place a light on a hdri?

[rmejia]

with gamma correction

Are you lowering the Gamma in the VRayHDRI?

Did some tests and the only way I get defined shadows using the Dome light was lowering the Gamma in the VRayHDRI map. The first image is Gamma 1, the second Gamma .5

[BBB3]

I either lower the gamma upon loading if using an .exr, or lower it in the VrayHDRI map if using an .hdr. I tend to use a gamma of 0.456.
What it does is basically pumping up the contrast of the map. Sometimes it works great and brings a map that would be too faint back into the right colour space. Sometimes, however, it's just too harsh and will affect the colours too much. I guess it's a case-by-case thing, but I found it to work quite well and give really nice saturation and light definition when using the sIBL samples (only the hi-def ones, though).
My render settings are very basic. A high subdivision on the domelight - 30 or so - and, since these were just quick tests, just an IR at medium factory settings. No second bounce for those.
B

[rmejia]

An issue with shadows is also LWF which washes out the image somewhat. Here are 3 renders, the first with the gamma set to 1.0, the second with LWF set to 2.2 and the third is the 2.2 image photoshoped. The shadows can be popped out more in ps, but they still are not that sharp as lowering the gamma on the .exr or HDRI map.

[jujubee]

Hmmm, so you can get much sharper shadows through knocking the gamma down to .4545 with subs to 30 - at the cost of blowing out the background and possibly blowing out highlights/reflections. I'm trying to do the math in my head regarding that - that shouldn't have to be done for more natural results using LWF. But hey, if it works for you and you get good results then that's all that matters. I personally prefer sIBL because it won't blow out my highlights and renders extremely fast. There needs to be some additional support and fixes still for the Vray template since the latest release - the North parameter and a few other items/bugs. Several months ago, I had requested support for the Dome Light and Dschaga started messing around with the template - but he's got his hands full of real work. If someone knows Maxscript and wants to volunteer/help out it would be much appreciated...

Here's something I was messing around with last night (I kinda wanted to test the new video function on this forum lol.) Keep in mind the reflection and env gamma is slightly off because of VrayRT Activeshade versus using VFB:

[BBB3]

Yes, I guess it boils down to personal taste. The render times and the inability to use the hdr as backplate don't bother me as I have the processing power and I would never use a spherical map as backplate - too pixelated unless you get into the 20,000 pixel resolution region. The issue with the highlights is more problematic, though I found this workflow in many cases to deliver better highlights, at least in LWF.
Certainly for animations, where render times really matter, the sIBL way would be the way to go, it seems. Your VRayRT demo is certainly convincing.

[instinct]

It depends a lot on the HDRi dynamic range from my experience. If you're using Paul debevec's full day set for example you get razor sharp shadows. As the dynamic range is very high with the sun at the correct value (around 350 at noon if i recall right).

Kind Regards,
Thorsten

[peterguthrie]

I've been pretty happy with the shadows from my homemade HDR images with have a very large dynamic range of 18EVs (going by the table on hdrlabs.com anyway) but even then sometimes I need to supplement them with a vraysun.

Since this thread isn't explicitly about sIBL does anyone know how to wire the rotation angle of a vray sun (angle based on where it and it's target are in relation to each other) to the 0-1 rotation range in the HDR map? Best answer gets a free HDR! http://www.flickr.com/photos/pg/3459089563/

[instinct]

Most straight forward would be linking both to a dummy in "full north" position. Then rotate the dummy to turn the sun around and wire the rotation of the dummy to the offset (dummy.rotation/360. Am not sure if the controller holds degrees or rad, so you might need a diff factor or use rad2deg())

Regards,
Thorsten

[instinct]

Oh. And this only works properly if your HDRIs are north centered (or centered to any axis for that matter) of course

Kind Regards,
Thorsten

[peterguthrie]

Thanks Thorsten! HDRs are all north centered, will try it out.

[tammo]

I am still not convinced.
All the sIBL renderings here look more like CGI then the vray domelight renderings with only the HDR.

Maybe the 3d sun is not in the right position,
or the size, color or intensity is different?

If you look at the shadows and lighting of the "real" objects in the background image like the cars or trees
and compare that to the shadows and lighting of the teapot you can see that there is a difference even with another gamma value.

[jujubee]

I am still not convinced. All the sIBL renderings here look more like CGI then the vray domelight renderings with only the HDR.

Perhaps you have an outdoors scene of your own that you would like to share showing what looks correct to you?

But hey, if you don't like it, then don't use it. I would say however, to at least give it a try before forming opinions/conclusions since it's free and there's some free libraries out there. Worst case scenario, you can rip out the original HDR and use it any way you like to. On the other hand, I see plenty of plain HDRs which look like they're being used wrong and not casting strong enough shadows without resorting to some of trickery. I think there's a difference in what we are "used to" seeing and what we "should" be seeing. But if the original question was how to match a light with the sun position - then sIBLedit is one way to do this.

We should come up with an outdoor test scene to test the various methods and what lighting looks most correct - no overcast days and no post-processing.

[BBB3]

Jujubee, is the sIBL methode only geared towards sunlit maps (i.e. one strong light source), or will it adapt to, say, an interior hdr with potentially dozens of strong light sources of different intensities.

These are the kinds of maps I find harder to use with my methode as the calculations made by the dome light seem to be much more complex and the render time shoots up (although the final result is good, if you have the patience).

By the way, I tried some of the earlier maps with a subdiv of 10 on the dome light and they seem to work fine. Maybe 30 was overkill, though it affects render time less than I thought.

[jujubee]

or will it adapt to, say, an interior hdr with potentially dozens of strong light sources of different intensities.

There is support for multiple light sources, however multiple light sources doesn't currently work for the Vray exporter until somebody adds additional code (I can't code well plus I don't know Maxscript.) It shouldn't be difficult, but it has been difficult to find a volunteer who uses both Vray and knows Maxscript. The only person that has worked on the code from here is Dschaga. Multiple lights will work currently in Lightwave, Maya, and a few other packages that some of the coders have been more regular with. Ideally, you could add multiple lights to a room at varying intensities. Otherwise, it should more or less work like an ordinary HDR setup - an efficient HDR environment blur combined with a non-blurred, higher resolution HDR for the reflections override.

Slightly off-topic, but in the future there are plans for backplate support as well - you'll be able to choose from a series of images and everything should automatically line-up.

[snivlem]

I can't even get the bloody thing to find HDRI's. Everytime I point it to a directory it forgets, asks me where my HDR's are saved, I tell it then it says it already exists and shows me an empty selection!