Environment Artist: Highpoly Zbrush or regular 3d program?
polycounter lvl 8
I have been wondering this for a little bit. For environment artist do they create their highpoly in Zbrush or do they just create a high poly version in 3ds or Maya, etc?
I would think this also has to do with what they are creating. If you are creating anything that is Organic, I would think Zbrush is the way to go. Something like rock, wood structures or tree's, etc. I guess for Hard Surface you would stick to 3ds or Maya? Metal objects, or buildings, etc. Or maybe Sub-d modeling? I'm not sure.
Just wanted to ask this question to see what is most common work flow for the different kinds of assets?
I would think this also has to do with what they are creating. If you are creating anything that is Organic, I would think Zbrush is the way to go. Something like rock, wood structures or tree's, etc. I guess for Hard Surface you would stick to 3ds or Maya? Metal objects, or buildings, etc. Or maybe Sub-d modeling? I'm not sure.
Just wanted to ask this question to see what is most common work flow for the different kinds of assets?
Replies
ZBrush to make more organic stuffs, details, using a base mesh made in 3ds Max (ex: temple, rock, etc...).
Gears of War 3
I pretty much make the low-poly (adhering to the grid in the app for modularity), take it into Zbrush to make it uniform and then subdivide, sculpt away, export 2 versions (a high and a medium/decimated version), import the medium/decimated and the original low-poly into 3ds/maya, snap the low-poly to the medium/decimated and adjust things as needed (usually while keeping to the grid), make sure my uvs on the the low-poly are done, re-export that low-poly and bring it into xNormal with the high-poly for baking maps.
Things change depending on what you want and specific cases etc, like if you already have a pretty uniform low-poly then you can just subdivide that without remeshing and export the intact, adjusted low-poly from zbrush rather than doing the whole medium/decimate snapping/retopo thing. You can also do some diffuse painting in Zbrush if you find that saves time, and paint directly to the polys(verts). When you export the high-poly it should include this vert paint info, which can be baked to a diffuse using xNormal.
Some things to keep in mind is try reusing you textures between assets, that includes the UVd ones. So if you can get a clean bake the first time and have some solid textures, make multiple mashes and pieces from that same texture instead of starting from scratch every time.
The only reason I see these days to use 3ds max or maya is for animating and rendering purposes.
You don't work in the industry do you?
Same as him but I use Maya instead.
it's just my opinion on what should be used, if the industry is lagging behind so be it, but zbrush models can be transferred to maya or 3ds max with a click of a button usually so I don't see what the big issue is. I do realize many studios tend to stick with what they know though, doesn't mean it's the right/fastest way to do it.
For architecture it doesn't matter that there are clipping brushes and ways to do hardsurface in Zbrush. The measurements need to be snapped exactly for modular pieces, and in general it's a lot easier to do quality hard surface (at least the base of it) in a modeling application.
Trust me, I made a point of making architecture exclusively in Zbrush once 4R2 came out, and while you sure can do hard surface from scratch, certain things (like architecture) are just plain easier to model by hand (at least for the main form). You need the precision of modeling by hand and working with a grid if the goal is to have modular pieces (almost always the goal for games and architecture in general).
Zbrush is still used but it doesn't do the whole process for everything. There will always be an aspect of modeling somewhere else (either the base or the retopo).
This is how the industry works: Zbrush? Do you have time for that?
Believe me, if you do any architectual work inside of zbrush, I'm pretty sure you're out of job rather quickly save working for EPIC. Making something look good for portfolio when you can masturbate in zbrush for 2 weeks? That's easy. Making something look good in an afternoon using nDO and MAX? That's impressive.
As far as using it in production, you just need to learn when you are sculpting needless detail, if you are spending alot of time sculpting details that bake out to 1x1 pixel you're wasting time.
As far as the EPIC thing, I'm fairly certain they do alot of subd modeling and use zbrush for damage or organic things.
Thanks
Well perhaps as far as your personal skill level is concerned, but I dispute that as the general rule.
Hard surface from scratch in Zbrush is still just an unnecessarily tedious process, which at best, will turn out a mesh which will then have to be retopo-ed and properly subdivided in an alternative package in order to get anything which will bake down cleanly.
Currently, Zbrush simply does not have the chops to compete with a traditional modeling program in hard-surface modeling. You can't make quick changes, so if your edges are too sharp in a particular portion of your model, screw you, you have to essentially re-sculpt it to get the look you want and then polish it endlessly to get some level of consistency across the surface. Alternatively, in 3ds Max, I can just take 5 seconds to scoot my control edges over, and problem solved.
The day I can sculpt a Ferrari with precise to-the-millimeter dimensional accuracy and make small scale changes in an intuitive way in Zbrush like i can in a traditional modeling program, I would maybe start to consider a single-package workflow. But why would you needlessly restrict yourself by using an inferior method? Autodesk is practically giving away personal licenses nowadays, so I don't understand the reluctance to use it. If available, I would always like to use the best tools for the job.
This is coming from a guy who got into modeling because of Zbrush, and I love the program, but there are simply things which it does not do well, like any other software.
So yes, to the OP, like others have said, a combo of the two is currently the best way to go.
Or you can break your architecture into couple of modular chunks, mesh them in 3d app, sculpt them and bake proper Norm/AO/Curvature/height maps so that you have consistent basis for texturing/vertex blending e.t.c, forms are properly reading, edges are nicely chiseled and it does not look like someone photoslapped bunch of bsp surfaces :poly124:
Ha! That cracks me up.
I totally agree with you as I had submitted my work to a game studio and they came back with, "You need more lowpoly work."
So still in the process of doing that with nDo instead of Zbrush as Zeebrush was taking too long.
obviously depends on the specific prop but I love this workflow.
This.
Don't let the bluntness of some of the other posts confuse you(the OP) because using something like Zbrush only makes things inefficient if you are working inefficiently with it.
What's amateurish is pumping out diffuse textured stock assets that you slap with some photos. Low-poly is the idea here, but the goal is to make it look a lot better. If people look at your work and think that it's high poly then that should be a good thing.
Zbrush doesn't actually have to add a lot of time at all if you have a good workflow; in fact it's very good for making things that look like they would take forever to make in a short amount of time. In the case of architecture it lets you quickly add in unique details that have some character and quality.
Once you get an asset done you can reuse the textures for a whole modular setup. Add in some vert painting to break things up and you have a bunch of quality assets done in a short amount of time.
NDO wont handle making something low poly look like it has the form of something high poly, such as a low poly character model or pretty much anything more complex than a smooth surface. What picture to normal map software lets you do is quickly add in extra details directly from textures, but while it seems faster it pretty much gives the same results as using an alpha and painted directly to the normal map (which you can do on the actual model in something like 3dcoat). In fact alphas will give you much more predictable results often in the same amount of time.
Honestly you have to figure the workflows out on your own because everything can be done in so many different ways and there are not always clear winners. It depends on preference, perception, and the specific needs of what you are working on.
Also from my own experience, even with using Zbrush I can do an environment asset in an afternoon, so it really depends on how you work.
alphas in zbrush would be just as much work, since you will be re-useing the same few several times or you will have to, make them in PS or from existing sculpts.
but in the end there is no point in arguging one persons workflow over the others, since im sure were all going for different outcomes, and work in very different ways.
i know some like working in a very linear way, hp>lp>bake>texture and others like to work everything at once.
Some really inspiring work and awesome technique.
If you are doing hard surfaces and you can get the form down with the low-poly, then as long as the normal maps are made as surface details (like you get from NDO), then you can reuse them all over the place. Making things can be very quick and easy if you can find ways to reuse textures with different meshes, reuse meshes with different textures, and utilize materials in ways to add variation.