ZBrushCentral

Maya / Zbrush Godzilla Multi-UV Set Tutorial (ongoing)

Hi guys, I’ve been working on a multi-day Tutorial/WIP for setting up a creature using the Maya->Zbrush->Maya pipeline. It deals with multiple UV/Texture maps on a single watertight mesh and focuses on getting good animatable topology.

http://www.treyharrell.com/v3/?cat=7

I took a break after day 4 (initial surface sculpt) waiting to get up & running on Z3 for surface detailing (got it last night when the store opened!), but there’s a ton of info there already that I hope some of you will find helpful, and a timelapse massing/rough sculpt video.

Day 5 will be coming soon, once I get the hang of exporting displacement by hand again and play with full 3D alpha projection a bit.

Attachments

day5preview.jpg

i’ve been waiting for z3 to work on my big-g model too, yours is looking great so far with the detailing. GMK was by far the best of the G films with the exception of the original, i decided to model mine on a watercolor i did. I’m interested to see how the multi-uv’s and polypainting work out. Maybe we can have a kaiju battle when we’re both finished. oh i made mine originally from zspheres then went into silo to add the teeth and spines also made with zspheres. godzilla1.jpg

The kaiju battle thing actually sounds fun. Would make a great challenge on CG Talk! :smiley:

Spent the evening getting a feel for some of the new features in Z3. Learned to have a healthy respect for using the snakehook brush in 3D with 10k polys and 45 second undos.

Cavity mask + invert + gravity is a cool trick. Or inflate. Or both. I can think of a bajillion uses for cavity mask.

Still got a little bit of cleanup, and some unwise experiments to undo on the backspikes, but I’m pretty close to my “day 5” stopping point and I should have it up on the site this weekend.

Side note: I really like being able to record directly in Z, but why on earth is quicktime export capped at realtime playback?

Anyhow, teaser image from where I’m stopping tonight and starting my day’s writeup:

this is looking great so far, just in case you haven’t noticed you can’t pose all subtools at this time so i hope when you add the teeth and nails that doesnt present a problem, at least thats what ive read. those spikes are so great, i love how organic they are, with mine i just copied one over and over, but i still havent done any sculpting on it yet, hopefully soon.

The teeth and nails aren’t worrying me at this point… it’s getting anything out of Z3 with sane UVs or any surface detail whatsoever that has me perplexed. I can do the accessories directly in maya.

So, I can get a 16 bit displacement map (that will melt 3/4 of my detail), or I can export a 20 million poly .OBJ and wrap deform it to rig. So, no go for either of those for real production work, or teaching people how to do things “correctly” for real productions… because you can’t now.

Let’s not even get into texturing quirks with anything but Z3 generated UVs at this point. Multi-UVs don’t appear to be working at all. Any honest question on the boards tells you to “RTFM… beta users struggled, so can you!” … yeah, well, so RTFM doesn’t do any good if it doesn’t actually work and the manual is still talking about ZB2 workflow or features that were removed before release.

I’m kinda at a strange point. My tutorial series is going to either stall until the DE3/MD3 plugin is released or I’m going to have to roll back to Z2 and start railing on the WTF-variety omissions in Z3 (MD, Mapper, AppLink)… same with my real work. I’ll say that the stability, memory issues solved and upgrades are fabulous… but until I can actually use what I create in Z3, there’s no point whatsoever in even launching the thing.

cannedmushrooms posted this about polypainting on multimaps maybe this could help

http://www.zbrushcentral.com/zbc/showthread.php?p=356730

Rockin! Thanks a ton for that. Splitting the model into subtools seems to be the key here before export. This will probably allow me to do multi-displacements, too, although I almost gave in and did a giant 8k autoUV .map instead of tearing my hair out.

Upgraded to 8 gigs of ram, a core 2 quad and xp 64 over the weekend… I can actually export the entire 18 mil poly .obj and get Maya to read the thing without displacements at all now, but that’s a bad habit to potentially get into :slight_smile:

Anyone figure out how to get color->texture to do anything over a 1k map yet in Z3? No matter what I try, or what settings my “new” texture has, it always comes out at 1k when I try it.

–T

this is probably the wrong answer but i know with z2 if you just choose color texture it gives you that default 1k map, i think you need to pick your texture size first then hit new then do the color transfer, try that.

can i inquire how much a pc like that costs? i plan on upgrading next year.

spaceboy,

Thanks for the idea…

No go on that front. Don’t know if it’s something up with the .ztl I imported from v2 to 3 or what, but I create a new texture like in Z2 at the desired resolution, use color->texture and nothing happens (it stays white).

If I have texture “off” and I do this, I get a 1k map… which is about 1/16 the rez I need for closeups on this map :slight_smile:

–T

try asking cannedmushrooms in this thread
http://www.zbrushcentral.com/zbc/showthread.php?t=45316&page=2&pp=15

A few notes on how to get Multi-UV displacements as good as you can get from Zbrush 3 to Maya. It seems that the alpha factor will only get calculated accurately for each map if they’re separated as subtools. Well, reliably at any rate. Sometimes it will do it correctly, sometimes it’ll calculate the same depth for each part if you do the old-school shift-control-click to hide each piece workflow.

  • Before you start your export, save your ZTL. This is essential because you’ll need to get your watertight mesh back for tweaks after this point.

  • If you’re using the UV Smooth option in displacement, go to the lowest subdivision level and calculate a 4k displacement over the entire model, never minding the overlapping UVs. We’re not going to save this one, we’re merely interested in smoothing the UVs on our watertight mesh before it is split.

  • If you did the above action, export your new .OBJ file

  • Under the subtool palette, use group->split. This will separate each of your UV sets into subtools.

  • Hide all of the mesh and show each piece individually. When each piece is shown, create a displacement map at your desired settings. I used adaptive sampling, mode highlighted, 4k map, no UV smooth as I smoothed the UVs before exporting my geometry above, we don’t want them relaxing again.

  • As each map is created, go to your alpha palette, and click on each map. Take a note of the alpha depth factor and write it down for each piece (if you don’t save it as part of the filename). This is crucial to create a seamless mesh.

The next part is optional, but I do this to sharpen my displacements that get a little muddy with 16-bit output.

  • Go to two subdivision levels from your maximum (ie if your model is max at 5, go to 3), and generate a second set of maps with the mode button unchecked. There’s no need to note the alpha depth this time as we’re using this as a bump map (because normal map generation is broken right now).

Now we go to Photoshop

  • Open all your maps, flip them vertically (image->rotate canvas->flip vertically), and save as uncompressed .TIF

Now to Maya

  • Import your model, and reverse the normals. For some reason, Z3 likes to implode polymeshes. Your bump map will be reversed if you don’t do this step.

  • Go to each of your shading groups and set up your displacement shaders. When you import each file node, go down to the color balance section.

  • In the color balance section we’re setting the number we wrote down as our alpha gain, and negative half of that value as our alpha offset.

Example:

Alpha depth factor of 28.1217 from Zbrush

Alpha gain in Maya: 28.1217

Alpha offset in Maya: -14.06085

  • Now you can attach your bump maps with a bump value of 1 as we didn’t use auto-levels or anything that would have shifted the midtone gray. As you do test renders, shift all of the bump values up or downward to taste, all with the same value. This will help to make the model sharper and more detailed.

  • Click on your model, and in the attribute editor turn off feature displacement

  • With the model highlighted, go to Window->Rendering->mental ray->Approximation Editor and create a new subdivision approximation for your model. The values in the Zpipeline guides are still valid on this point, so I’d follow their guidelines. One small note is that if you use Spatial mode, be mindful of the scale of your model. The pipeline guide may use a length of .010, but if your model is 9 meters tall, you’ll be subdividing needlessly. With my mesh at 9 meters tall, length of .5 was the best balance of detail, rendering speed and memory usage. The moral: start high (like with a value of 5), save at every step in case MR bombs, and do test renders each time cutting your length in half until you reach the sweet spot. (in my case, I went 5->2.5->1->.5->.250 (twice the rendering time, no visible difference in the render) -> .5.

The 16 bit maps are somewhat muddy and soft, but the addition of the bump maps make them a little crisper. Not ideal, but it appears this is the best we’re going to get until MD3 is released.

The same basic workflow seems to be working with multi-UV texturing. Basically save your model then do a UV smooth before you chop it up into subtools and you should be able to get your maps out individually.

Hope this helps someone who’s stuck in trial and error mode on deadline.

–T

So, for some reason when I generated some maps last night, they were taking 50 times too long to render and coming out really bloated.

It seems that some strange combination of Maya 8.5 and Z3 will sometimes want alpha maps with the gain set to 1, which is Maya’s default, except their gain defaults to 0 (which should be -0.5).

So extremely odd, I’d been producing maps all night in 16 bit format, recording the alpha depth religiously, then one model decided it wanted to work at Maya defaults, and not the alpha depth that was recorded from Z3.

Hope this helps someone out there, and I was wondering if anyone else has experienced this and knows how to reliably reproduce these results? Would save some of the more annoying hassles related to dealing with the 16-bit workflow.

–T

Just to let everyone know, I’ve updated my series with Day 5 / Zbrush detailing for my Godzilla model. There are two videos on that page you can follow along with that show some volume raking through to getting the first pass of the surface done.

Day 5 - Initial Detailing in Zbrush 3 (2 videos + journal)

[cavity.jpg]

Day 6 will be getting the surface detail to POP in displacements, and a little bit of texture mapping. Waiting on the new plug-in bundle before I record that point as I’m still a little muddy heading into Maya land at 16 bit displacement + 16 bit bump, and if I’m recording this for posterity, everyone will want the new displacement workflow in there anyhow.

If anyone has any questions about techniques I use, or would like to see portions in greater detail, I can acommodate. I’ve been avoiding doing a voiceover as a huge percentage of the audience aren’t english speakers.

Hi from Bulgaria. I realy like your work it was so helpfull for me. Even i put a link for your work there :cool: .
While you wait you could see this workarounds it maybe helpfull to finish early:
http://3d-tips.blogspot.com/2007/06/zbrush-3-maya-85-maxwell-render-15.html