LuxRender v0.1 CVS build of 09-06-2008

[Radiance]

Hi all,

Built with VS2005 (microsoft compiler) for SSE2 architecture of CVS HEAD on 09-06-2008.

Download the DLL's to run it in a separate archive here: viewtopic.php?f=30&t=553

For use with CVS luxblend exporter v1.82(CVS) here (UPDATED): http://cvs.savannah.gnu.org/viewvc/*che ... root=ecume
(Please report any issues with the exporter in the luxblend development forum)

UPDATES SINCE LAST WEEKLY BUILD (last weekly build was 03-06-08)
-------------------------------

NOTE - LuxBlend has had many updates too this time, don't forget to update a new copy.

summary (too much to describe ):

- many bugfixes
- improved bidirectional integrator
- significant speedup when rendering with imagemap textures.
- more displacement mapping options / normal smoothing etc...
- support for UV mapped displacement mapped loopsubdiv meshes
- SPD interfaces working on point, spot,distant and infinite lights (infinite lights still broken when using .exr files, due to bug in texture mipmap code though)
- lots of work on new gui by ratow (not in this binary yet unfortunateley, should be in next one)

All the new stuff needs lots of testing, please help

See commits for details:

COMMITS
-------

radiance29 Tweaked normalization of L SPDs master
radiance29 ported SPD interfaces to distant light
radiance29 cleanups and LSPD destructors
radiance29 ported SPD interfaces to point and spot lights
radiance29 cleanups
radiance29 Ported SPD interface to 'infinite' area light
radiance29 made BILINEAR filtering for imagemaps default
ratow Added OpenGL output viwer and small fix to thread control.
dade916 Added the new imagemap option 'string filtertype' and ...
ratow Fixed a wxluxgui compile problem on win32
jeanphi Correct fix for bug 176
ratow Embedded png icons, splash image and some other small ...
dade916 Added the new imagemap option 'integer discardmipmaps' ...
dade916 Added gain and gamma options to texture imagemap (bug ...
dade916 Commented out expf definition for __APPLE__ because ...
dade916 Applied Ratow's patch to the harlequin texture
radiance29 fixed stype<>noisebasis2 bug in distortednoise texture
radiance29 fixed null material
radiance29 fix for build on win32 in harlquin.cpp
radiance29 added anisotropic u and v roughness params to ALL gloss ...
dade916 Patch submitted by jim (bug #176), order to avoid a ...
dade916 Added Harlequin texture
dade916 Patched lux/CMakeLists.txt with Debolaz's changes for ...
dade916 Added dmsharpboundary option to loopsubdiv objects
ratow -Removed bogus toolbar
dade916 Fix: replaced few incorrect delete with delete[] in ...
dade916 Added a new dmnormalsmooth parameter to displacement ...
dade916 Added a new dmoffset parameter to displacement maps
ratow Added missing strings header
jeanphi Nearly working perspective camera interfaces for bipt.
dade916 Written the support for UV coords in loopsubdiv objects
dade916 Fixed dislpacement direction in displacement maps
dade916 Correct ObjectToWorld transformation is now used in ...


Have fun!,
Radiance

[supraserv]

Thanks for this release. I have a "small" problem though. There is an image attached to this message. I tried rendering this scene with various settings and I always get the same black meshlike elements. Also the previous CVS release gave me the same result while that of April did not.

[Radiance]

Hi supraserv,

I have noticed similar problems in a scene a couple of weeks ago,
i though it was a modeling error then.

i'm starting to believe there could be a bug in the new 'tabreckkdtree' module written by dade.
it's very important that we fix this before sunday

can you please post your lux scenefiles so we can use it to test/debug please ?
these don't seem to happen a lot so test files are difficult to get

I think this might be the same problem pinko had in a render he posted yesterday...

Radiance

[dougal2]

I think I've seen this too. Some single plane emitters don't emit, and some box emitters emit on one face only.
I thought it was a modelling error I had made.

I would knock up a test, but at the minute savannah's not letting me check out my own code

[Radiance]

i don't think it's involving emitters,
it seems a general problem in the new kdtree:

i made a bug: http://www.luxrender.net/mantis/view.php?id=222

radiance

[dougal2]

hmm I mis-read slightly. Maybe my emitter problems were bad modelling - *doing some tests now*

[Radiance]

Hey supraserv,

I've located the problem with the black bands/boxes, it's a bug in dade's new tabreckdtree indeed.

a temporary fix is to swith to the old kdtree code,
change to 'unsafekdtree' in the luxblend system panel (see screenshot attached)

we'll fix this bug asap, should be in the next weekly cvs build if all goes well

greetz,
radiance

[Dade]

I've located the problem with the black bands/boxes, it's a bug in dade's new tabreckdtree indeed.

The problem was both in the scene and waldtrianglemesh.

The scene includes degenerate triangles (i.e. a triangle where 2 or 3 vertices have the same position). waldtrianglemesh was returning a false hit on degenerate triangles with t = NaN.

I added a fast path to correctly handle degenerate triangles ... we have now the fastest degenerate triangle render in the world

[Radiance]

supraserv,

discard my previous post, it's fixed properly in the new weekly cvs build i just posted in this forum.

Radiance