Mitsuba renderer

Discuss stuff not about Indigo.
User avatar
dakiru
Indigo 100
Posts: 746
Joined: Mon Jun 22, 2009 6:51 am

Re: Mitsuba renderer

Post by dakiru » Tue Feb 01, 2011 4:43 pm

StompinTom wrote:
dakiru wrote:The scarf is extremely nice! Interesting to know how it was made, what do you think, guys? 8) Simulation of the detailed mesh, maybe? The image is in the volume rendering section, so I have no idea.. :?:
It's a volume rendering :) No mesh.
Well, uhmm, but how? :o

StompinTom
Indigo 100
Posts: 1821
Joined: Mon Sep 04, 2006 3:33 pm

Re: Mitsuba renderer

Post by StompinTom » Tue Feb 01, 2011 7:31 pm

dakiru wrote:
StompinTom wrote:
dakiru wrote:The scarf is extremely nice! Interesting to know how it was made, what do you think, guys? 8) Simulation of the detailed mesh, maybe? The image is in the volume rendering section, so I have no idea.. :?:
It's a volume rendering :) No mesh.
Well, uhmm, but how? :o
"The underlying volumes can be represented as sparse voxel octrees or as hierarchical grids, where grid cells are directly mapped from files into memory."

User avatar
SaphireS
Indigo 100
Posts: 134
Joined: Tue May 05, 2009 9:06 am
Location: Austria
Contact:

Re: Mitsuba renderer

Post by SaphireS » Tue Feb 01, 2011 10:10 pm

It was a one-man-project, mainly for researching algorithms, as stated.
IIRC it was used for the master thesis and PhD-research of Wenzel Jakob (the developer).
But he's still working on it, now under a open source licence.
And he is closely working together with some Blender Devs to get a very good COLLADA support between Blender and Mitsuba, which of course means good interaction with every other app that can properly export COLLADA-files (unlike Blender now :P )

To sum it up: One-Man Project, only based on the newest cutting-edge algorithms and open source = awesome, imho.
Your ad here!

User avatar
snorky
Indigo 100
Posts: 327
Joined: Fri Apr 17, 2009 8:54 pm
Location: Parma, Italy

Re: Mitsuba renderer

Post by snorky » Wed Feb 02, 2011 4:38 am

My Norton hates Mitsuba...

mtssrv.exe

http://securityresponse.symantec.com/se ... 00-2232-99

:x

User avatar
dag
Posts: 189
Joined: Tue Feb 02, 2010 2:28 am

Re: Mitsuba renderer

Post by dag » Wed Feb 02, 2011 5:16 am

snorky wrote:My Norton hates Mitsuba...

mtssrv.exe

http://securityresponse.symantec.com/se ... 00-2232-99

:x
My brain hates Norton.

User avatar
hcpiter
1st Place Winner
Posts: 242
Joined: Fri Feb 19, 2010 9:19 am
Location: Poland
Contact:

Re: Mitsuba renderer

Post by hcpiter » Wed Feb 02, 2011 6:46 am

who use Norton :D ?! I remember one time when I made scan my hdd. All files from .max was deleted :p

User avatar
fused
Developer
Posts: 3636
Joined: Fri Sep 22, 2006 7:19 am
Location: Berlin, Germany
3D Software: Cinema 4D

Re: Mitsuba renderer

Post by fused » Wed Feb 02, 2011 6:49 am

hcpiter wrote:who use Norton :D
Snorky! :D

User avatar
hcpiter
1st Place Winner
Posts: 242
Joined: Fri Feb 19, 2010 9:19 am
Location: Poland
Contact:

Re: Mitsuba renderer

Post by hcpiter » Wed Feb 02, 2011 7:07 am

fused wrote:
hcpiter wrote:who use Norton :D
Snorky! :D
hehe lol :mrgreen:

User avatar
snorky
Indigo 100
Posts: 327
Joined: Fri Apr 17, 2009 8:54 pm
Location: Parma, Italy

Re: Mitsuba renderer

Post by snorky » Wed Feb 02, 2011 9:11 pm

hcpiter wrote:
fused wrote:
hcpiter wrote:who use Norton :D
Snorky! :D
hehe lol :mrgreen:
my head office loves norton... :mrgreen:

User avatar
CoolColJ
Posts: 1734
Joined: Mon Jun 25, 2007 1:47 pm

Re: Mitsuba renderer

Post by CoolColJ » Sat Feb 05, 2011 11:11 am

The spirit of Radium renderer lives on :)


Ah no MLT in this release

User avatar
CoolColJ
Posts: 1734
Joined: Mon Jun 25, 2007 1:47 pm

Re: Mitsuba renderer

Post by CoolColJ » Tue Oct 02, 2012 7:15 pm

Mitsuba 0.4 gets two types of MLT among lots of goodies, seems pretty fast too 8)
Especially the Veach method. The other simplified one is like Indigo's, which piggy backs Path or Bidirectional path tracing.

Tried the Veach MLT method on the Sponza scene and it will hit 1024 samples in 6+ minutes on my i7 3930k at 4.2ghz system!


http://www.mitsuba-renderer.org/devblog ... nt+Blog%29

Veach MLT method from the manual
Like pssmlt, this integrator explores the space of light paths, searching with preference for those
that carry a significant amount of energy from an emitter to the sensor. The main difference is that
PSSMLT does this exploration by piggybacking on another rendering technique and “manipulating”
the random number stream that drives it, whereas MLT does not use such an indirection: it operates
directly on the actual light paths.
This means that the algorithm has access to considerably more information about the problem
to be solved, which allows it to perform a directed exploration of certain classes of light paths. The
main downside is that the implementation is rather complex, which may make it more susceptible
to unforeseen problems. Mitsuba reproduces the full MLT algorithm except for the lens subpath
mutation15. In addition, the plugin also provides the manifold perturbation proposed by Jakob and
Marschner

Path Space Metropolis Light Transport (MLT) is a seminal rendering technique proposed by Veach and Guibas, which applies the Metropolis-Hastings algorithm to the path-space formulation of light transport.

In contrast to simple methods like path tracing that render images by performing a naïve and memoryless random search for light paths, MLT actively searches for relevant light paths. Once such a path is found, the algorithm tries to explore neighboring paths to amortize the cost of the search. This is done with a clever set of path perturbations which can efficiently explore certain classes of paths. This method can often significantly improve the convergence rate of renderings that involve what one might call “difficult” input.

To my knowledge, this is the first publicly available implementation of this algorithm that works correctly.

Primary Sample Space Metropolis Light Transport (PSSMLT) by Kelemen et al. is a simplified version of the above algorithm.
Like MLT, this method relies on Markov Chain Monte Carlo integration, and it systematically explores the space of light paths, searching with preference for those that carry a significant amount of energy from an emitter to the sensor. The main difference is that PSSMLT does this exploration by piggybacking on another rendering technique and “manipulating” the random number stream that drives it. The Mitsuba version can operate either on top a unidirectional path tracer or a fully-fledged bidirectional path tracer with multiple importance sampling. This is a nice method to use when a scene is a little bit too difficult for a bidirectional path tracer to handle, in which case the extra adaptiveness due PSSMLT can bring it back into the realm of things that can be rendered within a reasonable amount of time.

This is the algorithm that’s widely implemented in commercial rendering packages that mention “Metropolis Light Transport” somewhere in their product description.

Energy redistribution path tracing by Cline et al. combines aspects of Path Tracing with the exploration strategies of Veach and Guibas. This method generates a large number of paths using a standard path tracing method, which are then used to seed a MLT-style renderer. It works hand in and with the next method:

Manifold Exploration by Jakob and Marschner is based on the idea that sets of paths contributing to the image naturally form manifolds in path space, which can be explored locally by a simple equation-solving iteration. This leads to a method that can render scenes involving complex specular and near-specular paths, which have traditionally been a source of difficulty in unbiased methods. The following renderings images (scene courtesy of Olesya Isaenko) were created with this method:

User avatar
CoolColJ
Posts: 1734
Joined: Mon Jun 25, 2007 1:47 pm

Re: Mitsuba renderer

Post by CoolColJ » Tue Oct 02, 2012 9:41 pm

12 minute renders with the Veach MLT, on my i7 3930k at 4.2ghz system

Very impressive!
That Torus scene is like a pool/caustics scene, very difficult in the standard MLT in Indigo/Luxrender etc
Veach_MLT_Torus.jpg
Veach_MLT_Room.jpg

User avatar
CoolColJ
Posts: 1734
Joined: Mon Jun 25, 2007 1:47 pm

Re: Mitsuba renderer

Post by CoolColJ » Wed Oct 03, 2012 1:08 am

Veach MLT - direct light sampling off, about 15 mins or so render time

Manifold Perturbation option on vs off
Turning it on slows down the sample rate, but improves the convergence of the torus and refracted caustics - handy for pool scenes!

I also found it better to set number of bounces to -1, infinite, works better, more detail, and is more unbiased :)
VeachMLT_Torus_Manifold-Perturbation-on.jpg
Torus.jpg
from the PDF manual
To explore the space of light paths, MLT iteratively makes changes to a light path, which can either
belarge-scalemutationsorsmall-scaleperturbations. Roughlyspeaking,thebidirectionalmutationis
usedtojumpbetweendifferentclassesoflightpaths, andeachoneoftheperturbationsisresponsible
for efficiently exploring some of these classes. All mutation and perturbation strategies can be mixed
and matched as desired, though for the algorithm to work properly, the bidirectional mutation must
be active and perturbations should be selected as required based on the types of light paths that are
present in the input scene. The following perturbations are available:

(a) Lensperturbation: thisperturbationslightlyvariestheoutgoingdirectionatthecameraandprop-
agates the resulting ray until it encounters the first non-specular object. The perturbation then
attempts to create a connection to the (unchanged) remainder of the path.

(b) Caustic perturbation: essentially a lens perturbation that proceeds in the opposite direction.

(c) Multi-chain perturbation: used when there are several chains of specular interactions, as seen
in the swimming pool example above. After an initial lens perturbation, a cascade of additional
perturbationsisrequireduntilaconnectiontotheremainderofthepathcanfinallybeestablished.
Depending on the path type, the entire path may be changed by this.

(d) Manifold perturbation: thisperturbationwasdesignedtosubsumeandextendthepreviousthree
approaches. Itcreates a perturbation at anarbitrary position along thepath, proceedingin either
direction. Uponencounteringachainofspecularinteractions,itnumericallysolvesforaconnec-
tion path (as opposed to the cascading mechanism employed by the multi-chain perturbation).

User avatar
CoolColJ
Posts: 1734
Joined: Mon Jun 25, 2007 1:47 pm

Re: Mitsuba renderer

Post by CoolColJ » Wed Oct 03, 2012 1:10 am

Veach MLT - 10min render - only source of light is indirect lighting through windows and geometry seams :P
No exit portals or anything here :shock:
VeachMLT_RoomPlant_10mins.jpg
Wanted to see how many samples to reach a clean shadows with Veach MLT
8192 samples here - 30 min render
VeachMLT_Spheres_8192samples-30mins.jpg

User avatar
Pibuz
1st Place 100
Posts: 2646
Joined: Tue Dec 11, 2007 7:58 am
Location: Padua, Italy
3D Software: SketchUp

Re: Mitsuba renderer

Post by Pibuz » Wed Oct 03, 2012 3:07 am

Hi cool!
Mitsuba seems to be a very cool rendering engine, even more after this very interesting last release!
..sadly there is no direct support for exporters, I suppose, so there's no way it is usable to us poor fiddlers.. :lol: I'm following with interest btw!

Post Reply
37 posts

Who is online

Users browsing this forum: No registered users and 5 guests