this is a classic problem where the way you’ve specified the input just makes it too hard for the renderer to “find” the light. Most of the algorithms that are used these days are pretty dumb, i.e. they cannot figure out by themselves that the only light that actually makes it into the room is coming through the window.
You have two options: switch to a smarter algorithm like MLT or PSSMLT, which have a 1-step memory of “relevant” paths. But you must make sure to set their “directSamples” parameter to -1 so that MLT is even used for the direct illumination component (this is handled with another technique by default which, in your scene, would just add a lot of noise).
The other option is to make an area light which approximates your environment and put it in front of the window.
About speedup - best solution would be using area lamp inside room, but JW refused to make transparent area lights , since this is physically incorrect. But maybe if you ask him (again), he will change mind.
I requested earlier today, but WJ still does not want to incorporate non-rendering emitter sources. It is against physical laws, and Mitsuba, is strict about being physically correct.
The next bug-fix release of Mitsuba is available, which has the following improvements:
Volumetric path tracers: improved sampling when dealing with index-matched medium transitions. This is essentially a re-implementation of an optimization that Mitsuba 0.3.1 already had, but which got lost in the bidirectional rewrite.
Batch tonemapper: due to an unfortunate bug, the batch tonemapper in the last release produced invalid results for images containing an alpha channel. This is now fixed.
Shapes: corrected some differential geometry issues in the “cylinder” and “rectangle” shapes.
MLT: fixed 2-stage MLT, which was producing incorrect results.
MEPT: fixed the handling of directional light sources.
Robustness: got rid of various corner-cases that could produce NaNs.
Filenames: to facilitate loading scenes created on Windows/OSX, the Linux version now resolves files case-insensitively if they could not be found after a case-sensitive search.
Python: added Python bindings for shapes and triangle meshes. The Python plugin should now be easier to load (previously, this was unfortunately rather difficult on several platforms). The documentation was also given an overhaul.
Particle tracing: I’ve decided to disable the adjoint BSDF for shading normals in the particle tracer, since it causes an unacceptable amount of variance in scenes containing poorly tesselated geometry. This affects the plugins ptracer, ppm, sppm and photonmapper. See the commit for further details.
Subsurface scattering: fixed parallel network renderings involving the dipole model.
Homogeneous medium & dipole: added many more material presets by Narasimhan et al.
OBJ loader: further robustness improvements to the OBJ loader and the associated MTL material translator.
i had a similar problem with UV’s, but i think it was related to my own build of COLLADA, or something along those lines. i set the UV scale to 1,-1 respectively for textures and textures would export correctly
“- Export only one light of type HEMI or SUN.”? this doesn’t mean you can only have a hemi OR a sun, does it? (Hemisphere doesn’t work with latest exporter…)
Sun does’t work if rotation is 0,0,0
“- Reutilize meshes from serialized file when needed.” should this mean it doesn’t export already “serialized” meshes? (cause it does re-export)
In the tests scenes I used, Mitsuba terminated with this error when more than one light of type Sun or Hemi was present:
Caught a critical exeption: ERROR main [Scene] The scene may only contain one environment emitter.
What do you mean Hemisphere doesn’t work with latest exporter? What error do you get?
I will check about Sun rotation. Thank you!
It means that if two or more objects (duplis, linked objects) use the same mesh, it’s exported only once and used on both objects. Before that, all objects were exported as unique meshes and the serialized file contained duplicate meshes, which was inefficient but was the fastest way I could implement the serializer to replace collada exporter.
Right now, the scene is always re-exported, I am looking into changing this.
With bug fixes in mitsuba 4.2 I was able to render this in half time compared to previous version in mitsuba 4.1 - 3minutes with PT. And with less noise. Exporter seems to work great - but there in no partial mesh export right? Car is bit heavy, and it get exported every time, even though there is no mesh change.