[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / r / s / t / u / v / vg / vr / w / wg] [i / ic] [r9k / s4s / vip / qa] [cm / hm / lgbt / y] [3 / aco / adv / an / asp / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / qst / sci / soc / sp / tg / toy / trv / tv / vp / wsg / wsr / x] [Settings] [Home]
Board
Settings Home
/3/ - 3DCG



Thread archived.
You cannot reply anymore.



File: torbool.png (837 KB, 1904x688)
837 KB
837 KB PNG
Ok, so we all know booleans are a thing that exist and are real good for speed modeling and everything but, uh

what's the cleanup process for them? To get them to game ready etc? Any time someone's showing off Booleans they never show it.
>>
you clean them up by retopologizing like you would any other retopo task
>>
>>597532

Boolean to your hearts content, only the unemployed cry about them.

Just retopo them like everything else when the time comes to clean up the mesh
>>
As the others already said, you retopo it afterwards.

THEORETICALLY could use the resulting messy mesh in most cases but
>they're so messy they might fuck with your game engine
>texturing it will be a massive pain in the ass because of the mess
>if there are parts that are supposed to deform it will deform disgustingly
>light reflexes will massively fucked up if the mesh is not clean
>subdivision will usually not work or result will look vomit inducing

If you're lazy you could try several auto-retopo ways but usually the results lack details in certain parts and usually still need some work.
>>
>>597532
what other anons said. you retopo the mesh afterwards.
Imho the best thing about booleans is that it is so quick and stacking them on each other gives you awesome results, wich you can see before you commit to it. using booleans smartly, gives very good results. because of all this, Booleans is really powerful tool for designing/concepting hard surface, and not only hard surface..
for example, Zbrush became way more verstile tool for hard surface since r8 update.

but you are too lazy to retopo, then it has no meaning
>>
>>597542
Agreed. Booleans in combination with dynamesh in zbrush are a fucking godsend. It's one of my favourite things to use for making mouthbags and eyeholes. Outside of Zbrush I don't really use booleans, I think it's kind of stupid to rely on them because they cost so much time due to retopo. You can easily do without if you keep your models low poly for getting the basic shape, and then adding details. That said, if you must use bools, better do it in the low poly phase. Then, cleanup won't be as big of a deal. Otherwise look for rounded shader options or something..
>>
>>597542
>>597535
>>597541
Only thing I'm sort of wary about with regards to retopo is a lot of the retopology tools I've used so far seem really poorly suited to hard surface? Since they mostly consist of fairly rough and inaccurate quad drawing a lot of them seem to be made with more natural/less precise stuff in mind.
I get that I'll be retopologizing but it seems pretty clear that it's going to be a different process to what I'm used to and would typically use for character sculpts, and I'm not really sure what's out there since I rarely see anyone talking about it. With hard surface at the moment, I find it pretty easy to extract low-poly from the early stages of the pretty typical box modeling/subdiv process I use but there are a lot of situations where booleans would be insanely useful and the topology they produce is the one worrying aspect keeping me from fully utilizing them.
>>
>>597532
Depends on how you intend to use the resulting mesh, for games it's actually less work because you don't have to worry as much about topology as you're not subdividing anything. If the intersection is with a planar surface, anything goes as ngons don't matter, but if you have two intersecting curved surfaces, like some of the grooves on the gun barrel, it might be a bit harder. You could use a trim sheet with normals that you could lay over the edges, chances are that if you're doing hard surface modeling for a game, you're already using trims.
>>
>>597532
Don't know if it's been mention ITT yet, but I would highly recommend to all anons to check out using Fusion360 as a tool to create your hipoly hardsurface assets.
It gives a perfect, consistent, machine-like look to all your bevels and fillets. It can output in a bunch of formats that you can take into whatever app to perform retop, as well as using as your bake-source, obviously.

Seriously - if I ever do a hardsurface object (like a handgun) that would IRL be manufactured by CNC milling or whatever, I'd start in Fusion360 every time now.
>>
File: 333.jpg (73 KB, 1609x759)
73 KB
73 KB JPG
>>597544
I mean Booleans are really big for me, in design process, in Zbrush.
smartly stacking them on each other can generate very good shapes, which with just modelling will be good and will take too much time to get to shape you want, and of course there is very big chance that design won't be good, with Live booleans you manipulate separate subtools or polygroops and change the outcome very fast.

BTW, outside of Zbrush, I don't even model )) I know this is not popular workflow, especially on 3 and with oldschool people, but I model everything in Zbrush.


I modeled/designed this, almost 100% with live booleans. (Pic related)
>>
>>597565
>BTW, outside of Zbrush, I don't even model ))
I've been using ZBrush more and more lately and I've been heading more and more in this direction - just starting out in ZBrush for more or less everything.
When I started reading 3D magazines interviews with artists and I'd see one after the other just say 'zbrush, zbrush, zbrush' over and over again for *every* kind of work - hardsurface, environment - everything, I just didn't get it; why the fuck would you do hardsurface machine-made stuff in Zbrush, but the more I use it, the more it makes sense and the new LiveBoolean just made it even easier.

I see now that a 100% Zbrush workflow is possible and the people who do it swear by it... but you gotta be goooooooooood before you have that level of autonomy in this one application.

3DS Max is boxing.
Zbrush is Jeet Kun Do.
>>
>>597573
I had very little modeling experience when I started Zbrush, and I was awed by it.
it's exactly a two years since I started 3d, but almost whole first year I was doying Vaporwave shit in cinema 4d. with 0 modeling.

then switched to Maya and Zbrush. I also was using maya for box modelling, or to correct some things, but I stopped. I really don't see any reason not to use Zbrush, since I learned Zmodeler. it needs good retopo and UV tools tho., UVs most of all.

If you reaaaallly want to learn Zbrush, and all it's tools. I'd suggest to watch
this playlists : https://www.youtube.com/watch?v=5jZwtBtvIIY&list=PLMjnnUF3eJFdn1vTMFZd-gaY-hXGzwK9H

and this:

https://www.youtube.com/watch?v=PewNbS__W50&list=PLMjnnUF3eJFe37oKxPWLGR0A3j4moKpNj

they are both awesome Instructors and zbrush Developers.
and also good artists. they both have refined flow of explaining tools, you are 100% learning, if interested.
>>
>>597587
Thanks, I may take a look at those. I have Pluralsight and Gnomon right now though.

I think maybe going straight to Zbrush instead of Maya/Max is an interesting way of learning, because there's a certain amount of 'unlearning' that a box-modeller has to do before he can really get his head round zbrush. That;s what it was like for me anyhow, it was like learning to read backwards or something.

Zmodeler could do with some improving and yeah the retopo workflow is horrible, and dealing with scale in Zbrush is just a total headache - beyond that though, yeah it has the potential to be the photoshop of 3d modeling if pixologic commit to making the whole thing a bit friendlier.

Then.... if they could integrate a proper baking mechanism, a PBR painting environment and advanced PBR materials... then we'd really be talking.
>>
>>597592
yeah, I'm sure if I were already good with box modelling, I would also resisted this. I knew how to draw and that also helped.

Retopo tools is the worst thing. they must change it first, because this software depends on retopo more than any other software. some people are even using Zbrush's retopo tools, but that is really too much..


Gnomon and Gumroad, tutorials are more artistic based things, this two "shows" has very cool format, they are one hour livestream shows, every two week. one explains topics, like booleans, or fibermesh, etc, in depth and viewers choose what topic will be next show.

and second is, viewers ask questions live, and he explains these questions.

the good thing is in these two shows that, they explain the best ways to use Zbrsh, since I found these I've got hold of so many tools which I thought were not important. when I watch tutorials, I find that many artists/tutors don't know so much awesome little things.
oh and BTW whole Zbrush youtube channel is Gold,
>>
I just bake them when it comes to game models.
>>
are booleans as bad in maya and max as they are in blender?
>>
>>597765
Don't know about Maya, but Max has a decent boolean system, and Modo apparently has the most robust options. Unless you're using software that doesn't give a shit about topology, like Zbrush or some curve-based editor like MOI 3D or Fusion360, I doubt there will ever be a satisfactory poly-based system. You just have to use common sense about what kind of shapes mesh well together and clean up accordingly, most of the time the anger and confusion with booleans is more on the end of the user not knowing how to use them properly, rather than the fact that using them sucks in general.
>>
>>597765
I've got no point of reference for how bad booleans are in Blender, but in Maya they seem to be pretty barebones but functional. I have to clean up the mesh a whole lot but I've never "broken" a mesh with one.
>>
Within the scope of this thread, what do you guys consider "booleans"?

Because for me, a defining feature of a boolean-based workflow is that no (or minimal) cleanup is required. Think of Tor Frick's rounded edge modeling, or live booleans + a dynamesh pass.

When people complain booleans require cleanup, I'm assuming they're using them as part of sub-d modeling. Which at this point, is still sub-d.
>>
>>597828
I think of booleans as anything where most of the final shape is achieved by subtracting one shape from another, but how you manage to smooth the result is a bit of a secondary consideration, whether it’s a turbosmooth, an edge shader, chamfers with weighted normals, or normal trim sheet. In fact, most workflows outside of remeshing in a sculpting app seem to require some amount of fixing before use.
One type of “boolean” that many people seem to gloss over is drawing out shapes using splines and then generating a mesh from them, as when dealing with any type of curve-based system all booleans are “free” so to speak, so one should always consider whether they should start with a shape rather than a poly that you cut into later.
Anyways, to me Sub-D is usually more synonymous with edge-extrusion modeling, because one of the draws is being able to see the smoothed variant of your mesh as you model and create edges, and booleans don’t fit in with that idea as you can’t smooth the mesh until you’re done with it, making it more of an afterthought than an integral part of the workflow.
>>
>>597828
like I said already, in Zbrush it's great, because you can see booleans without commiting to it, you can sub D and smooth mesh as much as you want, you can sub d any boolean object, you can sculpt on boolean objects. and stack many boolean shapes on each other as layes .use IMM brushed as boolean subtractions.
so for me it's great designing tool, because you can see the results imediately. and can change the result by modifying differents shapes in boolean hyrarchy.
>>
What do these "cleanup" and "retopologizing" words even mean? The tutorials never mentioned anything about this. What do?
>>
>>597879
Retopo is the steps you take to make a low-poly mesh you can bake your high onto, basically making your fancy boolean prop useable within a game.

The process depends highly on the workflow. For something like rounded edge, even if you use live booleans, retopo might be as simple as taking the highpoly geo and deleting a few unneeded loops here and there (since with rounded edge, your high and low poly don't really differ).

If you're doing booleans with subd models (the zbrush workflow), retopo will require more effort, but could basically involve redoing your boolean operations with the same meshes at a lower polycount, then manually fixing the topo wherever the booleans produced something fucked, or sub-optimal.
>>
>>597879
Cleanup: Booleans (and sometimes other modeling processes) often leave behind lots of excess edges and vertices, and you have to double-check your mesh to get rid of them. Some programs have built-in tools for aiding with this, for instance in Max selecting all vertices with only two connecting edges (i.e. junk) is as easy as pressing a button. Merging and reflowing the remaining verts and edges is more of a hands-on thing. There is also a modifier called STL Check that sees if your mesh has any "illegal" topology.

Retopo: Taking a wastefully-made, but detailed piece of geometry and making a more primitive, but widely-usable variant. This applies more often than not to sculpting, as the process leaves you with a millions-of-polys model that can't really be used anywhere directly, so you have to trace over it with a more basic low-poly variant, and then bake the high-poly down, either as a normal map (for games) or displacement map (for CG). This low-poly is the one that's rigged/animated and upon render it gains the "look" of the high-poly through the baked information.

Oftentimes being wasteful in the design process is generally faster than making sure your topology is perfect as you go along, as you'll be wasting time on a design you may not even be sure you like yet, so having to essentially remake the model a second time isn't an issue, especially since you're basically just tracing something you already have right there in 3D space.
>>
>>597535
>>597540
>>597541
>>597542
I think the question is "how to retopo hard surface shit". I doin't get it either.
Quad Draw is excellent for soft bodies, Instant Meshes can be decent if you're in a hurry, but neither take good care of perfect edges and curves. How do you deal with basic geometry and hard edges when you retopo?
>>
>>597560
fusion is aight indeed, but i would recommend it only to design if you actually able to do that
using fusion360 for modeling is not a good idea because it shits out garbage topos. if you know what you are doing it's better to stick to reguar booleans or at least use mesh fusion (modo). otherwise you ll just waste your time with fusion 360 when you could boolean it right away in regular 3d modelling app with better results and less topology pain.
>>
>>599737
i started with max, but never retopo'd there. my first retopo experience was modo and it has really good repoto tool with snaps to verts and angle snaps so retopoing hard surface out of fusion 360 was quite ez
>>
>>599751
Hm, nice. Might try that then.
>>
>>599737
Personally I wouldnt retopo I would just clean up the existing mesh by welding verts and cutting edges
>>
>>599750
but, if you are doing a retopo anyway, why wont you just take advantage of the fusion 360 precision tools for a manufactured look?
>>
http://polycount.com/discussion/168610/proboolean-dynamesh-hardsurface-workflow-tutorial/p1
>>
>>597532
Im sure Boolean is better now, but when i used to fuck with Boos the time gained was time lost later having to fix all that fucked up geo
>>
>>600005
>>Im sure Boolean is better now

You bet yer ass it is. Oldfag here - if someone seriously suggested a boolean workflow even as recently as 5 years ago - they would've been laughed out the room. But now, not only are the actual boolean modelling tools way better - so your actual hipoly object looks better than it ever did with like, no work - but now the export/cleanup/decimation tools in things like F360 are way better and less mysterious and produce more predictable results.
On top of this; RT engines like UE4 care less and less about polycounts with each iteration now. The idea of 'lo-poly' being an absolute necessity is declining - In a lot of situations, everything is what you'd call mid-poly.

So - certainly in a lookdev scene - I frequently get away with running two models out of F360, (one with lower tesselation settings, one with everything maxed out) - bring the cheaper one into Zbrush, apply AUV or PUV mapping with one click - run it back out for the bake and it's all ready for painting. I used to decimate and remesh but I don't even bother with that now.

No retop. No UV unwrapping: It works fine. Yeah - in a shipping product or a situation where you needed to be use pre-existing tilable textures this would *not* fly at all, but during lookdev iteration? Totally cool.

It's 2018; relax, crack open a Heineken, and be at one with the Boolean logic.




Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.