Manifold Sampling for Rendering Caustics and Glints!

Guys, look at this, FINALLY fast caustics way beyond MLT speed!
We need this in Cycles, this would finally allow true and FAST full light transport!

This is already implemented in GPU Optix on their github project!
The sample count of around 128 to 64 samples for FULL CLEAN render convergence is INSANE!
Imagine the speed cycles would have, OMG!

This would bring Cycles roughly some 100 billion lightyears ahead any render engine. :smiley:

UPDATE: Their github code:

@brecht @lukasstockner97


Not a day passes without me hearing about some new light transport method, and I never see them end up in Cycles. i guess it’s not as easy as plug-and-play… maybe this one will be the exception ? Here’s to hoping :beers:

(just watched the demo, it’s incredible)

Yes, you’re right, many things were proposed, but this is the first complete light transport solution that has no real compromises. Also GPU friendly. This is worth implementing.
Yes, it’s incredible. I’m playing with the github stuff now.


Let’s hope for this. Would this be conceptually spectral-friendly ? @smilebags hope it’s ok to ping you over here, see if you have thoughts on this. My understanding is -since this concerns sampling patterns, it would be mostly removed from the changes in the spectral branch, and thus sounds rather combine-able.
There’s a big chance I could be talking out my ass too. Do not hesitate to point it out.

Wenzel Jakob (Mitsuba) is one of the authors, last I checked his experimental render engine from years ago is spectral-based.

The header suggests Autodesk was involved as well, it is possible they will integrate it into the Arnold engine in the future, which means it would be wise for the Cycles devs. to implement this as well. Unfortunately, I can’t make an educated opinion because the visual demo will not load properly on my copy of Chrome (as it buffers endlessly after the first few seconds).

I haven’t looked into it but generally this sort of thing would be able to be combined with spectral rendering relatively easily. I would be surprised if Wenzel Jakob designed something which was not able to work well in a spectral engine.

Well, this sounds good.

I will take a look into this and come back with a more definitive answer. Though ‘it’s possible’ and ‘I can do it’ are two very different things, I’m not particularly savvy about light transport algorithms.

EDIT: It’s implemented in Mitsuba 2 so it definitely can play nicely with spectral rendering.


Oh, already in Mistuba main branch ? I totally missed that. Great.

I’m not sure whether it has been merged into Mitsuba 2, but I think an implementation of it was built for Mitsuba. As a research engine, I doubt they merge each new technique they come up with, it seems like it would add a lot of complexity to the engine if they did.

Can someone help me please to install Mitsuba + CPPM + PSSMLT etc…

Looks good. I hope something like this makes it into Cycles at some point - caustics and glints are important aspects of many scenes.