Main focus is creating game assets - weapons, props etc. My current workflow includes highpoly, retopo and UVing in Maya 2016, then doing all the texturing in Quixel.
I'm still a newbie in the CG area. I love Maya to pieces but I feel like I'm spending too much time making subdiv work and messing with topology I'm going to retopo anyway.
1. Are there other feasible alternatives for creating hard-surface highpoly models faster, other than subdiv?
2. Should I look into sculpting and what software might be the best for creating hard-surfaces?
Replies
I suggest to start with sculpt to see a fast preview of the shape, than go back to to subd for make it clean ( Im not a big fan of polishing too much the sculpt model)
cheers.
http://www.polycount.com/forum/showthread.php?t=128414
Also as Bek said, make HP with LP in your mind when you choosing number of loops, sides etc. It will save your time on making LP.
I haven't made the switch to Modo yet unfortuntately, but I make quite heavy use of the quad chamfer modifier in max. You can use this in combination with a turbosmooth by smoothing group for example.
How does Quad Chamfer with smoothing groups compare with just a regular double-smooth method?
I don't think any studio or freelance position would be okay with that.
Why not? if you can produce a clean quad mesh who cares how you got it?
I was under the impression that you were exporting the nurbs into zbrush and just smoothing the errors out and exporting a un-editable mesh back out for baking. My mistake.
It's a mess if you or anyone else have to make major revisions to the model. For you, you have to go all the way back to moi and redo the zbrush process. For someone else its more complicated. What if they don't know how to use, or don't have a license for moi? You've got your own proprietary workflow that only you can use then. The end result is only one factor when working in production/in collaboration with other artists.
Traditional sub-d means anyone can edit your mesh in any common 3D package.
The modo rounded edge shader is good but I wish we had a baking tool that could do it. MightyBake said they'd implement it but they're taking their sweet time and no word on how that feature is coming along yet. Imagine how much more work I could get done if I could just focus on the low-poly (which is the actual thing going into the game)... and not have to do twice the work just for some rounded edges. Ahhh... all this futuristic tech and normal maps are still a headache.
Thats the point of remeshing in zbrush. To quickly make a subdivision friendly mesh.
Umm, I think we probably have different ideas of what a sub-d friendly mesh is in that case. Being all quads isn't it, perhaps if the goal is to sculpt on it in zbrush.
However, if you're doing hardsurface work, a sub-d friendly mesh would be a mesh with the minimum amount of geometry placed in the right spots to make it easy to alter without going back to the drawing board.
Being able to do a blockout mesh then set the edge widths and not have to do support loops would save so much time, because both your high and low poly could be a single mesh just changed at the click of a button.
I imagine modo with rounded edge shader is far more useful but for now I'm making due with quad chamfer and it helps speed things up massively.
There's a lot of different ways you can use the quad chamfer modifier.
You can for example block out your shapes with it set to quad chamfer by smoothing group with a hardness of 1. So it basically just adds support loops. After that you can switch to edge creasing to fine tune the edge widths. And then you can always collapse the modifier and tweak the edge loops if needed.
Quad chamfer is great for blocking stuff out. Means you can keep your sub-d cages low and instantly see a preview of the smoothing. And afterwards you can just collapse the mod and start tweaking things. Or even just get rid of it entirely and manually do the loops once you're happy with the base shape.
Another workflow is doing a turbosmooth by smoothing group first to get nice round topology and then adding a chamfer on top of that. I find that this works just as well as doublesmooth, if not better. There are some cases where it doesn't work nicely ( tight corners for example). It's also nice that the meshes you end up with are nowhere near as dense as doublesmooth.
Or you can just do quad chamfer without any additional smoothing. Which works great for some meshes. For example I made some knurling a while ago and instead of smoothing everything and ending up with a really dense mesh I just put a quad chamfer mod on top.
You can set "edge bevel weight" or use "vertex groups' and put several different bevel modifiers each for each own group . Can make just support loops only, partly flat chamfers and partly rounded bevels ( with different vertex groups) , different width based on weight or vertex groups.
UVs are adjusted automatically (with minor errors sometimes)
There is a checkbox that prevents bevels overlapping
All that easily switchable and having base mesh editable while seeing it beveled.
http://community.thefoundry.co.uk/discussion/topic.aspx?f=77&t=86388
don't get me wrong, we use it a lot to stay flexible and fast, but it blows up way to easy
@bek: the uv update looks kickass, yeah yeah need to learn more blender, just, so lazy
Some additional info to where I'm at currently: I will improve my subdiv skills, since they seem to be essential, and also I took a look into MODO 901. Got to say, MeshFusion is great and the handy preset library even more. Easily dragging and dropping different bolts, panels and other floating geometry on to my high poly object is extremely time-saving. Everything else needs some getting used to coming from Maya, and also need to figure out the licencing, since Autodesk kindly provides students with a free 3 year licence.
Some other solutions I'm thinking of trying are: ZBrush and 3D Coat voxel sculpting. Unsure how great they are for hard surface models, but I'm interested to give it a go.
I even started looking towards Fusion 360, which is a simple CAD software.
Thank you all for your feedback!
That sounds.. awesome. I need to try it.
What ?!!
there is maybe a hand full of modellers out there on pernas level, don't even try to argue
that said, there are some seriously good modellers in maya, i couldn't stand how destructive it is, but thats just my oppinion on mayas modelling
Seems also like a button called "bake a high quality normal map without distortion" should be available now in every program. Theres only so many variations on what all of use are doing, only so many angles, etc. I once heard (could totally be wrong) that Autodesk makes most of its money from Architectural/industrial/civil service stuff so i wonder how small our voices are to 3d companies.
Like a skewmesh projection render, can't they just automate that now using the aforementioned togglable support edges on and off? Aren't we at the point where we can work with the HP and LP right on top of each other, toggling back and forth to make changes, the HP acting as sort of a interface to the LP and vise versa?
Source: http://www.bytehazard.com/articles/vertnorm.html
Like this example, vertex A's normal is shifted to the right because its connected to faces u and v on the right side, but only face t on the left, skewing the shading because the vert normals are all average, despite the fact that u and v are one quad plane. Correcting this kind of stuff is what computers are great at doing for us.
But im still learning all this stuff, perhaps this is all like second nature to some of you.
But don't mind me, im going to keep popping my crazy pills as i try to learn about and figure out normal maps. I just got a perfect normal map from a hp box in max after i reset the cage, even though I had "use cage" unchecked for the whole time. Crazy pills!
Can you explain a little bit more on what the face weighting is? I assume it's related to some technique with vertex normals, but after googleing I got a bunch of rigging related results.
1 - The key is to clearly separate design tasks from execution tasks. Any workflow that does not take this difference into account is bound to be inefficient.
2 - You *should* get into sculpting, but not with the hope of producing "clay" hardsurface models that can compare to traditionally modeled hard surface models, especially if you are requested to output cinematic/AAA quality content. However sculpting is a fantastic way to create blockout models very fast.
For instance, let's say you are given the hull of a hardsurface spaceship to model. If you jump straight into "subd" modeling you will have to deal with a lot of inertia coming from the technicalities of the process (building subdivision-friendly topology while attempting to capture the broad shapes of the design, and so on). However if you spend maybe an hour or two in a sculpting package to mass out the big shapes freely, you will have a very solid base to reference for the later modeling stages. Now of course if your modeling package allows it you don't necessarily have to sculpt - you could use booleans, edge creasing, creative modifiers to give thickness to paper thin planes, and so on. There are many ways to get things done ... and this the area where all 3d programs are definitely *not* equal.
Also, what is true for an alien spaceship is also true for some parts of even the most realistic assets - like the handle of any gun, the complex body shapes of p90, or the organic lines of a motorcycle gas tank.
You are essentially missing the first and probably most important step, which is to create a blockout model (sculpted or not, as just explained) that will be a useful guide for later modeling. It will also allow your AD to give you early reviews and provide feedback on what to edit.
Here's an example (I should probably rename the first step to something like "basemesh modeling", but you'll get the idea anyways) :
http://i.imgur.com/udytRFB.jpg
Now of course if you do have access to powerful modifiers like bend and FFD, you could very well bypass the organic sculpting phase altogether by modeling everything in a flat and symmetrical manner directly with regular subdivision, and bend it in place after the fact. Use whatever gets the job done fast !
Last but not least, some art styles allow you to bypass the highpoly step altogether, by relying a somewhat dense ingame meshes or even models with hard edges/smoothing groups (think of all the recent WiiU games). But that is far from the norm.
Now as far as Modo goes? I simply haven't had the time to fully get myself up to speed to be fast in it, but it's been pretty tempting to try out the smooth edge shader, maybe I can get something similar going in Maya.
https://www.youtube.com/channel/UC3PHtDtdiV6-FlLmqu8OUCA/videos