Some images from my renderer

[yiningkarlli]

Hello all,

I've been following the ompf2 (and before that, ompf) boards for some time now, but haven't really posted aside from one or two posts a long time ago. I thought for a first "real" post, I'd share some images from my renderer, called Takua Render. I wrote Takua Render from scratch and have been working on it for around four years now. It started out as a unidirectional pathtracer, but now is primarily a VCM renderer (which correspondingly means it also supports PPM, BDPT, etc).

Some images produced by my renderer:

Edit: apologies for the first four images; the first four images are rendered at 1920x1080, but I'm not sure how to add image resize tags in the forum's post editor thing.










Hope everyone likes them!

-YKL

[shocker_0x15]

Cool!!

How much vertices do the flowers consist of?

[ypoissant]

Very nice.

How do you handle the lackered material on the chair?

[beason]

Wow, very cool! Would love to see more renderings.

[yiningkarlli]

Wow, very cool! Would love to see more renderings.

More coming soon! I have a couple of cool things in the works. In the meantime, there's a lot more images (and also my writeups and stuff) on my development blog: http://blog.yiningkarlli.com/

How do you handle the lackered material on the chair?

My renderer has a layered shading system not too dissimilar from Vray's layered shader or Renderman 19/RIS's LM system. So the chair uses a layered BSDF with a base diffuse BSDF with the color, and then a clearcoat layer on top and then a couple of layers with fingerprints and scratches and stuff, although the fingerprints and scratches don't really end up being visible...

How much vertices do the flowers consist of?

The flowers are close to a quarter of a billion triangles, which took an awful lot of memory. Detailed writeup here: http://blog.yiningkarlli.com/2015/02/fl ... ender.html

Thanks so much everyone!

[koiava]

Very nice renders and layered materials looks great, fingerprint layer was highest level of awesomeness
VCM rocks, SDS paths is totally clear what was render time for second image(Lilacs in glass vase)?
Flower scene is not a simple case for BVH with object splits, because AABBs doesn't fits well on this thin rods, so which acceleration structure are you using? Do you use spatial splits?

[mpeterson]

Very nice renders and layered materials looks great, fingerprint layer was highest level of awesomeness
VCM rocks, SDS paths is totally clear what was render time for second image(Lilacs in glass vase)?
Flower scene is not a simple case for BVH with object splits, because AABBs doesn't fits well on this thin rods, so which acceleration structure are you using? Do you use spatial splits?

vcm rocks ??? where are you coming from ? it is for shure the slowest approach to handle complex lighting/illumination

[koiava]

vcm rocks ??? where are you coming from ? it is for shure the slowest approach to handle complex lighting/illumination

In common cases unidirectional path tracing with external light sampling and MIS is industry standard, but there is situations where this approach has problems.(for example "flowers in glass vase" scene)
In the set of unbiased techniques we has BDPT with MIS, which has really good results for difficult cases(I mean difficult for unidirectional PT), but it has can't simply find SDS paths efficiently, because at specular event Importance Sampling is hard to do.
VCM isn't unbiased, because it's combinations of PPM and BDPT, it's consistent, So for surfaces PT/VCM integrator is probably most useful at this time

[ypoissant]

My renderer has a layered shading system not too dissimilar from Vray's layered shader or Renderman 19/RIS's LM system. So the chair uses a layered BSDF with a base diffuse BSDF with the color, and then a clearcoat layer on top and then a couple of layers with fingerprints and scratches and stuff, although the fingerprints and scratches don't really end up being visible...

This piques my curiosity a bit more.

Do you have a compound layered BTDF that computes all the layer effects together or are each layer considered as a different medium which you trace through one after the other?