Corona and Kerkythea MLT Caustics comparison after 15 min of rendering.

[quadrays]

Here you are. Just simple question. What is wrong? What should I do, to achieve better Corona result in less time ? I hope caustics calculation will be only better in new official releases of Corona.

Notice that the most focused reflected light in KT 2008 Echo is more smooth, less noise than that Corona tube render.

[racoonart]

Well, your apples look perfectly fine compared to the steering wheel in that car.

What i mean: You're comparing normal pathtracing with bi-directional (+metropolis) pathtracing. Shadows look better in your corona image, caustics are worse. Exactly why one method is chosen for certain tasks and the other one for other things.

[Siahpoosh]

can u post your scene here ? (for test)

[iliyang]

Why don't you try bidirectional path tracing + vertex merging in Corona?

[quadrays]

Yep, it's just simple tube with reflection on fresnel IOR higher than 4,00 and one little planar light source. As alternative, there is also plane with CoronaLightMtl just to test two different light sources. And this additional plane is also useful to export it with rest of scene as OBJ to Kerkythea 2008 engine.

All was rendered on Core 2 Quad Q9300 @ 2.50 GHz

There is a scene file.

[Ondra]

Yep, it's just simple tube with reflection on fresnel IOR higher than 4,00 and one little planar light source. As alternative, there is also plane with CoronaLightMtl just to test two different light sources. And this additional plane is also useful to export it with rest of scene as OBJ to Kerkythea 2008 engine.

All was rendered on Core 2 Quad Q9300 @ 2.50 GHz

There is a scene file.

Corona BDPT, no metropolis, 1minute-ish.

Cheers

[Score]

Looks like we need better caustics calculations.