Hard Surface Modeling - Standard SubD Modeling VS ZBrush VS Houdini Booleans/VDB
Many months ago I felt that SUBD modeling was really holding me back to produce highly detailed objects/characters/vehicles, like the simple fact of, adding a manifold extrusion would create loops throughout the whole object, I'd spend 2-3 more time dealing with loops and fixing triangles that are messing up with the shader than actually modeling, every single new detail would lead to a bunch of new puzzles to be solved on my mesh.
From there, I decided to explore into other softwares to learn other workflows looking for faster and more effective ways, I started modeling in MOI3D and I could definitely see the potential, but eventually I found out all the limitations within the software, specially the difficulty to make a little more organic shapes, or god forbid when you try to extrude a face and dissolve that edge, or the simple task of filling/closing a hole, or when you can't calculate the bevel/chamfer your trying to do, MOI3D became a good tool for my arsenal depending on my need, but again I found myself stunned by the software nature and limitations.
I decided to give a try and do hard surface on Zbrush, it's my 1st week studying like a madman this software using ZModeler/Dynamesh/Zremesher combined with a lot of tricks, I am doing a course by Henry Chervenka: https://www.artstation.com/3dviper (Dude made a whole train and a Helicopter ONLY through Zbrush)
But then in the middle of this I ask myself which path should I take?
I am in a crossroad that the more I study, the more possibilities I see and the more uncertain I become on what should I do?
Should I keep studying as much as possible and learn all the tools in depth and then I can decide what to do? But then why you have people like Henry Chervenka and many others decide to make a whole helicopter in Zbrush instead of following the standard way of modeling in Maya/Blender/3DSMax? Just out of convenience?
If say your going to model something that, is going to take at least 100 hours of work, for me, I dread big time the fact that I might be doing something that could be done elsewhere maybe 2-3 times faster!
For hard surface modeling, what guarantees that, the pipeline adopted by companies/enterprises is actually the best? Or maybe I am just so impatient with the process? Should I conform with what I hear, or should I explore even further and find the best work flow for me? Maybe I am too impatient and should just stick to a single path and master it? I have so many questions, once you start studying and seeing the possibilities, I find it so, chaotic, what are your thoughts? What is your experience? Do you guys have any guidance?
Thanks in advance!
Replies
That guy using ZB to do the helicoptor tutorial is great if you want to learn how to use zmodeller/remesher/ and polygroup workflow to do pretty cool stuff, super useful to know when you are wokring in ZB, it's essential workflow paradigm and applies to all kinds of ZB work> But imo you can just poly model the helicopter stuff lot of faster, there are schematics, measurments and it's made of cylinders and normal crap. it just looks mega complex all put together.
For me the reason to use ZB for hard surface over another tool is the feeform concepting part that leads to things that wouldn't come up easily or arent easily solvable when using poly modelling tools. The whole polygrouping, extracting, boolean, grouploops, zmodeller, zremeshing, creasing or whatever 5 other tools in what ever order you need to use any give part to get the hard surface working is imo a bit tedious and convoluted in comparison to just normal poly modelling.
I think part of the reason people stick to one tool is actually to remove some of that chaos, it's just a distraction and leads to less productivity, grass is always greener etc.
Hello,
This is a tough topic to comment on because you seem to have already decided that polymodeling these detailled helicopter parts in Zbrush is somehow a good thing, or something impressive somewhat ... even though this is objectively pretty much the most tedious way one could ever chose to tackle such a task, and by all means the least desirable way to do it in a production environment for a multitude of reasons. When things as simple as importing a set of image paintovers showing some edits to perform, or even showing/hiding or transforming multiple parts at once requires dozens of inputs ... things get pretty slow :)
However not everyone is wired the same way and it is possible that this artist is somehow getting some pleasure out of the process when working in isolation - after all and as stated above there is indeed something to be said about being able to do something from start to finish in a single environment, without ever having to toggle apps and see ones desktop. So perhaps that's just what he is comfortable with. Who knows, maybe he spent years with only access to a low-end workstation with a CPU decent enough to run Zbrush but no video card and developed habits from there. No way to know.
But hey, up to you to decide for yourself anyways.
On a side note you might want to also show of your work and/or a link to your portfolio because this would help clarify where you are in your learning journey. What I mean by that is that once one is familiar enough with polygon-based subdivision modeling, it then doesn't take much effort to compare the workflows in different apps to see which one is the most appropriate for the task. Therefore I can only guess that you are still very new to this and probably not yet equipped to make these comparisons.
This
"... even though this is objectively pretty much the most tedious way one could ever chose to tackle such a task, and by all means the least desirable way to do it in a production environment for a multitude of reasons. "
@pior @oglu @Ghogiel
First of all thank you very much for taking the time to respond to my question, as requested from Pior I am linking my artstation and other things I did outside the Artstation for refering to where I stand in skillwise
PS, @pior what you think? Should I focus more in the standard way and don't even bother withh hard surface with Zbrush or?
Thanks in advance!
Now another thing I forgot mentioning @pior is that, certain parts of the modeling from that course, more specifically this one:
I can model this SOOOO much faster in zbrush than I would in Blender, and with amazing topo, I was like studying this specific piece and trying to recreate it in Zbrush VS Blender, Zbrush won by a far far margin, and I have only 5% of the experience in Zbrush compared to modeling in Blender, and by thhe way, great topo too:
Hello !
Well, this explains a lot :) First of all congratulations for your willingness to learn and study, you seem to have a great attitude towards all this.
I would say that the pictures confirm that you are in a bit of a self-made echo chamber so to speak. You've obviously found something that works well for you, and it is reinforced by this helicopter/tank tutorial you're looking at. But there are many factors just as important (if not more) than just speed - accuracy and ease of editability being two of the important ones.
Similarly to how you quickly put this part together in Zmodeler, someone could have done it very quickly in CAD too, or even using some kind of boolean approach with 3DCoat voxels ; but in production you also need to be able to edit things quickly and cleanly without having to redo everything from scratch all the time ... and the designs are more often than not way more complex than the part you show here :)
Now one could spend hours trying to explain that in prose but I think it becomes self-evident when looking at some examples. Here is the typical kind of modeling expected by studios when hiring a highpoly modeler :
Now of course there are subsets of that with some artists doing it fully in subdiv loops, and others leveraging things like creasing values and procedural bevels. But the overall point is : the workflow needs to accommodate any shape, and allow for clean edits at any moments notice. And the reason why most productions settled for the workflow shown in the video above is because it is the only one allowing all this. CAD may be extremely clean and precise, but it is also extremely tedious and slow to edit on the fly ; voxels/booleans may act like clay and allow all kinds of freeform details, but they make big soft-selection edits impossible. And so on.
(BTW, I don't see how the part you show could possibly be done faster + more easily + more efficiently in Zmodeler compared to a regular polygon modeling environment with equivalents tools, since Zmodeler is literally just polygon modeling, just retrofitted into the Zbrush UI ... with the essential operations like component selection and scene manipulation being very tedious. Or is this because of a reliance on things like booleans and Zremesher to get the holes and topology done ? Because if that's the case then it's not an apples to apples comparison at all. And most 3d apps have similar auto retopo solutions anyways. And lastly, this part doesn't look finished to me anyways, since it doesn't have any rounded edge quality along the sharp edges ...)
But all that said it all depends on context. If you are your own art director then nothing prevents you from using whatever method that you feel gets you to your end result faster. And indeed, some projects can very well be tackled with an unconventional approach. But as soon as the workflow introduces friction in the context of team work, then it becomes a problem.
If anything I believe that you'll naturally end up refining and adapting your approach in relation to the bottlenecks you'll be facing - either because of the specifics of a design, or because of the requirements of working with a team.
"Or is this because of a reliance on things like booleans and Zremesher to get the holes and topology done ?"
It certainly is. And that's the only way to make ZB work imo. I wouldn't hold it against it, as you say it's operations that can be done anywhere these days.
Anyway If one has followed the tutorial, that rotor part would have creases for most of the subdiv, but also some small beveled parts for some holes, but you can't see them very well in polygroup view (and that image is all one polygroup by the looks of it> seems like a mistake to clear all the carefully set up polygroups that are basically required to get model and zremesher to work)
in my mind I just have far better edge control Max/Maya. The precision and ease of editing and just being able to stack symmetry modifiers, ie to use both radial and planar symetry at the same time makes doing something like this there fast enough.
Plus it's already worked out when following a tutorial, so it's orders of magnetude easier and faster when you don't have to workout and breakdown something complex then use ZB tools to make while having accuracy, good topology, good subdivision, good modelling speed, ease of editing.
Not to mention that zBrush is "imprecise", as there aren't any unit measurements or precise movements and adjustments etc. At least it isn't immediately obvious or clear there is. If you need those holes to be exactly 3cm wide for example, it'd be a lot more cumbersome and strategy to get that information in your work in zbrush.
It's certainly fun, fast, and easy when done right, but shouldn't totally replace your DCC.
@Ghogiel @pior
Thanks again for taking your time to answer, though I do agree with you about accuracy and stuff.
The thing that made me impressed is the fact that, In Zbrush, I can simply model it and then prepare the mesh for the Dynamesh, then getting that really good topology, it sometimes creates some issues but it is a matter of taking it to Maya/Blender and fixing it in a matter of a minute, does that make sense?
Now when I am modeling say, in Blender/Maya, I can't just focus on creating, I have to model/create and at the same time be worried about the topology because that might cause shader issues, then be fixing triangles specially in curved areas, and do a lot of clean up specially if using Boolean.
In Zbrush I could make that piece really fast (12 minutes) and do not worry about the topology, because Zremesher will do a great job with it, to summarize, I can skip worrying about the topology in Zbrush (with that specific piece, at least) while in Blender/Maya I would have to worry about the whole topology does that make sense? And then if necessary because I want more accuracy, I could for example, throw the Zbrush remeshe'd obj back to Blender/Maya and fix the topology, but that would be something really quickly compared to creating that same object from zero, what is your thoughts?
Heya,
Well, now things are starting to make sense (in a roundabout way !) :
But as said, whatever works ! And it's not always a race anyways. Just keep in mind that just because it feels fast and efficient, doesn't mean that it necessarily is :D
To put my money where my mouth is, here's 12 minutes in CAD (mostly spend on the 2d drafting, and not quite finished overall - but with the advantage of dimension accuracy) :
Now I am not saying that CAD would be "better" in that case, nor "industry standard". Just making a point really :D
Agreed, but I am looking for Zbrush in this case because then I have good topology, something I can also unwrap it lately for texturing if necessary, in CAD were gonna have a problem just like I would have in MOI3D right?
Now say you use Autoretopo as you've mentioned, would it also have a good topology as I have shown in this picture? Or how should I prepare a CAD before doing the autoretopo?
I am looking for both speed and good topology, even if necessary to send it back to Blender/Maya from Zbrush, make sense?
By the way HUGE thanks, because all this things you've said gave me a better understanding and better foundantion of what I should do!
not sure if it was said already or not, but it's fine to use both workflows and software together.
If you know that you can make some complex shape quickly in zbrush you can just get it done, then send to blender or whereever and cleanup topology if needed to make the part compatible with rest of the model.
I'm not the expert but that's a workflow I did a few times - basically just hopping from one software to another based on what sort of thing I need to make. But in general for game models I am finding life easy if I avoid zbrush for many cases and stick with simple, straightforward modeling approaches.
I'd also ask you to question yourself a bit when you say things like "good topology."
what does that mean?
I think its better to think in terms of define specific technical criteria model has to conform to rather than some nebulous idea about best practices.
To the OP : the point of showing the same thing done is CAD wasn't to illustrate any kind of "best practice" ; it was to show that the 12 minutes argument is irrelevant, and that the point you are making about fun of use and automatic topology doesn't necessary prevails over, for instance, the necessity for the precise dimensioning of some parts, or the ability to create and edit models through a modifier stack. That's all :)
That's the killer for me.
Zbrush has its place in the process but for the final meshes it's just too loose and clunky for my liking. I like cad packages more and more these days - or use max because it's also very good at this sort of thing.
Understood, thank you very much, it was extremely insightful!
depends on what kind of role you like to be, if you are working as a concept artist, then ZBrush gives you more freedom to more organic shapes, but if you are working as an asset artist, the model needs to be production ready, it needs to be clean as it will be easier to texture, rigging and render in the down steam. In that case, the traditional sub-d model with nice edge flow, and edge loop is the way to go.
3ds Max/Zbrush: Proboolean + Dynamesh hardsurface workflow tutorial
Not sure exactly what you are after but the above polycount link was an eyeopener for me. If you don't want to go CAD then there will be accuracy issues as polys will always be an approximate. I think the above link was of interest to most people here because it provided a great method to produce a high poly for game asset baking. 3DSMax in the title isn't really necessary. Just about any app with up to date booleans will do.
Anyhow if you don't know the thread already, have a look.