So are you all using Brute Force now?

[M V]

I watched a few of Grant Marshall's videos and I saw he is using Brute Force for his Primary bounces. Are most folks using Brute now and not IR Map. Is this just a 3.0 thing?

[Scott Schroeder]

Brute force in 3.0 has been sped up dramatically to where you are almost on par with render times using imaps from previous versions. So it's now a question of why don't I use it? Combo that with the "universal" method of either adaptive or progressive rendering, and you may actually see your render times go down a bit from what you are used to in 2.4 or lower.

 

Imaps are still useful though on a per user basis. But with the speed increase and less fiddling, brute force starts to make more sense.

[RyderSK]

Since it's primarily a GI algorithm, it heavily depends upon scene type. While it's considerably faster now (since 3.0+ and Embree ) and with better universal approach (being able to directly affect primary and secondary rays ratio through single parameter of Min shading rate) makes for much better default choice for almost any still images, it's still behind in terms of speed the more complicated the GI in scene becomes.

 

And in animation case, specially regarding interior flythroughs with static geometry, IR will always reign supreme because it can be interpolated and reused across multiple frames. That alone even if they were speed similar (which they are not in this case) would provide astronomical speedup over overall animation. If you compute GI (like IR map) every 30 frames for example, that means 29 frames render in small fraction of single frame done with full GI.

 

So that makes interpolated solutions hardly obsolete, they won't go away, ever. It's just easier to use BF in more cases now (or even 100perc. times for still images).

 

Since Grant isn't from Archviz his use of BF is more than rational. Outside of Archviz, nothing outside of BF ever made sense. The speed difference in non-heavy-GI scenes is minimal, if not favourable towards BF solutions, and coupled with issues of intepolated algorithms in animation, their complicated and scene dependant setup, there is almost no reason to consider anything else.

[Ernest Burden III]

Good comments...

 

I have experimented with BF recently on a nasty interior full of downlights and shiny stone. It didn't make much difference, amd was hard to keep from being very noisy. Still, the times and quality were close enough, and I LOVE not having to ponder a thousand settings.

 

The answer of the best way to use vray is always "it depends", and while true in the sense that tweeking your arcane, non-intuitive settings can result in speedups (or NOT, per scene), there is a huge time hit in sitting there testing and comparing. I have ended up spending all afternoon poking at settings to save a half hour of final render time. And then felt like an idiot when I realized.

 

I probably will be animating that interior soon, so it will be back to settings...

[heni30]

I tried it because of positive things said here but then I'd be rendering something and I'd look over and would see that it was only 2/3 done whereas with IRR it would have been been finished. My machine isn't maxed out so maybe that makes a difference.

 

Irr settings are not that complicated and everything else is handled by universal settings. In the back of my mind "selective/adaptive" is better than "brute" (You'd think they could have come up with a more technical scientific name).

Edited January 25, 2015 by heni30

[Corey Beaulieu]

What Juraj said.....

 

I will add that while the Brute Force algorithm is faster in 3.0, I think the speed is due to meeting the noise requirement at lower settings. This is my own visual assessment, but if you push the engine as hard as was necessary before, it's going to take just as long from the few tests I've done.

 

Universal settings are a key to ending your render early (faster) or if you want to go your own way, I would try lowering the Subdivs a bit. Lately I've been trying set up a minimum level of acceptability and using the Global Subdiv Multiplier to step up my quality through the delivery phases. It isn't a perfect system, but it is a no brainer for settings.

 

I've also never tried it with Embree as many of my scenes push the memory limits as is and the write up says it will increase memory usage, but including that feature will be my next goal.