Max Normal Map Baking Color-Set
polycounter lvl 12
Hey to all,
I have several question, and hopefully someone can help, and mainly concerns with Max, ZB and XN to an extent.
1 ) How would I go about baking Normal Maps in different parts of my low-poly. Example; Girl in Fishnets and Boots with a Skirt and Jacket?
(High-Poly Fishnet, Boots, Jacket) - (Low-Poly Body, Skit, Jacket).
Naturally in this case, the Rays being shot out of the Low-Poly skirt and Jacket will hit the fishnets and body to a certain extent, and vice-versa. In ZB, I can fix this since it's all about exporting that each map individually and combing them together, but not always possible since sometimes, the low poly model won't match up perfectly or isn't a cage.
How would I go about this in Max and XN? I use XN alot, so a solution in that area would be grand, but I also heard that the new script called RTTAssist might just do what I need a la ZB style. Advice for this part?
B ) Are cages as simple as being a low-poly mesh that has been pushed out so that it's ontop of the High-Poly mesh? Is there something more to it?
Γγ ) What's the best way to Normalize a Normal Map? Especially after having been edited in Photoshop?
4.2) Is it true that ZB and Max only bake 128 channels of the Normal Map as opposed to the full 256 and that only software which does this 'correct' bake is XN?
Cheers for the answers, and ciao.
I have several question, and hopefully someone can help, and mainly concerns with Max, ZB and XN to an extent.
1 ) How would I go about baking Normal Maps in different parts of my low-poly. Example; Girl in Fishnets and Boots with a Skirt and Jacket?
(High-Poly Fishnet, Boots, Jacket) - (Low-Poly Body, Skit, Jacket).
Naturally in this case, the Rays being shot out of the Low-Poly skirt and Jacket will hit the fishnets and body to a certain extent, and vice-versa. In ZB, I can fix this since it's all about exporting that each map individually and combing them together, but not always possible since sometimes, the low poly model won't match up perfectly or isn't a cage.
How would I go about this in Max and XN? I use XN alot, so a solution in that area would be grand, but I also heard that the new script called RTTAssist might just do what I need a la ZB style. Advice for this part?
B ) Are cages as simple as being a low-poly mesh that has been pushed out so that it's ontop of the High-Poly mesh? Is there something more to it?
Γγ ) What's the best way to Normalize a Normal Map? Especially after having been edited in Photoshop?
4.2) Is it true that ZB and Max only bake 128 channels of the Normal Map as opposed to the full 256 and that only software which does this 'correct' bake is XN?
Cheers for the answers, and ciao.
Replies
B. Cages: Cages can be simply a "fat" copy of your low, however, sometimes the mesh normals on your cage will differ(say you want hard edges on your low, but NOT on the cage, as that will cause splits and seams). This tends to come down to exactly how your app deals with cages.
C. Nvidia normal map plugin for photoshop can normalize.
D. 128 bit normals etc: No. Just simply no. I'm real curious as to where you got this idea?
Going along with EarthQuake, you can probably fix most of your problem by manual adjusting your cage...but if you are looking to get AO that excludes sections (such as maybe not having the skirt cast dark shadows on the legs) then the exploded way would help for that too.
I remember some thread a LONG time ago and a debate on it that Xnormals normal maps where more, for a lack of better terms, pronounced as opposed to Max which were more subdued to it only using a limited channel set.
Naturally, many people argue that it's all about the Green Channel being flipped and there being no difference when played around in Photoshop, so I was abit, again, for a lack of better terms, dumb-founded that such an issue could exist in anyway or form.
Thanks, that was a follow up question I had in mind, much obliged.
They are just different, one isn't "better" than the other. The important thing is baking in an app that is synced to the tangent bias that the asset will end up in, for best results(least smoothing errors).