why TRIANGLES rather quad nowadays?

[Anonymous No.387135]

Easy to have issue with quad, not with triangles, video games use too triangles, isn't it, so why so much people say triangles is EVIL, because Zbrush who use only quad divisions ?

[Anonymous No.387136]

http://gamedev.stackexchange.com/questions/9511/why-do-game-engines-convert-models-to-triangles-compared-to-keeping-it-as-four-s

>Four points or more may not be on the same plane, but three points always are (ignoring degenerate cases). This has the interesting property that scalar values vary linearly over the surface of the triangle. Even when the triangle is projected on the screen, scalar values still vary linearly respect to x'/z and y'/z.
>This, in turn, means most if not all of what's needed to shade, texture map, and depth filter a triangle can be calculated using linear interpolation which can be done extremely fast in specialized hardware.

[Anonymous No.387138]

Quads subdivide better than tris so you want quads only when you are making your base mesh that you will sub-d in order to sculpt. Some novices get confused when they hear quads only and think it applies to in game meshes as well however this is not the case. Here is a picture of Nathan from Uncharted 2, notice the tris, tris are used all the time for in-game meshes.

[Anonymous No.387140]

thanks, but it's why I prefer Sculptris "modeling" rather Zbrush, even it Zbrush is thousand more developed and have more tools and undo, you can do whatever you want with traingles tesselation of Scruptris

[Anonymous No.387142]

>>387140
Can't make a hole in a mesh with sculptris, hence why 3d-coat is better because you CAN do whatever you want. With that said I still prefer z-brush for most thinkgs.

[Anonymous No.387143]

>>387142
I do some Zspheres, for the form, then send that on sculptris to finalise the sculpting, then again zbrush for zremesher, at least if it's not for animation, but manual retopo is so boring and a long process ! Zbrush do holes too now so it's ok

[Anonymous No.387158]

>>387142
>hence
>why
Pick one. They mean the same thing.

[Anonymous No.387164]

>>387142
You can do that in ZBrush too. And you can add edgeloops, bevels and extract surfaces as well. Plus the retopo is much better in ZBrush than 3DCoat now. Don't support those religious crusader assholes who make 3DCoat.

[Anonymous No.387195]

>>387138
Basically, this.
But I would push the tris further back.

[Anonymous No.387208]

>>387135
Easy to divide, easy to control and they just look better, i have more fun with modeling quads than tris.

[Anonymous No.387212]

It's not nowadays. It's ALWAYS been triangles in games, because a GPU only processes triangles. Quads and Ngons are a trick that 3D modeling software does, they just don't render the edge going across the quad that makes up 2 triangles, but it's still there. Keeping things as quads is mainly to allow for all your modeling tools to work without issue, since they can easily follow edgeloops for all sorts of operations. When it's triangles, it's just a mess of edges that your software can't do much with. Also, when your mesh is in quads, then you can be sure that it's going to subdivide, smooth out and deform without much issue (unless you have some seriously fuck topology), whereas if you have some triangles, it will often show some smoothing issues, especially when it comes to organic surfaces.

Most games models are modeled in quads and then converted to triangles before exporting to the game and maybe optimizing it a bit more since you're using triangles finally and can merge some points. If it's extremely low poly stuff, then it's often mostly triangles from the start.

[Anonymous No.387213]

how the fuck do you use zsphere rigging? i keep getting the most fucked up skinning no matter the configuration of the spheres. it just gets segments of my mesh and tears them off leaving the rest invisible.

[Anonymous No.387214]

Why does the OP's world revolve around video games? OP, there MORE to the cg world than games, shithead.

Quads are useful in every other type of polygonal/sub-d modeling methodology. It's standard practice for a reason, but then there's also an additional optimization pass for gaming done IF YOUR TARGET MARKET IS GAMING.

[Anonymous No.387215]

Are you dense OP? Quads don't exist, they are a convenient abstraction to model easier, like have edgeloops and proper deformation, but the hardware only works with tris.

[Anonymous No.387217]

Tris are just quads slit in half diagonally

Therefore quads > tris.

[Anonymous No.387322]

Here's a handy little picture to illustrate the problem: to properly draw a quad, you need to check if all the points are on the same plane. If they aren't, good luck depth-testing and shading that.

Subdividing the quads into triangles solves the problem, simplifies the graphics pipeline and allows us to make performance improving assumptions. Your special cases for handling non-planar quads would probably involve doing that anyway.

[Anonymous No.387325]

>>387215

This. Tris are not forbidden nor anything... you just have to understand where you can use them and you can use them if they don´t affect the flow or if they are needed for it.

A good flow is easier to read and to predict and is essential in production since you won´t be doing everything.
The other guy should see the mesh and know the shape a flatten UV map should have or how a mesh is going to deform in animation.

So... as longs as triangles are in places where the result is still easy to read and don´t interfere with stuff like this they can be used.

[Anonymous No.388402]

>>387212
>It's not nowadays. It's ALWAYS been triangles in games, because a GPU only processes triangles.
It's a bit less simple than that. While the hardware processes triangles, it is able to receive slightly higher level primitives, notably triangle strips, which are a way of organizing triangles that requires only sending one vertex per new triangle after the first until you encounter a triangle that can't be encoded into the strip. Guess what easily converts into triangle strips? That's right, quads! You can get up to nearly 60% compression on the vertex stage if you have good flow, plus it'll tesselate better.

[Anonymous No.388411]

>>387212
>It's ALWAYS been triangles in games, because a GPU only processes triangles


Sega Saturn and its bat-shit crazy architecture would like to have a word with you. It actually used quads rather than tris and forced them to be planar.

[Anonymous No.388427]

>>387135
Triangles for real time, quads for pre render. Simple as that.

In real time you can't waste resources determining which direction your polys should bend, but in rendered images you let it decide so you have smoother shading.

The higher the poly count the less it matters. Neither are evil, and anyone who says so it an idiot.

[Anonymous No.388428]

Modelers using quads is almost like cutting off the end of the roast...

Except that there are contexts where you actually should use quads.

Technically, anything that needs the surface to have a good "parameterization" is going to require quads Unfortunately that is just about everything these days.

You see, anything that is interpolated across a vertex that has more than 4 edges on it is going to have a discontinuity. That applies to subdivision and shading, which means that any point like that will have a subtle or not-so-subtle artifact.

The reason is simple, a surface has two dimensions (labeled U and V) and there is no way to tell what direction U and V are going at a vertex unless there are exactly two crossing lines at that point (like the graphs you'd make back in high school math).

It is better that the lines cross as perpendicularly as possible as well.

[Anonymous No.388429]

>>388427
Higher polygon count does not fix anything. It just makes the artifacts smaller and if you got those small polygons by doing subdivision, that doesn't help either.

[Anonymous No.388579]

>>387135
Quads ARE triangles (two triangles, to be precise), and you model with quads so that the mesh doesn't take a diarrhea shit on you when you give it a skeleton and animate it. Another way of putting it is to say that modeling in quads is a way of laying out your models triangles that won't fuck you over in the long run

[Anonymous No.388614]

>>387212

>It's ALWAYS been triangles in games, because a GPU only processes triangles

false

Any OpenGL 4.x or DirectX 11.x hardware supports quads.

[Anonymous No.388671]

>>388614
Any OpenGL 4.x or DirectX 11.x hardware supports quads.
Thats just triangles without the connecting edge rendered ,.,

[Anonymous No.388691]

>>388614
There has been quad support for far longer than that.

>>388671
True, however there are times that REAL quads are used such as in particle effects or in some forms of foliage like grass.

[Anonymous No.391623]

>>388671
DirectX doesn't support quads.

http://msdn.microsoft.com/en-us/library/windows/desktop/ff476189(v=vs.85).aspx

[Anonymous No.391655]

>>387135
Y'all niggers posting in a troll thread.

[Az No.391656]

>>388427

People say "triangles are evil" because they are hard to work with and sub-divide poorly. Quad is basically just two tri's with an invisible edge anyway, they just don't have to deal with them directly.

[Az No.391659]

>>388428
what you talkin' bout Willis?

[Anonymous No.391665]

Basically, quads are easier to work with when you care about topology. But quads are purely a working tool; They all get turned into triangles when they're done actually building the model.

[Anonymous No.391670]

Quads are only nicer to subdivide in the Catmull-Clark algorithm. Loop subdivision works perfectly for triangles. There are also alternatives that look good for both triangles and quads.
For example:
http://www.cise.ufl.edu/research/SurfLab/papers/04qtsub.pdf