A few Xnormal related questions and such

seforin

ok I was messing with xnormal over the weekend, in the past I had minor success but after seeing how well it handles bakes I have many questions I need to ask this time around




1- why is it every time I bake out my maps the green channel always needs to be flipped?


2- When Im working with my particular mesh the REALLY high detailed version will not render in the 3d viewer. So When projecting the cage I need to build its own cage via max separately (taking a copy of the low head adding a push modifier and projecting from there)

BUT when I bake there areas that are stretched and not showing details, but when I look at the cage it encompass the entire high poly head. Why would that happen?


3- Is there anyway to transfer the Cage information over from max instead of creating a copy of the low poly head and having that be a faked cage?

4- how do some of you other more experienced next gen modelers handle doing this process (maybe my process of doing this is backwards? lets rule out using max only due to limitations of max where it can only go up to displaying 5 million polys but not baking them.)

kary

1) You did not set the Y channel in the right direction when rendering the map.

2) Stretched/squashed UVs, not enough pixels to capture the detail, might need to show a picture of this to clarify the question.

3) Thats what a cage always is (a copy of the low res mesh that is deformed to have the light go the right distances), but you can use the sbm export installed by xNormal to carry the modifier cage info.

4) Not sure what you mean, look at EQ's weapon baking thread, Cholden/Adam's sketchbooks etc, you can pic up what they do via their work and ask specific questions to clarify problems.

Joao Sapiro

try not using the cage, set the ray distance to something like 2 or 3 , and if it gives you trouble then use the cage, but so far never needed to use cage

Pedro Amorim

dont forget to exploded your mesh if you have diferent pieces of the lowpoly

seforin

thanks, I just finished building the final low poly last night I'll post screenshots of some of the more individual specific questions I have with this mesh, thanks again guys

jogshy

seforin wrote: »

1- why is it every time I bake out my maps the green channel always needs to be flipped?

Because some apps and engines use the texture data otigin at top-left(Direct3D based) but other bottom-left(OpenGL based) and because ones use a left-handed coordinate system(D3D) and others right-handed(OpenGL).

If the app or engine you're using to visualize the normal maps generated with xNormal uses the OpenGL's conventions, in theory you won't need to flip it.

2- When Im working with my particular mesh the REALLY high detailed version will not render in the 3d viewer.

The graphics card has limited video memory. If the highpoly mesh exceeds the card's limits it cannot be renderered. That's why DX10/Vista and OpenGL can virtulize the video memory like they do with the system RAM... unfortunately DX9 lacks that feature.

4- how do some of you other more experienced next gen modelers handle doing this process (maybe my process of doing this is backwards? lets rule out using max only due to limitations of max where it can only go up to displaying 5 million polys but not baking them.)

Use a x64 OS like WinXP Pro 64 or Vista x64. The x86 ones ( WinXP Pro, Win98, etc ) are limited physically to 2 or 3Gb(this last one with a nasty trick ).

ZBrush can work with very large meshes... and xNormal can load ZBrush-exported OBJ files twice faster than any other app(because ZB uses a special .OBJ header).

Just my 2 cents.

seforin

well I built a new rig recently for a update to the times and to really have something powerful to run games / 3d based aps

currently im running


winxp64 bit
8gigs of ram
1 gigs graphics card


I find it weird that I can load the uber high poly mesh in max and show wire frame but it wont bake out for some reason, but in x-normal it bakes out but wont display in the 3d viewer so I cant play with the real time cage settings?

I think my system/ card can run DX 10 but im not sure, how can I find out/ test that?

also stupid question where is the settings to flip color channels prior to bake outs? (so I can flip green channel? )

kary

You don't need to use the absolute highest poly version to set up your cage. If you have a 1 or two mill poly version, set it up with that, then render it with the 10million version.

Max does a lot of things to not show all the polys you have in your scene (thats why 2008 and 9 are so much better in the viewport). I don't think xNormal does that.

DX10 should be irrelevant to making cages and stuff, but you would have to be running vista. Check you card manfucaturer's site for dx10 compatibility.

Y Flip = Baking options > green button with and elipsis on it ( the ... )

seforin

well I was working with the mid poly before , before going to high version and prior to bakes, but the high poly hada few wonky areas I needed to fix as far as the low poly goes with its cage, its just annoying tweaks that need to be done, I'll post more tonight with screenshots soon of what im doing/ mean


Thank you so much everyone

jogshy

seforin wrote: »

but in x-normal it bakes out but wont display in the 3d viewer so I cant play with the real time cage settings?

How many polygons have your lowpoly and highpoly models, pls?
And... are you sure the highpoly mesh is marked as "visible" in the highpoly list and also that you checked the "show highpoly meshes" inside the viewer? Do you see any error dialog telling you that your VRAM is not enough to show the mesh? With a 1Gb card you should be able to display a 16M poly mesh using DX9 with the latest drivers... Btw, whwat graphics card are you using?

Have you tried to change the default graphics driver from DX9 to OpenGL using the XNormal plugin manager? ( the plug-in icon near the close button ).

And yep... DX10/Vista can help there due to the VRAM virtualization of the new WDM Vista driver model.

I think my system/ card can run DX 10 but im not sure, how can I find out/ test that?

Yep, more than enough.

also stupid question where is the settings to flip color channels prior to bake outs? (so I can flip green channel? )

When you render the normal maps in xNormal use "X+Y-Z+" as swizzling.

seforin

well thanks to johny I had some more progress, and just got a better understanding of a few things

the head I was using as a test was roughly around 9 mill polys, currently im running a nvidia geforce gtx 280.


AS far as chaning drivers or anything like that no I have not done that yet. Im rather newbish to your program, I Appologise for that, Im trying to learn curve this as fast as I can. I think with johny kicking me a few strings of information to my face I kinda figured out alot of things.


I stoped using cages and used the ray trace playing with the numbers and got some nice results though the control of the distance is rather confusing at first (Trial and error mostly)

something I dont get is For example I have the hair of my character and he has this little swoosh thing and I basically unwrapped it as a rounded piece, (look at pic for example)

And the mustache was basically the same principale.

now the high and low are identical and matched up in every retrospect as far as the models lining up so it shouldnt be a issue with the model itself.

but for some reason I get these weird stretch bakes I cant quite figure out, im still trial and erroring it right now

oh also when I was making the mustache it was a seporate unit baked from the head (Exploded mesh) originally I had tried ray tracing but it didnt work, since the high poly of the mustache was smaller I looked in the viewer and played with its cage manually that way, but for some reason the results still baked out weird versus the cage (Yes I saved the new low poly cage from the 3d viewer and replaced them in my setup etc) I have a feeling alot of this is just things people more experienced have dealth with or worked around, but I just want to make sure its nothing program/ performance/ workflow wise Im doing wrong at this point.



well these are the results I got so far







thanks for actually responding to me personally about this matter, guys like you and ryan clark kick ass for that reason!

Michael Knubben

Those are the edge padding. I don't know what would make them so prominent on your model, though. Anyway, have you tried just painting them out? You'll have to learn to do so eventually, as I've never had a 100% satisfactory bake. The amount of time it takes to get something to bake out absolutely perfect is a lot more than it takes just to paint out a few errors.

seforin

yea I figured to paint the edges but before I did any sort of painting to the normal map I wanted to find out if theres something process wise im doing incorrect.

Joao Sapiro

yeah, you are trying to bake something that didnt even need normal bake and tht could be achieved with crazybump

lol i love to make useless posts

EarthQuake

Make sure that if you didn't set up a cage, that you click off "Use Cage" as its on by default(dumb!) if thats not the issue with some of this stuff here, try adjusting your ray distances until you get better results.

As far as the edge padding showing up on your model, i'm not entirely sure if thats what is happening, but try and make sure your uvs havent shifted at all on your mesh you're baking with, and hte mesh you're previewing with.

jogshy

Yep, the strange border expansion is the "edge padding", dilation, etc... and it's ok.
It's used to avoid problems with mipmaping when you use normal mapping.

About the mustage, I think you need to use cages to avoid those artifacts. The uniform ray distances method is only good for very simple simple objects.

seforin

earthquake: I also had cage unchecked before I messed with the raytracing.

jogshy: I did mention I used the cage for the mustache but it still caused issues and came out like that which Im a bit unaware on?

I will continue to use xnormal and test more tonight when im home, the head was really the only piece by itself I really had a issue viewing/ using a cage on and after a few bakes I got it to work pretty close ( though the swoosh piece of the hair I cant figure out still..)






off topic, asside from the vids on the website of xnormal, are there any artists who show workflow use of the xnormal that they use specifically online/ tutorial wise?

I was watching the old tutorials when I was originally using the cage as a 2ndary mesh and such and was unaware that xnormal has its own format now that can export from max etc.


If there updated xnormal tutorial based stuff Please let me know I wish to view and see if I can integrate things into my workflow with this