[ QUOTE ]
My vid card is fairly old, but supports pixel shader 2.0, so I thought it would be ok.
[/ QUOTE ]
The shadows are disabled if you use a pixel shader 2.0 card + DX9 ( due to instruction limit and speed ). However, if you enable the OpenGL graphics driver(in the Plugin manager, default graphics driver combo box) shadows will be enabled ( but fear the speed if you move the light )
[ QUOTE ]
xnormal just keeps telling me to install directx 9.c! Ive installed direct x 9.c as it told me to, so whats up with this? is there anything I can do?
[/ QUOTE ]
You need DX9.0c August 2007 or above edition. See if you have a file called d3dx9_35.dll in the system folder(usually c:\windows\system32 ). If not, you could have installed DX9.0c, but not the August 2007 edition. Download it from the xNormal download page or directly from
I thought your installer would update my old version but it put itself in another dir so I was still running the previous version.
[/ QUOTE ]
In theory I should add the DX9 to the installer, yep.. but I didn't want to add other 60Mbs to the package... so had to remove that and the .NET 2.0(other 50Mbs) from the install package
@! The 4.0 gonna skip any damm M$ library! OpenGL 3 + GTK+ statically-linked ftw!
[ QUOTE ]
Is it impossible to render 2048x2048 ww 4x AA? Here everything renders, then at the end before writing to disk I get a memory out error.
I've got 4gigs of ram (/3gb setting in boot.ini), 32bit winXP and a 6 gig vmem file
How come so much memory is needed to render? I can't imagine what kind of data is needed.
[/ QUOTE ]
Basically is 2048 pixel x 2048 pixel x 4(rgba channels) x 4AA x 20bytes(nx,ny,nz,h) x 2(to allow a bitmap copy for the dilation, temporary buffers, etc) = 2,6Gb just for the normal map...
The highpoly model is harder to calculate... depends on geometry and topology. Usually 1M poly mesh occupies like 150Mb.
Well, these quantities are not completely real... because all is compressed into memory... but at some point I need to decompress a great volume of data.
You mention the error is at end... have your tryed to disable the dilation filter? Are you just rendering one kind of map? This should help... disable filters and render maps individually( i mean... don't render AO map + normal map at the same time... will double the memory used... render first the normal map... then exit the program... reload program and render AO ).
I suspect what crashes is the preview window... the .NET picture boxes uses a lot of RAM... but the output bitmaps should be saved ok into disk before to crash.
Btw, i'm optimizing a lot the memory usage for the next version though.
[ QUOTE ]
hey jogshy, have you sorted that stuff about core2duo processors taking long time to render ao maps?
[/ QUOTE ]
Yep, was solved in the 3.11.1. The error was a loop asking for an incorrect random number that never reached the proper value. Now i'm using a precomputed cycled table so it won't enter into an infinite loop.
Also gonna change the progress bar to be updated better for the next version... it currently jumps too brutally... perhaps gonna add an "estimated finish time" too.
To be sure it works load the smiley example and render an AO map with the default settings. It takes like 6s on a decent computer(reference: OCd E2140,superPI 1M 19s). If takes more or halts(and you have an AMD) then try to install the WinXP dual core patches(one from M$, other from AMD)... and if you have a Core Duo sure you install the famous microcode patch via Windows Update.
However, AO comptutation, by nature, can be very slow for dense(1M+) highpoly meshes.
Some advices to improve speed:
- Sure you use cages to limit the ray distance(is faster than the uniform distance).
- Adaptative AO can help the speed too(to set min and max rays to a dirrent value)... but can make serious artifacts if you set the min rays too low.
- Disable antialiasing because multiplies linearly the time used to render.
- If you have some computers there try to use the hxGrid distributed renderer.
- Always do test with 16/16 rays, 512x512 and absolutely no AA. Once you have the values well set render the final version and go sleep/lunch
The new version I'm preparing gonna be a bit faster computing all the maps. You could also use the GPU to render the AO faster.
I loaded up the smiley example, it reset some settings and rendered ao fine
I set up my old high and low models again and this time around they rendered fine.
So I guess there was some setting somewhere that the program didn't like. Anyway, just so you know.
[/ QUOTE ]
Well, feel free to save and send me your XML settings if you notice it freezes again. Perhaps some parameter is saved incorrectly and tells the renderer to use 10000M of rays per pixel or something strange.
thx for the comments!
jogshy , another thing, that i might sound like a total noob, but i run ( fortunately i had someone to help me with it ) with a somesort of problem in the matter of cages, i was getting incorrect values of the normal map in one areas, so i saved the cage i was using on 3ds max to render correct , but it whined about the number of vertex not beeing the same , wich was odd, since i imported it back to max and made a vert count and was exactly the same , so i guess something must have gone weird
Also getting some memory problems, but its understandable since i have only 1 gig of memory ehehe....
[ QUOTE ]
i saved the cage i was using on 3ds max to render correct , but it whined about the number of vertex not beeing the same , wich was odd, since i imported it back to max and made a vert count and was exactly the same
[/ QUOTE ]
Are you using the 3.11.1? Solved a problem related to that in the last version.. that's why I ask you this first.
In affirmative case...could you send me your cage and lowpoly meshes (as a .max is possible) for debugging the problem pls?
xnormal does seem to have rendered out the AO as seen in the small images when it is done, but the occlusion file is just the AO Bent-N one
[/ QUOTE ]
Bent normals are saved into RGB, occlusion into Alpha channel. If you choose a file type without the proper alpha channel support you won't get the ambient occlusion...
BMP, JPG -> No official alpha support ( can contain it but is non-standard )
I'm getting the following error when i try to view an approx. 16k triangle model in the 3d viewer in 3.11.1.7199 (64-bit). Also receive the same error when i try to render the mesh.
I've been able to render by cutting the model into small sections. But haven't been able to view
"invalid function parameter 'msh->UVs.size()' : SOrry cannot calculate tangent basis for meshes with more thean 32k UVs"
any ideas what i can do to display the model short of cutting it into little pieces?
also, when in windowed mode, i can't seem to resize the window, anyone else run into this and found a solution.
"invalid function parameter 'msh->UVs.size()' : SOrry cannot calculate tangent basis for meshes with more thean 32k UVs"
[/ QUOTE ]
16k triangles mesh and 32k UVs looks like the exporter you are using is messing some data. Have you tryed other format? Are you sure you welded well the UVs in your model? Is your UV map composed of very small sepparated tiles(zbrush AVTiles perhaps?)? Usually the UVs should be like 8k for a 16k tris model with joint triangles. With 16k tris and 32k UVs means each triangle is not sharing almost 2 vertices with all the neighbors.
Well... the 3d viewer cannot manage more than 32k verts or faces for lowpoly meshes(because to calculate the tangent basis + deindex is too slow).
If you want I can send you a modified version with 64k limit to test... but 16k tris / 32k UVs indicates the exported data has been corrupted in any way. I had problems with Modo exporter in the past in case you exported the mehs from it.
If you wanna send me your lowres model I could debug it to see what's happening.
[ QUOTE ]
any ideas what i can do to display the model short of cutting it into little pieces?
[/ QUOTE ]
Well, if your mesh has individual objects like a hat, sword, cloth, etc... I recommend you to detach them from the main body. If you detach some faces those belong to the same object then the shading will be affected.
Other possibility, in case your normals are not averaged, is to detach the faces with the same smoothing group.
[ QUOTE ]
also, when in windowed mode, i can't seem to resize the window, anyone else run into this and found a solution.
[/ QUOTE ]
Currently, the Direct3D window cannot be resized in windowed mode.
If you change to the OpenGL graphics driver in the Plugin Manager you will be able to resize it(but the Direct3D one is usually faster, has extended shadow modes and usually Forceware/Catalyst drivers have less bugs in D3D).
there is something wrong with the amboc, i have a model with like 2 to 3 mil max , i have 1 gig of ram DDR2 , dual core processor , the normal map renders cool in 2048x2048 with 2x antialisaing, the amboc wont even render with 512x512 and 150 samples min and max, the program simply gets weird, i cant abort , nor close the program only ctrl + alt + del will do , the progress bars is stuck in a part and it will keep there, the program apparently is still running...
[/ QUOTE ]
That's a lot . Have you tryed with 20 samples per pixel and without AA?
About the abort thing I am solving it as somebody mentioned before and adjusting better the progress bar increments.
Te 3.12.0 will bring two new tools:
Cone map tool for ultra-realistic parallax mapping and the HW-accelerated AO computation. Also a realtime raytracing graphics driver for the 3d viewer and more!
probably the Beta 1 will appear by the end of the next month.. i'm introducing massive code changes.
i couldn't find SBM exporter plugin for Max 9 64 bit ?
looks like the 64-bit installer have the max plugin disabled,
do you know where i can download the SBM exporter (for max 9 64bit) ?
Is the sbm exporter supposed to work with Max 6? At the moment I get a dll error when I start Max. I've looked around for a solution to this problem, but haven't found it.
The .EXE xNormal installer autodetects 3dsmax 64 bits ( looking a file called max64.task in the 3dsmax folder ). It also autodetects if you have max 6/7/8/9 installed looking at the Windows's registry. The option to install the 3dsmax plugins won't be enabled unless it autodetects the presence of max 6/7/8 or 9 in your hard disk.
Perhaps the autodetect mechanism has some kind of bug...
On the other hand, if you download the xNormal 7zipped package(instead of the .Exe installer) you will find the SBM plugins into the x64/3dsmax_plugins folder. Copy the xNormal_3dsmax9_sbm_mesh_exporter_x64.dle manually to your stdplug or plugins 3dsmax's folders and will work ok.
About the 3dsmax 6 compatibility i'm not sure. I just own a 3dsmax 8 and 9 so only can guarantee the SBM mesh exporter works there. In theory, I compiled the plugins with 3dsmax 6 and 7 compatiblity, but i'm not sure if would work ok because I cannot test/debug them... and would be only useful for max 7 or above because the projection modifier does not exist in max 6.
Thanks for the response. Looks like I'll be exporting .ase files or editing my cage in xnormal until my poor company can upgrade me.
Thanks for providing such a useful tool for the community. Keep up the good work.
Maybe a crazy suggestion, but any plans for shaders in the viewer? It would be nice to be able to view a model with a .fx created in shaderFx etc. I could see how this might be beyond the purpose of your viewer, but I just thought I'd ask.
theres another issue, that might be my fault but doesnt hurt asking ahah , whenever i drag a .tga created in xnormal to photoshop the map is like 10% only of opacity , and whatever alpha i cant view it, it views fine in other programs , but the inability to acess alpha is weird
Ok I'm using xnormal and getting some odd results that seem pretty standard, but have no idea what I am doing wrong. Is it the LP model, hi poly model uv layout? Can anyone explain some theories of what is going on on here please?
i cant quite figure out whats in that image, maybe diusplaying the highpoly next to the lowpoly ? also dont forget that lowpoly must have a single smoothing group.
right, here is a better shot. I am just checking "smooth normals" is this not a desirable way of doing things? I'm sure it has something to do with it, but have seen countless examples of the same stuff that do work.
Johny, i've dropped the parallax wall normal map TGA into Photoshop CS2/CS3 and also in the DX9 texture viewer and seems it works ok... Perhaps is something related to this known Photoshop bug? http://nicolaescher.com/journal/article/20/photoshop-7-tga-plug-in-alpha-channel-bug
Any other having this problem? Perhaps you copyed the ps7 tga plugin to your CS1?
About the AO file you sent me, i'm investigating it.
Ape, have you tryed to separate the conflictive part from the lowpoly/highpoly model? Sometimes that works.
Ok here's more of en explination(I need to get better at posting these things, lol)
Anyway, basically its geometry that looks smooth but coming out oddly, the bevels at the front are on 90 degree surfaces, which I guess can be tricky, but the edges seem like there is too much tension or something causing those faded effects on the side and bottoms... this is warping out that flat surface.
Further at the top, I realize I have triangles in my Low poly mesh, but figured they would be ignored and not screen up that smooth corner on there.
Check out the image to see what I mean after I wrote over it...
Have you tried rendering/shading it yet? Most of these "weird" gradients aren't really a problem, the tangents actually need them to give you the proper shading.
Just with my experiance in the engine I'm using, especially when this is going to be right up next to the camera they are going to be issues, but maybe I'll continue for now and work it out a little later.
Are you importing the mesh from Xnormal, or just the texture? You'll need the tangents + bitangents (stored in the verts by Xnormal when it creates the normalmap) to counteract the gradients. In fact you'll need them regardless, to remove shading seams along uv borders.
Export the hipoly mesh and lowpoly mesh from maya. Sorry how how do you export tangent info? Isn't it just stored in the mesh? But not that experianced with editing that in maya.
I've got an xNormal question. Is there a way to use the Height Map to Normal Map Tool in xNormal to batch a large directory of height maps?
I know that xNormal has the ability to do batching and from the help doc it looks like you run xNormal from a command line which points to several .xml files, but I'm not sure exactly the format of the .xml files.
Note that my first thought was to use the height2Normal photoshop filter in an action, but I'm asking this for an engineer I know who doesn't have Photoshop.
Alright, more tests are proving to be confusing... anyone want to tackle why the crappy maya 7 surface sampler is rendering out a smoother normal map then Xnormal?
[ QUOTE ]
Is there a way to use the Height Map to Normal Map Tool in xNormal to batch a large directory of height maps?
[/ QUOTE ]
Not currently, sorry.
Perhaps in a future, but will be hard to previsualize the results... and without previsualization will be hard to set well the parameters.
[ QUOTE ]
... several .xml files, but I'm not sure exactly the format of the .xml files.
[/ QUOTE ]
You can find the XSD schema file on the SDK. A programmer could use it to auto-generate a file for the command line execution.
biggest problem i'm getting in xnormal, is that it's either not bringing the meshes in properly, or they are TINY! doesn't matter where i export them from, they are either so small i can't see them in the viewer, or aren't there at all. and this is confirmed by any attempt to render out a normalmap giving me nothing but blank textures...
hey man, just was wondering if anything changed on the photoshop dialation filter. It use to work great, now for some reason when i try to run the filter it flips the mask vertically and the end result looks really messed up.
[ QUOTE ]
hey man, just was wondering if anything changed on the photoshop dialation filter.
[/ QUOTE ]
Yep, changed it a lot for the 3.11.1
[ QUOTE ]
It use to work great, now for some reason when i try to run the filter it flips the mask vertically and the end result looks really messed up.
[/ QUOTE ]
You are right, it flips the mask accidentally.
Is solved for the next release.
I think the 3.11.0 works ok but uses the old system that requires to select the alpha mask manually, i'm not sure.
If you need it urgently I can send you by email the corrected plugin.
The 3.12.0 beta 1 is almost ready with tons of new things.
I hope could release it this month.
wtf, Microsoft now forces the DX9 runtime to be Geniune-validated? Dammmm.... perhaps I should include it in the .EXE installer for the evil people but + 60Mb sux
- GPU ambient occlusion generator ( 100x faster than typical way to render AO )
- Ogre3D mesh importer/exporters
- Realtime ray tracing graphics driver for the 3D viewer + a new example showing RT reflections.
- PRT-p and PRT-n map rendering
- Video capture inside the 3D viewer
- Radeon HD 2XXX OpenGL problems solved
- Photoshop dilation filter fix
- 3dsmax 64bits detection in the EXE installer fixed
- 10% map rendering speed improvement
- OpenEXR B44A compression support
- PLY mesh importer improved compatibility
- New cavity map EMBoss method
- Improved rendering abortion response time
- Minor bugs fixed and small improvements.
For the final version I want to add too the "cone map" tool ( improved parallax mapping generator ) and perhaps a DX10 graphics driver with real HDR/fs glow/depth of field effects.
Stay tuned!
ps: Finally embedded DX9 into the .EXE installer.. but the good compression made it to occupy only +15Mb.
btw, some important details of the new GPU AO tool:
1. Requires a Shader Model 2 card ( Radeon 9XXX, GeForce FX or above ) with decent OpenGL/GLSL/FBO support.
2. The blending "precision" option is critical... Use FP32 for GF8 or Radeon HD 2XXX. Use FP16 for GF6/7 or Radeon 1XXX.
Use CPU for Radeon X800/600/500/400/9XXX or GeForceFX.
Older cards like GeForce 3/4, Radeon 8XXX, rivaTNT/voodoo/savage are not supported so you need to render the AO as before.
If you set FP32/FP16 on a graphics card that does not support it then will enter into a super-slow software emulation mode(is an OpenGL thing... and there is no way to autodetect it well )
3. The meshes need to be UV-mapped.
4. That tool cannot apply distance attenuation, cannot calculate bent normals and neither to project the highpoly into the lowpoly mesh. It just accepts one mesh. If you need to perform that bake the AO with the GPU tool and then use the "bake base texture feature".
5. Can manage 1M polys objects well. Past that number the mesh can or cannot be rendered fast depending on your graphics card and video memory.
I've been using the 3.12.0 today, and I noticed that if I open the model viewer in windowed mode, just viewing a lowpoly mesh and normalmap, and then ALT-TAB out to another window, after a while Windows gave me a Low Virtual Memory warning.
Then the windowed viewer crashed and dropped me back to the main xNormal screen. I think there's a memory leak there somewhere.
Replies
My vid card is fairly old, but supports pixel shader 2.0, so I thought it would be ok.
[/ QUOTE ]
The shadows are disabled if you use a pixel shader 2.0 card + DX9 ( due to instruction limit and speed ). However, if you enable the OpenGL graphics driver(in the Plugin manager, default graphics driver combo box) shadows will be enabled ( but fear the speed if you move the light )
[ QUOTE ]
xnormal just keeps telling me to install directx 9.c! Ive installed direct x 9.c as it told me to, so whats up with this? is there anything I can do?
[/ QUOTE ]
You need DX9.0c August 2007 or above edition. See if you have a file called d3dx9_35.dll in the system folder(usually c:\windows\system32 ). If not, you could have installed DX9.0c, but not the August 2007 edition. Download it from the xNormal download page or directly from
http://www.microsoft.com/downloads/detai...;DisplayLang=en
and sure you have Administrator priviledges when installing it.
EDIT never mind, found it
I thought your installer would update my old version but it put itself in another dir so I was still running the previous version.
[/ QUOTE ]
In theory I should add the DX9 to the installer, yep.. but I didn't want to add other 60Mbs to the package... so had to remove that and the .NET 2.0(other 50Mbs) from the install package
@! The 4.0 gonna skip any damm M$ library! OpenGL 3 + GTK+ statically-linked ftw!
[ QUOTE ]
Is it impossible to render 2048x2048 ww 4x AA? Here everything renders, then at the end before writing to disk I get a memory out error.
I've got 4gigs of ram (/3gb setting in boot.ini), 32bit winXP and a 6 gig vmem file
How come so much memory is needed to render? I can't imagine what kind of data is needed.
[/ QUOTE ]
Basically is 2048 pixel x 2048 pixel x 4(rgba channels) x 4AA x 20bytes(nx,ny,nz,h) x 2(to allow a bitmap copy for the dilation, temporary buffers, etc) = 2,6Gb just for the normal map...
The highpoly model is harder to calculate... depends on geometry and topology. Usually 1M poly mesh occupies like 150Mb.
Well, these quantities are not completely real... because all is compressed into memory... but at some point I need to decompress a great volume of data.
You mention the error is at end... have your tryed to disable the dilation filter? Are you just rendering one kind of map? This should help... disable filters and render maps individually( i mean... don't render AO map + normal map at the same time... will double the memory used... render first the normal map... then exit the program... reload program and render AO ).
I suspect what crashes is the preview window... the .NET picture boxes uses a lot of RAM... but the output bitmaps should be saved ok into disk before to crash.
Btw, i'm optimizing a lot the memory usage for the next version though.
[ QUOTE ]
hey jogshy, have you sorted that stuff about core2duo processors taking long time to render ao maps?
[/ QUOTE ]
Yep, was solved in the 3.11.1. The error was a loop asking for an incorrect random number that never reached the proper value. Now i'm using a precomputed cycled table so it won't enter into an infinite loop.
Also gonna change the progress bar to be updated better for the next version... it currently jumps too brutally... perhaps gonna add an "estimated finish time" too.
To be sure it works load the smiley example and render an AO map with the default settings. It takes like 6s on a decent computer(reference: OCd E2140,superPI 1M 19s). If takes more or halts(and you have an AMD) then try to install the WinXP dual core patches(one from M$, other from AMD)... and if you have a Core Duo sure you install the famous microcode patch via Windows Update.
However, AO comptutation, by nature, can be very slow for dense(1M+) highpoly meshes.
Some advices to improve speed:
- Sure you use cages to limit the ray distance(is faster than the uniform distance).
- Adaptative AO can help the speed too(to set min and max rays to a dirrent value)... but can make serious artifacts if you set the min rays too low.
- Disable antialiasing because multiplies linearly the time used to render.
- If you have some computers there try to use the hxGrid distributed renderer.
- Always do test with 16/16 rays, 512x512 and absolutely no AA. Once you have the values well set render the final version and go sleep/lunch
The new version I'm preparing gonna be a bit faster computing all the maps. You could also use the GPU to render the AO faster.
I loaded up the smiley example, it reset some settings and rendered ao fine
I set up my old high and low models again and this time around they rendered fine.
So I guess there was some setting somewhere that the program didn't like. Anyway, just so you know.
[/ QUOTE ]
Well, feel free to save and send me your XML settings if you notice it freezes again. Perhaps some parameter is saved incorrectly and tells the renderer to use 10000M of rays per pixel or something strange.
thx for the comments!
Also getting some memory problems, but its understandable since i have only 1 gig of memory ehehe....
thanks for the awesome program !
i saved the cage i was using on 3ds max to render correct , but it whined about the number of vertex not beeing the same , wich was odd, since i imported it back to max and made a vert count and was exactly the same
[/ QUOTE ]
Are you using the 3.11.1? Solved a problem related to that in the last version.. that's why I ask you this first.
In affirmative case...could you send me your cage and lowpoly meshes (as a .max is possible) for debugging the problem pls?
xnormal does seem to have rendered out the AO as seen in the small images when it is done, but the occlusion file is just the AO Bent-N one
also, is it in any way possible to have the normals on the highpoly kept as hard and not smoothed?
edit: oh nm about the occlusion, I found it in the alpha channel :x
Oh, and thanks for a truly awesome application, it imports and handles huge meshes with ease :x
xnormal does seem to have rendered out the AO as seen in the small images when it is done, but the occlusion file is just the AO Bent-N one
[/ QUOTE ]
Bent normals are saved into RGB, occlusion into Alpha channel. If you choose a file type without the proper alpha channel support you won't get the ambient occlusion...
BMP, JPG -> No official alpha support ( can contain it but is non-standard )
PNG -> Problematic alpha ( 1 bit, 8 bits premultiplied masked alpha )
TGA, DDS, TIF, OpenEXR -> Good
[ QUOTE ]
also, is it in any way possible to have the normals on the highpoly kept as hard and not smoothed?
[/ QUOTE ]
The normals on the highres mesh are fully respected in the map generation process.
However, for the 3D viewer I need to smooth them to get good rendering speed and to keep the loading process fast enough.
I think mudbox exports smooth by default so it's a bit of a hassle to switch it and re-export.
I've been able to render by cutting the model into small sections. But haven't been able to view
"invalid function parameter 'msh->UVs.size()' : SOrry cannot calculate tangent basis for meshes with more thean 32k UVs"
any ideas what i can do to display the model short of cutting it into little pieces?
also, when in windowed mode, i can't seem to resize the window, anyone else run into this and found a solution.
[ QUOTE ]
16k triangle model in the 3d viewer...
"invalid function parameter 'msh->UVs.size()' : SOrry cannot calculate tangent basis for meshes with more thean 32k UVs"
[/ QUOTE ]
16k triangles mesh and 32k UVs looks like the exporter you are using is messing some data. Have you tryed other format? Are you sure you welded well the UVs in your model? Is your UV map composed of very small sepparated tiles(zbrush AVTiles perhaps?)? Usually the UVs should be like 8k for a 16k tris model with joint triangles. With 16k tris and 32k UVs means each triangle is not sharing almost 2 vertices with all the neighbors.
Well... the 3d viewer cannot manage more than 32k verts or faces for lowpoly meshes(because to calculate the tangent basis + deindex is too slow).
If you want I can send you a modified version with 64k limit to test... but 16k tris / 32k UVs indicates the exported data has been corrupted in any way. I had problems with Modo exporter in the past in case you exported the mehs from it.
If you wanna send me your lowres model I could debug it to see what's happening.
[ QUOTE ]
any ideas what i can do to display the model short of cutting it into little pieces?
[/ QUOTE ]
Well, if your mesh has individual objects like a hat, sword, cloth, etc... I recommend you to detach them from the main body. If you detach some faces those belong to the same object then the shading will be affected.
Other possibility, in case your normals are not averaged, is to detach the faces with the same smoothing group.
[ QUOTE ]
also, when in windowed mode, i can't seem to resize the window, anyone else run into this and found a solution.
[/ QUOTE ]
Currently, the Direct3D window cannot be resized in windowed mode.
If you change to the OpenGL graphics driver in the Plugin Manager you will be able to resize it(but the Direct3D one is usually faster, has extended shadow modes and usually Forceware/Catalyst drivers have less bugs in D3D).
Hope it helps.
About the cage ill see what i can do to send you
sorry for my rant
150 samples min and max
[/ QUOTE ]
That's a lot
About the abort thing I am solving it as somebody mentioned before and adjusting better the progress bar increments.
Te 3.12.0 will bring two new tools:
Cone map tool for ultra-realistic parallax mapping and the HW-accelerated AO computation. Also a realtime raytracing graphics driver for the 3d viewer and more!
probably the Beta 1 will appear by the end of the next month.. i'm introducing massive code changes.
i couldn't find SBM exporter plugin for Max 9 64 bit ?
looks like the 64-bit installer have the max plugin disabled,
do you know where i can download the SBM exporter (for max 9 64bit) ?
EDIT:
btw.. I'm using Xnormal 3.11.1
thanks
Perhaps the autodetect mechanism has some kind of bug...
On the other hand, if you download the xNormal 7zipped package(instead of the .Exe installer) you will find the SBM plugins into the x64/3dsmax_plugins folder. Copy the xNormal_3dsmax9_sbm_mesh_exporter_x64.dle manually to your stdplug or plugins 3dsmax's folders and will work ok.
About the 3dsmax 6 compatibility i'm not sure. I just own a 3dsmax 8 and 9 so only can guarantee the SBM mesh exporter works there. In theory, I compiled the plugins with 3dsmax 6 and 7 compatiblity, but i'm not sure if would work ok because I cannot test/debug them... and would be only useful for max 7 or above because the projection modifier does not exist in max 6.
even with 1 sample it wont render and will clog
[/ QUOTE ]
But min is 8... hehehe
Well, feel free to send me your meshes and XML settings, i would test on my E2140 and debug the problem.
Thanks for providing such a useful tool for the community. Keep up the good work.
Maybe a crazy suggestion, but any plans for shaders in the viewer? It would be nice to be able to view a model with a .fx created in shaderFx etc. I could see how this might be beyond the purpose of your viewer, but I just thought I'd ask.
Normal Problems in XNormal?
Example 2
http://nicolaescher.com/journal/article/20/photoshop-7-tga-plug-in-alpha-channel-bug
Any other having this problem? Perhaps you copyed the ps7 tga plugin to your CS1?
About the AO file you sent me, i'm investigating it.
Ape, have you tryed to separate the conflictive part from the lowpoly/highpoly model? Sometimes that works.
Anyway, basically its geometry that looks smooth but coming out oddly, the bevels at the front are on 90 degree surfaces, which I guess can be tricky, but the edges seem like there is too much tension or something causing those faded effects on the side and bottoms... this is warping out that flat surface.
Further at the top, I realize I have triangles in my Low poly mesh, but figured they would be ignored and not screen up that smooth corner on there.
Check out the image to see what I mean after I wrote over it...
Example Exp
I know that xNormal has the ability to do batching and from the help doc it looks like you run xNormal from a command line which points to several .xml files, but I'm not sure exactly the format of the .xml files.
Note that my first thought was to use the height2Normal photoshop filter in an action, but I'm asking this for an engineer I know who doesn't have Photoshop.
BTW, thanks for xNormal, it's an awesome tool!
Normal Oddities
Is there a way to use the Height Map to Normal Map Tool in xNormal to batch a large directory of height maps?
[/ QUOTE ]
Not currently, sorry.
Perhaps in a future, but will be hard to previsualize the results... and without previsualization will be hard to set well the parameters.
[ QUOTE ]
... several .xml files, but I'm not sure exactly the format of the .xml files.
[/ QUOTE ]
You can find the XSD schema file on the SDK. A programmer could use it to auto-generate a file for the command line execution.
biggest problem i'm getting in xnormal, is that it's either not bringing the meshes in properly, or they are TINY!
[/ QUOTE ]
Try this:
- Scale them up ( radius > 1 for the win!). Re-export it scaled or just use the scaling option on the mesh slot.
- ResetXForm+Collapse+Edit mesh triangulation in 3dsmax, then export.
Freeze transformations+Triangulate in Maya, then export.
- Disable backface culling in the 3D viewer options ( to be sure objects like cloth, walls, windows, etc... are displayed correctly ).
- Press the autocenter button inside the 3D viewer with a non-black background and with the grid enabled.
- See the radius units and # of tris in the 3dviewer on the right-bottom part of the screen. If you get 2 tris or radius too small something is bad...
- If you export from ZBrush remember to set correctly the flipX/Y options before to export as OBJ.
- Sure the mesh path and file you are specifying is REALLY the one you want ( I know, sounds stupid but that occur a lot to me
- Sure the "Show lowpoly meshes" and "Show highpoly meshes" are checked. Also sure your mesh is "Visible"(in the mesh list)
Hope it helps.
hey man, just was wondering if anything changed on the photoshop dialation filter.
[/ QUOTE ]
Yep, changed it a lot for the 3.11.1
[ QUOTE ]
It use to work great, now for some reason when i try to run the filter it flips the mask vertically and the end result looks really messed up.
[/ QUOTE ]
You are right, it flips the mask accidentally.
Is solved for the next release.
I think the 3.11.0 works ok but uses the old system that requires to select the alpha mask manually, i'm not sure.
If you need it urgently I can send you by email the corrected plugin.
The 3.12.0 beta 1 is almost ready with tons of new things.
I hope could release it this month.
The 3.12.0 Beta 1 has been released!
- GPU ambient occlusion generator ( 100x faster than typical way to render AO )
- Ogre3D mesh importer/exporters
- Realtime ray tracing graphics driver for the 3D viewer + a new example showing RT reflections.
- PRT-p and PRT-n map rendering
- Video capture inside the 3D viewer
- Radeon HD 2XXX OpenGL problems solved
- Photoshop dilation filter fix
- 3dsmax 64bits detection in the EXE installer fixed
- 10% map rendering speed improvement
- OpenEXR B44A compression support
- PLY mesh importer improved compatibility
- New cavity map EMBoss method
- Improved rendering abortion response time
- Minor bugs fixed and small improvements.
For the final version I want to add too the "cone map" tool ( improved parallax mapping generator ) and perhaps a DX10 graphics driver with real HDR/fs glow/depth of field effects.
Stay tuned!
ps: Finally embedded DX9 into the .EXE installer.. but the good compression made it to occupy only +15Mb.
btw, some important details of the new GPU AO tool:
1. Requires a Shader Model 2 card ( Radeon 9XXX, GeForce FX or above ) with decent OpenGL/GLSL/FBO support.
2. The blending "precision" option is critical... Use FP32 for GF8 or Radeon HD 2XXX. Use FP16 for GF6/7 or Radeon 1XXX.
Use CPU for Radeon X800/600/500/400/9XXX or GeForceFX.
Older cards like GeForce 3/4, Radeon 8XXX, rivaTNT/voodoo/savage are not supported so you need to render the AO as before.
If you set FP32/FP16 on a graphics card that does not support it then will enter into a super-slow software emulation mode(is an OpenGL thing... and there is no way to autodetect it well )
3. The meshes need to be UV-mapped.
4. That tool cannot apply distance attenuation, cannot calculate bent normals and neither to project the highpoly into the lowpoly mesh. It just accepts one mesh. If you need to perform that bake the AO with the GPU tool and then use the "bake base texture feature".
5. Can manage 1M polys objects well. Past that number the mesh can or cannot be rendered fast depending on your graphics card and video memory.
- GPU ambient occlusion generator ( 100x faster than typical way to render AO )
- 10% map rendering speed improvement
[/ QUOTE ]
O_O awesome!
I can't wait for the gloom effects
Then the windowed viewer crashed and dropped me back to the main xNormal screen. I think there's a memory leak there somewhere.
Apart from that it all seems good!