I noticed that after installing Houdini 11 some of the lighting/rendering parameters and workflow has changed, and that the default settings for the environment light for example, still produces fast results, however they basically look like TV noise now.

So I've played around a little with the settings and tried to make a (as far as I know) fair, simple scene to compare performance and try settings in.

I have probably/hopefully missed something somewhere, but I have a hard time reproducing the same image quality in Houdini 11 without it impacting the render times quite a bit.

A simple scene with a sphere on a plane and a environment light produced the following results after I've matched the noise by adjusting the light and reflection samples, leaving all other settings the same in both scenes:

H10, Clay only: 2sec
H11, Clay only: 5sec
H10, Reflection/Checker: 4sec
H11, Reflection/Checker: 10sec

Sure these are short renders thus a bit unpredictable but so far in my experience Houdini 11 takes quite a bit longer to produce similar results, especially with smooth shadows and reflections.

Also the texture filtering is very different looking at the checker in the back, in Houdini 11 it seems to just remove all contrast whereas in Houdini 10 the checker stays crisp.

oh yes i found the problem too,when i render glass,the specular and reflecty look nicer,but the speed is much slower

In the middle of something atm, but I would point out that 10 second renders and very simple scenes aren't a good metric for this. Where the power of the new lighting, especially with PBR(which I assume you're not using), can give you very good looking stuff quite quickly. I understand you're comparing 10 to 11 directly, and that's perfectly valid, but I'd definitely look at something a little more detailed for before's and after's.

Managing noise is indeed important with area and env lights, and has it's own little tricks. I find that for some things, actually using PBR will give me better looking results faster than micropoly. There are definitely growing pains, though, with the light changes, I'm not trivializing that. When you're doing stuff for final, with larger datasets than a couple of prims, I think you'll see the advantages.

Cheers,

J.C.

JColdrick
In the middle of something atm, but I would point out that 10 second renders and very simple scenes aren't a good metric for this.

J.C.

yes thankyou i agree

I think the challenge is for SESI to get as many as possible the HOWTOs, caveats and gotcha's into the docs so setting up quite sweet looking renders isn't too harsh. I know that at the moment if you just dive in and start throwing down env and area lights and fire up some of the existing sample shaders(beware marble! ), you may well be hit hard with questionable render times and results.

There is some good reading in the docs at the moment, worth reading the new stuff about lights rendering and PBR.

Cheers,

J.C.