The best of all worlds rendering technique?

[CoolColJ]

Well I have found Stochastic progressive photon mapping, in Luxrender unmatched when it comes to difficult lighting, and refracted caustics. Even MLT can't match it, especially refracted caustics from the Sun! But it's slow to converge.
But if you could combine Bidrectional Path tracing and SPPM then it would could handle every lighting scenario?

How about Bidirectional Light Transport with Vertex Merging
https://graphics.cg.uni-saarland.de/201 ... x-merging/

We present vertex merging - a bidirectional path sampling technique for Monte Carlo light transport integration. Vertex merging is simple and more computationally efficient for specular-diffuse-specular effects than the currently available techniques in bidirectional path tracing. It brings the advantages of photon mapping to the path integral framework, while avoiding the concept of density estimation altogether. This makes it possible for the first time to quantitatively reason about the efficiency of two rendering approaches that have been historically considered conceptually different. The practical result is a combined bidirectional rendering algorithm that efficiently handles a wide variety of lighting conditions, ranging from direct illumination and diffuse inter-reflections to the notoriously problematic reflected caustics. This algorithm also has a higher order of convergence than progressive photon mapping.

And a technical report - Unified sampling framework for Monte Carlo path integration and photon density estimation
http://cs.au.dk/~toshiya/ups_tr.pdf

[StompinTom]

Looks very sweet from the pictures

[CoolColJ]

The power of SPPM - from luxrender. Combine this with Bidirectional Path tracing and you have a winner....as shown in the above papers

Both 15 min renders - SPPM, he underwater caustics and it's reflection in the sphere rendered within 1 min!
And yes it is unbiased

SPPM - filtering is off

MLT

And yes I've done this render in Indigo and it's nowhere near the speed of SPPM with caustics

[lycium]

SPPM isn't unbiased (it's consistent though), and its order of convergence isn't as good as BDPT in most respects, reflections of caustics being an exception

[CoolColJ]

SPPM isn't unbiased (it's consistent though), and its order of convergence isn't as good as BDPT in most respects, reflections of caustics being an exception

That's why you combine the two together

[lycium]

It does seem like a promising method, worth exploring

[CoolColJ]

supplemental PDF showing a comparison between the differnet methods
http://cgg.mff.cuni.cz/~jaroslav/papers ... mental.pdf


And something else of interest? No one has put out a GPU accelerated BPT renderer out yet... should be able to now

Combinatorial Bidirectional Path-Tracing
for Efficient Hybrid CPU/GPU Rendering

This paper presents a reformulation of bidirectional path-tracing that adequately divides the algorithm into pro-
cesses efficiently executed in parallel on both the CPU and the GPU. We thus benefit from high-level optimization
techniques such as double buffering, batch processing, and asyncronous execution, as well as from the exploitation
of most of the CPU, GPU, and memory bus capabilities. Our approach, while avoiding pure GPU implementation
limitations (such as limited complexity of shaders, light or camera models, and processed scene data sets), is more
than ten times faster than standard bidirectional path-tracing implementations, leading to performance suitable
for production-oriented rendering engines.

http://www.irit.fr/~Anthony.Pajot/publis/cbpt/cbpt.pdf

[CTZn]

yes this ! Sounds cool. IIRC Ono mentioned GPU acceleration for bidir to happen at some point.

[StompinTom]

Do want! It looks like it would handle scenes with lots of glass and tiny light sources (i.e. typical badass arch-viz scene )

[Zom-B]

And something else of interest? No one has put out a GPU accelerated BPT renderer out yet... should be able to now

Already sent that paper to Glare on 28 February, I'm sure they started to read it