I've also went into Substance Painter to bake them, while it does work it comes out fairly janky all the time, even with 8K baking resolution on, because of what I assume are the intersecting hairs.
My question is, what's the best way to bake xGen hair onto Planes where I can get good results? The best I've tried is to just project pre-rendered alpha textures made in Maya on Substance Painter, but again I would ideally like to bake these for better accuracy. Any help is appreciated
The below picture is the xGen version just for reference sake.
Replies
you are welcome to check my timelapse of the hair cap baking - not the lashes and brows - but the same workflow - maybe it could be helpful.
https://youtu.be/_IU4DkEWTwI
(sorry for the audio)
For the flow map - imho, the best way is to draw it using Substance Painter https://docs.substance3d.com/spdoc/flow-map-painting-143327274.html
There's also a great tool for creating lashes/brows from Olivier Lau HairTG - Surface, Facial Hair Edition - this one is free, for hair caps I believe he sells another version.
Anyway, it's a procedural tool, creates all the maps you would need, flow map as well - and he uses this SP flow map painting method.
does this also work on something thats not just a plane? say i have a creature full of hair, how good are those bakes?
need to test that, xgen/maya sucks big time
ah this is a for haircards workflow. still interesting but i also need a body full of actual groomed hair to bake down as a base (and lowest LOD). so far i never really had luck with haircards baked down
arguably you could get much the same information out using your flowmap + a root to tip gradient, just a question of how may things you want to bake i guess
Sorry for the thread necromancy. Just tried this out myself and it seems to work beautifully. @Neox - Great trick man, thanks for posting it. Should also work for Zbrush Fibermesh hair if you've remembered to assign UVs to it ( doesn't happen by default).
Just a quick sanity check - I assigned the 128, 255, 128 map as a tangent-space normal to the highpoly, baked that to the lowpoly tangent space normal then renamed it as a direction map and assigned it to the right slot. Did the rest of the bakes as usual after turning off the down vector map on the highpoly. It looks right, but just wanted to see if that's the correct process.
Crappy test hair cap here :
looks good to me
is it accurate? what do i know :D haha
so my thoughtprocess was, hey the cheapo way of flowmaps is, uvmapping the strands straight, top to bottom and apply the green map / down vector, why should this not work for the individual hairs if mapped this way?
Did a quick test in Fibershop using the hand draw mode and roughly drew out a similar pattern as my hair cap above. Flow map looked pretty much the same, so I'm calling this a win unless some mathematician turns up and tells us it's all wrong :D