Is IBL faster?

[simmsimaging]

I've used IBL in various ways with Vray since I started with it, but I'm wondering if lighting a scene *only* with HDR (in a vray light for example) would be generically faster than using actual lights? ( I almost exclusively used image-mapped Vray plane lights)

I ask because I'm learning that several other engines boast much quicker render times if you just use IBL and scrap the localized lights.

So, I could model a studio setup, render a spherical HDR of that studio, then dump the studio and light the scene with the HDR version of it, and gain quite a bit of render speed for the same quality, sometimes better. I'm wondering how this applies to Vray? Would I be better off (speed wise) to switch to an HDR only workflow, if I could do what I needed for lighting using that route?

Would the same apply to RT as well as the production rendering?

b

[vlado]

It would be best to make some tests to see which method is faster for your particular case. Using IBL will not give you quite the same results as actual lights, since it assumes that the light comes from infinitely far away, whereas actual lights have a definite position and therefore proper light decay. This wouldn't matter much for outdoor lighting or lighting with distant sources where the light decay is negligible, but in other cases the decay might be important to emphasize object dimensions.

As far as the RT goes, there won't be much of a speed difference; the noise levels might be different, although probably not too different, since the dome light does proper importance sampling.

Best regards,
Vlado

[simmsimaging]

I never considered the fall-off factor - thanks. My feeling is that I would probably need to do key lighting for smaller things with lights, but try to take care of fill lighting, generic reflections (things like walls/floors) with IBL. I can see that it would likely be faster to have the reflections of a soft highlight on a wall coming from an image map rather than calculating the glossy reflections on the wall and then back to the object.

I will test it, but I'm sure I have accumulated about 6 years worth of back-testing so I'm always hopeful that someone already has a definitive answer

b