is fucking horrible, someone please delete this board
>>560133well congrats you made something that looks like dog shit
>>560145>>560153Diss all you like, but box modelling would have a much harder time adding detail at this low a mesh resolution.
>>560155Good luck animating that family.
>>560155Are you trolling or do you legitimately think that has any detail, will be easy to add detail to, looks good in any way, or will deform correctly?
>Is it viable at all to model entirely with triangles in Blender?no.because>you have to create each triangle manuallystick to quads if you want to be "viable". Use tris if you want to be an autist and place every poly.the whole point of having edge loops is so you can easily add resolution and not have to place every poly individually.
>>560197I may not be able to use edge loops, but unlike when using quads, I can subdivide.Subdividing a region of quads will turn all neighbouring quads into pentagons, which cannot possibly become quads, but subdividing a region of triangles turns all neighbouring triangles into quads, which can become triangles in three different ways.Sure, it's a little less clean than using edge loops, but as I said, edge loops affect too large an area.Subdividing means I can add resolution only to the chosen area.
>>560155At this point, you're supposed to load the low-poly model to a sculpting program and make the details there. Even Sculptris is enough.
>>560129It's viable for still image rendering, but animating an all triangle mesh will be really difficult.
>>560129Lightwave uses verts and triangles to model Too bad it seems to have be stopped development for. I learned to model/animate in LW but I found Blender to be much better, but modeling without triangles is weird as shit. Well... it was .
>>560219>initially used triangle modelling out of something other than naivetyWell, you're a unicorn.Think it might be worth giving Lightwave a go, to see if I can pick up any tricks to bring to Blender?
>>560212>Subdividing a region of quads will turn all neighbouring quads into pentagons, which cannot possibly become quads...what fucked up subd algorithm are you using? Catmull-Clark will always give quads, even when using tris even though it spits out garbage.
>>560228A region of quads, not the whole model.Imagine a mesh that is two connected squares.One square is subdivided.The other square now has 5 vertices, because the 2 vertices it shared with the other square became 3.
>>560230not gonna make it
>>560262I'm not saying that picture is detailed, I'm saying it's easy to add detail to.If I add resolution to a region with box modelling, I have to worry about how I affect everything else that the edge loops travel through.Edge loops often traverse the whole body, something difficult to avoid.If I add resolution to a region with triangle modelling, I only have to worry about how I affect the border of that region, because the border will turn into quads that then have to be split into triangles.
A quad is really just two triangles.
>>560271you should be sculpting anyway you're wasting your time. You're doing something that nobody in the industry would do. You would get laughed out of town
Holy christ you fucking morons could at least do a tiny bit of research before you talk about shit you don't understand.https://www.youtube.com/watch?v=ckOTl2GcS-E
>>560277Oh yeah, nobody here understands the most basic premise of what sub dividing a mesh is. This tutorial is worthless. What does it explain that contradicts any replies in this thread?
>>560279The rager didn't reply to any specific posts, but I assume they were talking about the people who didn't seem to understand that quad modelling doesn't allow subdivision while triangle modelling does.
>>560277Would all the polygons that were used to make the boy's legs be considered redundant polygons?
>>560313See if they disappear when you turn on cull faces.
What am I watching and why is a senior 3D artist from Ubisoft doing this?https://www.youtube.com/watch?v=Y83FLL6TqF0&t=642s
>>560396>grenade tutorialbefore even watching that I can tell you hard surface topology doesn't matter as much as much as character topology.
>>560397Interesting, I don't know much about it yet, I've been learning this for a couple of weeks now. His workflow is pretty chaotic, though.
>>560397While technically true you are practically wrong. "Hard surface" modelling refers to constructing a high poly model using a sub-d workflow to construct mesh with optimal normals for baking, topology here isn't less important, just different, compared to character modelling where the topology is constructed to stretch optimally for animation. The reason the tutorial artist is connecting triangles is because he's making the low poly/bake target mesh, not because he doesn't care about his topo.
>>560355I admit that's one way you can add mesh resolution to only a single area while still strictly doing box modelling, but making your topology more complicated makes edge loops harder to use, and it's still more difficult to add topology like that than it is to do with triangles.
>>560403> "Hard surface" modelling refers to constructing a high poly model using a sub-d workflow to construct mesh with optimal normals for bakingThat is 100% not what hard surface modelling it.hard surface modelling means exactly that- hard surfaces- as opposed to organic forms. There are plenty of hard surface workflows; they don't define the definition.