I'm having a few problems with Xnormal, currently when i download and install Xnormal the Photoshop Filters do not show up in my Filters tab in Photoshop.
Ive tried everything from downloading different versions of Xnormal to moving the filters from the download to the filters folder in photoshop but nothing works.
If anyone could help with this issue i would greatly appreciate it
The xN's installer should copy them automatically, but if it does not work try this:
Copy the .8bf files from [xN install dir]\x64\photoshop_plugins to [PS install dir]\Adobe Photoshop CS5(64 bits)\Plug-Ins\Filters
it seems that the plugin is still not installing with your technique am i doing something wrong?
Have you tried to re-install xN with the PS plugin components options check in?
The xN's installer should setup all automatically, unless your PS or Windows SxS .NET config is broken. To see is .NET is correctly installed just try to execute the xN's main app. If it runs correctly then it's ok. PS might need a reinstall perhaps. Btw, which PS and OS version are you using, pls?
Can you post a screenshot showing where you copied from/to the 8bf files, pls?
Don't mix x86 and x64 versions !
I have a problem. I cant get the ambient occlusion tool to work. Once saving the file it just comes up with an error saying "Attempted to read or write protected memory. This is often an indication that other memory is corrupt.
I'm on Windows 7 64bit. This is using either the x86 version or the 64bit version, and in administrator mode.
I have a problem. I cant get the ambient occlusion tool to work. Once saving the file it just comes up with an error saying "Attempted to read or write protected memory. This is often an indication that other memory is corrupt.
Are you using the default renderer, the CUDA/Optix or the hxGrid one?
Can you render the examples without problems?
Does it happen when the CPU is at full workload? Run an OCCT/linX test to see if your computer is really stable.
This is a little embarassing - I've always wanted to use XNormal for baking normal maps, but the hard edges don't bake properly. I always get seams as if all my UV's were connected.
Do I need to set the padding higher than 1 pixel? (I just realized I haven't tried this).
I'm sure someone has asked this before... but I didn't find anything asking about it.
I don't know if it has been already asked (to be honest I haven't browsed the 86 pages), but would it be a way for xnormal to keep the last processed hi-poly/low poly in cache, so it wouldn't need to reload it each time we launch a bake. And then it would flush the cache when the meshes list is altered or so.
The reason would be to save a lot of time when you are adjusting a cage and have to rebake all the time to see the changes.
From a technical point of view....dunno if it's possible but well I had to ask .
Not sure how to describe this, but I'm getting some really weird results out of xnormal.
I did a hipoly sculpt in sculptris and used xnormal successfully (the 1024 file attached) but later decided it needed more work.
I resculpted from scratch, and the mapping went all to hell. I did remap the UVs but that didn't help either. (512 map attached) Here are some examples of the snafu. Any insight?
Santiago, thanks so much for coming up with this AND making it free.
Thanks, that lead to a fix. Turns out my origin got screwed up when I added some geometry in Sculptris. I was able to fix my origin point for the lowpoly. I then used the cage method (oh my gosh, so much faster!) and the maps are baking out way nicer than they did before.
I didn't work off a straight subdivision, but I was wondering what the matchUV was for... I guess that would give the cleanest results possible?
I have little probem with normal maps generated with Xnormal, I have high poly and low poly model, I exported parts of model to obj and put it to Xnormal, I generated maps and I go back to 3DS MAX 2012 and hier is problem, how can you see on enclosed pictures, I have problem with normal map in render, I try it in UDK, in Static mesh viewer is lock good, but in game its look bad. Low poly have one smoothgroup, also high poly.
I have little problem with mormal map generated with Xnormal.
I have high and low poly model, expoted to xnormal in OBJ, low poly and high poly have one smoothgroup. Normal maps in 3ds max 2012 in viewport and in render window dont look good. I try it in UDK. You can see on enclosed pictures. I don't know where is problem, what I'm doing wrong.
I don't have much practice using xNormals, so I wondered if I am doing something wrong, or is it normal for the Program to generate AO Maps this streaky...and if I can do something to prevent/fix it...
So for some reason the ray distance calculator is getting stuck analyzing the low poly mesh. I tried triangulating but it didn't help.
This is especially weird because just yesterday I was able to quickly get a calculation for several high poly meshes being projected onto one low poly mesh.
edit: nevermind. Remember to check your relationship editors to make sure whatever you're putting in xNormal is separate from anything else in the hierarchy....at least if you have a shitty computer with no RAM.
@Piflik, seems you must increase the #rays. Also sure your ray distances/cages are well set. @MattQ86 The ray distance calculater takes some time to get values. You must stop it manually.
I wanted to suggest a feature for xNormal, that has to do with tiling textures.
So a usual workflow for me would be to make a mesh that is setup for tiling, bring that into zbrush, sculpt it, and when I am done, export an obj.
I would take this obj into maya and duplicate neccesary pieces to make it tile, and then export another obj which i would use in xNormal for baking purposes.
This was great until I started polypainting the pieces in zbrush, and realized that the polypaint data was lost when using maya as a go between. My solution was to offset the subtools in zbrush, but this is cumbersome and not the most precise process.
So I though it would be a cool feature if xNormal could offset a bunch of instances of the high poly mesh by a user defined value to allow for easier tiling.
I'm no programmer and obviously haven't thought of the feasability or practicality of this, but thought it maybe useful.
Hopefully this image illustrates what I am asking better.
A workaround is to export your polypainted meshes from ZBrush into a OBJ. Load them into xNormal and save the mesh as SBM. Then use the SBM importer for Maya so the vertex colors can be load into Maya.
- Recompiled using Optix 2.1 ( == faster rendering speed and faster data precomputing )
- Added an option to offset the UVs.
- Added a new tangent basis calculator to improve the quality and compatibility across applications ( thx to Morten S. Mikkelsen ).
- Added support for Maya / 3dsmax 2012.
- Added more options to the triangulator ( shortest diagonal, UVs, etc... ).
...
- Added a new tangent basis calculator to improve the quality and compatibility across applications ( thx to Morten S. Mikkelsen )...
This sounds awesome!
So after I tried to RTFM and poked around for some settings in xNormal itself (I found it in the plugin manager but "Configure" is grayed out), how do we actually make use this feature?
Say for instance I want to render a normal map using xNormal that will sync up with UDK's tangent basis. How would I do this? Will it just work automagically?
Maybe my understanding of what this is and does it wrong? :poly136:
I guess I should just test it and see what happens...
So after I tried to RTFM and poked around for some settings in xNormal itself (I found it in the plugin manager but "Configure" is grayed out), how do we actually make use this feature?
Say for instance I want to render a normal map using xNormal that will sync up with UDK's tangent basis. How would I do this? Will it just work automagically?
Morten is contacting Autodesk, Epic and other people to standarize the tangent basis. Meanwhile, you can use the xNormal SDK to perfectly-match your tangent basis ( you need to write a C++ plug-in as Naughty Dog or 8MonkeyLabs did ).
If you're a Maya user, the SBM exporter allows to use the Maya's tangent basis directly in xNormal. Maya's API exposes that functionality ( unlike 3dsmax ) so I just used it.
If you're a 3D engine writter you can also use the xNormal SDK to import directly my tangent basis into your engine.
today i toyed around with the optix renderer as part of exploring sculpting-repoto-bake workflows.
workes very nicly, but ive run into a problem. the occlusion i got out of optix has some problems in comparison to the standard bucket rendering occlusion.
i made a picture to illustrate the issue
am i missing some settings or is it due to the experimental state of the optix implemetation?
i keep getting this message when i try and bake out my meshes.
"attemped to read or write protected memory. this is often an indication that other memory is corrupt"
is this an issue with my comp not having enough memory? because both .obj's work fine when i throw them into maya, and i have definitely baked much higher res meshes on this computer before.
i keep getting this message when i try and bake out my meshes.
"attemped to read or write protected memory. this is often an indication that other memory is corrupt"
is this an issue with my comp not having enough memory? because both .obj's work fine when i throw them into maya, and i have definitely baked much higher res meshes on this computer before.
and also, when i try to got to the 3d viewer with it to see what's going on, it gives me this
"invalid function parameter 'msh': the mesh default has an UV face (5383) pointing to a non-existent vertex 602224."
Does that point to a problem with the low res mesh? or did zbrush screw up my high res?
Okay so I've been getting this after a couple of renders on and off what are these and does this mean the rays are missing?
Thanks.:poly117:
Nevermind, newest xnormal and redoing cage fixed problem.
Most recent update to XNORMAL has a black axis instead of the nice colorful version I am used to maybe this was an error in packaging?
today i toyed around with the optix renderer as part of exploring sculpting-repoto-bake workflows.
workes very nicly, but ive run into a problem. the occlusion i got out of optix has some problems in comparison to the standard bucket rendering occlusion.
Yep, it's very experimental yet. Also, GPUs have some problems like the reduced numerical precision to win some speed.
However, the problem you describe sound smore like a bug. Could you send me your mesh files for testing, pls?
"attemped to read or write protected memory. this is often an indication that other memory is corrupt"
"invalid function parameter 'msh': the mesh default has an UV face (5383) pointing to a non-existent vertex 602224."
Seems you OBJ exporter is messing your geometry. Pls, don't use the old 3dsmax's max2obj exporter ( use the Guruware's one ) and notice Blender's new OBJ exporter might be a beta version.
Try with other exporter or other mesh format ( try to use .SBM if it's available ).
Although the xNormal's SBM can export the Maya's tangent basis, that's only a 50% of the equation. The other 50% is the pixel shader's tangent transformation. I'm not sure which one uses Maya, xNormal uses this one in DX9c:
half3 n = tex2D(NormalTexSampler,uv)*2.0h - 1.0h;
const half3x3 l_mTB = { normalize(inTang), normalize(inBiNorm), normalize(inNorm) }; //inTang,etc... are goureaud-interpolated without centroid
n = mul(n,l_mTB); //pass normal from tangent to object space
n = normalize(n);
I'm neither sure about how Maya deals with cylindrical coordinaters or UV mirroring.
Can someone explain to me what the fudge is going on here?! This is driving me insane!
It's like the baker just randomly picks a triangle in the uv's and collapses one of the points in either U or V to 0. I checked my UVs, they're perfectly fine with no overlaps. I froze the history before exporting, tried several FBX versions, etc... All to no avail.
And worst of all, this occurs at random places on the UVs, it's never the same spots that get this wonky stuff...
Is this a bug in the fbx importer jogshy? I'm using xnormal 3.17.5
edit: Yeah... I reckon this is totally an FBX bug... Using OBJs for my low res targets rather then fbx yields perfect results, which is a pity; because I really prefer fbx over OBJ...
Anyway, some more testing is needed; but can anyone confirm that using FBX for low res meshes goes foobar?
Replies
Btw, I've just uploaded the final 3.17.4.
I'm having a few problems with Xnormal, currently when i download and install Xnormal the Photoshop Filters do not show up in my Filters tab in Photoshop.
Ive tried everything from downloading different versions of Xnormal to moving the filters from the download to the filters folder in photoshop but nothing works.
If anyone could help with this issue i would greatly appreciate it
They show up fine in my 64bit.
Copy the .8bf files from [xN install dir]\x64\photoshop_plugins to [PS install dir]\Adobe Photoshop CS5(64 bits)\Plug-Ins\Filters
it seems that the plugin is still not installing with your technique
The xN's installer should setup all automatically, unless your PS or Windows SxS .NET config is broken. To see is .NET is correctly installed just try to execute the xN's main app. If it runs correctly then it's ok. PS might need a reinstall perhaps. Btw, which PS and OS version are you using, pls?
Can you post a screenshot showing where you copied from/to the 8bf files, pls?
Don't mix x86 and x64 versions !
Thanks for the high and low profile buttons on the max exporter, big click savings
I'm on Windows 7 64bit. This is using either the x86 version or the 64bit version, and in administrator mode.
Can you render the examples without problems?
Does it happen when the CPU is at full workload? Run an OCCT/linX test to see if your computer is really stable.
Sorry for messing you about jogshy.
Do I need to set the padding higher than 1 pixel? (I just realized I haven't tried this).
I'm sure someone has asked this before... but I didn't find anything asking about it.
Its best to:
A. export a cage mesh from your app of choice
B. set up a cage in the 3d viewer
The reason would be to save a lot of time when you are adjusting a cage and have to rebake all the time to see the changes.
From a technical point of view....dunno if it's possible but well I had to ask
Maybe make it less cryptic: There seems to be verts without UV's.
I did a hipoly sculpt in sculptris and used xnormal successfully (the 1024 file attached) but later decided it needed more work.
I resculpted from scratch, and the mapping went all to hell. I did remap the UVs but that didn't help either. (512 map attached) Here are some examples of the snafu. Any insight?
Santiago, thanks so much for coming up with this AND making it free.
If your LP mesh is the subdiv0 of the HP you can use the MatchUV feature, so you won't need to setup distances/cages.
I didn't work off a straight subdivision, but I was wondering what the matchUV was for... I guess that would give the cleanest results possible?
what's your object scale and origin? Remember to perform a ResetXForm in 3dsmax and Freeze transformations in Maya before exporting your model.
What xN's graphics driver are you using? DX9, 10, OGL, ?
Can you load and view any of the xN's examples correctly?
I have little probem with normal maps generated with Xnormal, I have high poly and low poly model, I exported parts of model to obj and put it to Xnormal, I generated maps and I go back to 3DS MAX 2012 and hier is problem, how can you see on enclosed pictures, I have problem with normal map in render, I try it in UDK, in Static mesh viewer is lock good, but in game its look bad. Low poly have one smoothgroup, also high poly.
Thanks for your help...
By rengar at 2011-04-25
I have little problem with mormal map generated with Xnormal.
I have high and low poly model, expoted to xnormal in OBJ, low poly and high poly have one smoothgroup. Normal maps in 3ds max 2012 in viewport and in render window dont look good. I try it in UDK. You can see on enclosed pictures. I don't know where is problem, what I'm doing wrong.
Thanks for your help.
By rengar at 2011-04-25
The center scene button fixed it thanks, I dont think I moved anything to start with though.
This is especially weird because just yesterday I was able to quickly get a calculation for several high poly meshes being projected onto one low poly mesh.
edit: nevermind. Remember to check your relationship editors to make sure whatever you're putting in xNormal is separate from anything else in the hierarchy....at least if you have a shitty computer with no RAM.
http://eat3d.com/free/xnormal_overview
@MattQ86 The ray distance calculater takes some time to get values. You must stop it manually.
So a usual workflow for me would be to make a mesh that is setup for tiling, bring that into zbrush, sculpt it, and when I am done, export an obj.
I would take this obj into maya and duplicate neccesary pieces to make it tile, and then export another obj which i would use in xNormal for baking purposes.
This was great until I started polypainting the pieces in zbrush, and realized that the polypaint data was lost when using maya as a go between. My solution was to offset the subtools in zbrush, but this is cumbersome and not the most precise process.
So I though it would be a cool feature if xNormal could offset a bunch of instances of the high poly mesh by a user defined value to allow for easier tiling.
I'm no programmer and obviously haven't thought of the feasability or practicality of this, but thought it maybe useful.
Hopefully this image illustrates what I am asking better.
I thing this is a great idea
+1 from me
A workaround is to export your polypainted meshes from ZBrush into a OBJ. Load them into xNormal and save the mesh as SBM. Then use the SBM importer for Maya so the vertex colors can be load into Maya.
- Recompiled using Optix 2.1 ( == faster rendering speed and faster data precomputing )
- Added an option to offset the UVs.
- Added a new tangent basis calculator to improve the quality and compatibility across applications ( thx to Morten S. Mikkelsen ).
- Added support for Maya / 3dsmax 2012.
- Added more options to the triangulator ( shortest diagonal, UVs, etc... ).
This sounds awesome!
So after I tried to RTFM and poked around for some settings in xNormal itself (I found it in the plugin manager but "Configure" is grayed out), how do we actually make use this feature?
Say for instance I want to render a normal map using xNormal that will sync up with UDK's tangent basis. How would I do this? Will it just work automagically?
Maybe my understanding of what this is and does it wrong? :poly136:
I guess I should just test it and see what happens...
http://wiki.blender.org/index.php/Dev:Shading/Tangent_Space_Normal_Maps
Morten is contacting Autodesk, Epic and other people to standarize the tangent basis. Meanwhile, you can use the xNormal SDK to perfectly-match your tangent basis ( you need to write a C++ plug-in as Naughty Dog or 8MonkeyLabs did ).
If you're a Maya user, the SBM exporter allows to use the Maya's tangent basis directly in xNormal. Maya's API exposes that functionality ( unlike 3dsmax ) so I just used it.
If you're a 3D engine writter you can also use the xNormal SDK to import directly my tangent basis into your engine.
Eek!!!! I'm more than a few months away from learning C++...
Thanks for the info anyway, that clears things up
Hopefully we'll see a standard tangent basis in the near future.
You and Morten are doing a great job, keep it up!
Loosing the polypaint is a huge obstacle in the workflow I would LIKE to use.
Which brings me to a xnormal baking problem I am having.
There should be 4 pics on this link that show details of my xnormal problem.
http://img857.imageshack.us/g/xnormalproblemhp.jpg/
EDIT: It looks like my bake is "looking" at the back faces of my High poly instead of the front?
today i toyed around with the optix renderer as part of exploring sculpting-repoto-bake workflows.
workes very nicly, but ive run into a problem. the occlusion i got out of optix has some problems in comparison to the standard bucket rendering occlusion.
i made a picture to illustrate the issue
am i missing some settings or is it due to the experimental state of the optix implemetation?
anyways - thanks for the awsome app
"attemped to read or write protected memory. this is often an indication that other memory is corrupt"
is this an issue with my comp not having enough memory? because both .obj's work fine when i throw them into maya, and i have definitely baked much higher res meshes on this computer before.
and also, when i try to got to the 3d viewer with it to see what's going on, it gives me this
"invalid function parameter 'msh': the mesh default has an UV face (5383) pointing to a non-existent vertex 602224."
Does that point to a problem with the low res mesh? or did zbrush screw up my high res?
I'm curious to know if this is possible.
Thanks.:poly117:
Nevermind, newest xnormal and redoing cage fixed problem.
Most recent update to XNORMAL has a black axis instead of the nice colorful version I am used to maybe this was an error in packaging?
Letting you know joshy, thanks.:thumbup:
However, the problem you describe sound smore like a bug. Could you send me your mesh files for testing, pls?
Try with other exporter or other mesh format ( try to use .SBM if it's available ).
half3 n = tex2D(NormalTexSampler,uv)*2.0h - 1.0h; const half3x3 l_mTB = { normalize(inTang), normalize(inBiNorm), normalize(inNorm) }; //inTang,etc... are goureaud-interpolated without centroid n = mul(n,l_mTB); //pass normal from tangent to object space n = normalize(n);I'm neither sure about how Maya deals with cylindrical coordinaters or UV mirroring.
It's like the baker just randomly picks a triangle in the uv's and collapses one of the points in either U or V to 0. I checked my UVs, they're perfectly fine with no overlaps. I froze the history before exporting, tried several FBX versions, etc... All to no avail.
And worst of all, this occurs at random places on the UVs, it's never the same spots that get this wonky stuff...
Is this a bug in the fbx importer jogshy? I'm using xnormal 3.17.5
edit: Yeah... I reckon this is totally an FBX bug... Using OBJs for my low res targets rather then fbx yields perfect results, which is a pity; because I really prefer fbx over OBJ...
Anyway, some more testing is needed; but can anyone confirm that using FBX for low res meshes goes foobar?