material problem issue

[patro]

hi all,
I'm 3dsmax user and I have a problem with material. the more the material become shiny the more the issue is to see!
more than two months a go I open a task in mantis... but the issue is still unassigned.
as you can notice from the attached image I'm using metal with gold presets!

the issue is visible with all material.... with a mirror material is more visible.

in the past, we had this issue with one of the first LuxMax exporter version done by perpixel!
he gave me a solution that at moment fix the issue!

this was the PM exchange between me and him:

hi,
I have a strange problem, when I apply the Lrmirror to a geometry like chamferbox, box sphere all work well, but if I rotate completely the scene like "z"+/-90°, the lrmirror become totally black. this strange problem does not affect the sphere. I try and retry but nothing seems to fix this problem.
.............................
thx

Thats a problem with your system languages. Some languages use , instead of . for floating point numbers. Can you tell me the language of your Windows an the currency and time format you use. Changing to English(US) format for time should solve the problem.

well I just change from Italian to English(US)... but this does not fix the issue now!
any way..... could the issue related to something similar?

thank you in advance for any tip and though

Edit: sorry I forget to mention to have try with Luxrender 0.7 version... also with all weekly testing builds... the latest one the obj are completely black!
the problem have nothing to do with normals and or flipped faces.

[binarycortex]

I have experienced the same issue, and have bug #833 open in mantis. http://www.luxrender.net/mantis/view.php?id=833

[patro]

my issue is:
http://www.luxrender.net/mantis/view.php?id=816
here the same issue problem with matte material!

[patro]

I forget to mention that Luxrender log report repetitively:

[2010-09-09 19:40:21 Warning: 0] Parameter 'relsize' not used
[2010-09-09 19:40:23 Debug: 0] Compiled scene size: 265KBytes
[2010-09-09 19:40:23 Warning: 0] Parameter 'pixelsamples' not used
[2010-09-09 19:40:23 Warning: 0] Parameter 'reject_warmup ' not used
[2010-09-09 19:40:23 Warning: 0] Parameter 'initsamples' not used
[2010-09-09 19:40:23 Warning: 0] Parameter 'micromutationprob' not used
[2010-09-09 19:40:23 Debug: 0] Mesh: accel = none (global), triangles = 32 wald, quads = 0 quadrilateral
[2010-09-09 19:40:23 Debug: 0] Mesh: accel = none (global), triangles = 1772 wald, quads = 0 quadrilateral
[2010-09-09 19:40:23 Debug: 0] Mesh: accel = none (global), triangles = 1772 wald, quads = 0 quadrilateral
[2010-09-09 19:40:23 Warning: 43] Inconsistent shading normals
[2010-09-09 19:40:23 Warning: 43] Inconsistent shading normals
[2010-09-09 19:40:23 Warning: 43] Inconsistent shading normals
[2010-09-09 19:40:23 Warning: 43] Inconsistent shading normals
[2010-09-09 19:40:23 Warning: 43] Inconsistent shading normals
[2010-09-09 19:40:23 Warning: 43] Inconsistent shading normals
[2010-09-09 19:40:23 Warning: 43] Inconsistent shading normals
[2010-09-09 19:40:23 Warning: 43] Inconsistent shading normals
[2010-09-09 19:40:23 Warning: 43] Inconsistent shading normals
[2010-09-09 19:40:23 Warning: 43] Inconsistent shading normals
[2010-09-09 19:40:23 Warning: 43] Inconsistent shading normals
etc................................................
the problem for luxrender seams to be the normals... but I try by flipping normals and faces without any result!

[subpixel]

I'm not sure if it helps, but it sure helps with locale decimal separator issue.

Code: Select all

setlocale(LC_NUMERIC,"english");

It's for C++ code, but I'm sure equivalents should be found in all scripting languages. Anyway, I use it in IndigoMax source and it works. If someone knows a better solution, please share.

[Lord Crc]

the problem for luxrender seams to be the normals... but I try by flipping normals and faces without any result!

You need to identify the faces where one vertex normal points "out" and the other two points "in", or vice versa. This often happens if one triangle points "in" while other triangles point "out". Then you'll get the warning for all the triangles which share a vertex or two with the "in"-triangle.

[patro]

You need to identify the faces where one vertex normal points "out" and the other two points "in", or vice versa. This often happens if one triangle points "in" while other triangles point "out". Then you'll get the warning for all the triangles which share a vertex or two with the "in"-triangle.

hi Lord thank you for suggestion... but as I just say the problem isn't related to normals in max.
inward oriented vertex are displayed in max in green.
the attached image show the "edit normal" modifier applied to the chamfer-cylinder, as you can notice all vertex are showing out.

this mean that the problem is related to something else.
also consider that I'm using only native 3ds max geometry and they have obviously no error.

[Lord Crc]

Perhaps there's an issue with the exported geometry? Could you try narrowing down the geometry by deleting triangles until the error goes away?

[patro]

I try more than once to increase smoothing, faceting, grouping, joining, scaling...
the one that worked for just one time was strangely joining! the log didn't show any error for normals... and all scene obj were nicely rendered with gold.
also..if apply a modifier like smooth the error narrow. but this don't resolve the problem... and a sphere for example will rendered faceted.

I have an idea about the error... but it's impossible to be right.
anyway.... it's like that the normals of the obj are exported pointing in direction to the view in which the obj are created!
so if for example I create the obj in front view... and I don' modify it.... than I apply a camera to the view(don't orbit), the obj are rendered nicely... but if I orbit the camera, the normal don't point no more perpendicular to camera(original creation view)... so that I see the black part.
I know that my idea is very strange.... but I try more than once before to trying to share it.

probably it's in a way related to the exporter... to me it would be nice that all LRmaterial have a double sided flag able function. so to have not to bother with strange normals problem.

[patro]

the image show objs created in top view... they were not modified(rotation etc.), as you can see they are nicely rendered.
only the chamferbox show black area, just because they are more complex from a chamfercylinder or a sphere!
I made a test on the fly just to explain better my idea. in the scene objs rotated by 45°are to see black faces.
luxout front

luxout back

luxout front rot 45°

luxout back rot 45°

for the image of the first post I used the same geometry... but a play with them.