what the...

[analogueWay]

VRay IS better than that, but I still get great results using convolved HDRI for lighting and reflection. Downsizing decreases render times since sample-lookup is quicker on a smaller map. Blurring reduces the precision and softens the shadows (a benefit or a detriment depending on the project) - I forget, but I think that this also reduces render times somewhat.

That said, lately I have been using full-resolution HDRI's for both lighting and reflection because it looks so nice.

And, in the case of the sun, your idea is great. (Would you still use the map with the sun in it for reflections?)

-Jeremy

[cpnichols]

And, in the case of the sun, your idea is great. (Would you still use the map with the sun in it for reflections?)

Yes... the sun needs to be reflected. BTW, you can replace the word "Lighting" in your last statement with "diffuse" and the word "reflection" with "specular" and you are all set. See where I'm going?

[analogueWay]

Absolutely.

My last question was based on still being fairly new to VRay. I should have asked it in this way: "Will VRay show the light-source-as-sun in the reflection, or do we need to use an hdri with the sun still in it for the reflection map?"

I love the VRay material and how it handles reflection and specular (they are intertwined, but generally treated as separate entities in most computer graphis apps. I am just not familiar enough with VRay to know if light sources will (as they should) show up as bright points/areas in reflections.

I the time it took me to type that, I could have tested it to find out.

-Jeremy

[analogueWay]

I set up a simple scene with a greeble surface and four spheres. The greeble surface has a default material applied to it, and the spheres are glass, default, chrome, and a bead-blasted aluminum.

The scene was rendered using the default GI settings within VRay with the exception that the Mitchel-Netravali filter was used instead of the Area filter.

Default lights were disabled and HDR images were used for the VRAY GI Environment (skylight) overrides. The HDR image that was used is of a clear blue sky with a few clouds on the horizon and a bright, 11:40am sun (obtained from Paul Debevec's website). A second version of it was produced by downsizing and blurred it using HDR Shop (prior to me sitting at this desk - we have not yet purchased a license).



This first image is "standard" HDRI. I placed the same HDR image of the clear blue sky in both the skylight and the Reflection/refraction slots in the VRay DI Environment overrides. Ahhh, the disco sun.




This is with "convolved" HDRI. The skylight HDR image was replaced with the downsized and blurred version of the sky without changing the reflection HDR image. It's certainly nicer than the first, but the shadows are soft.




I went back to the "standard" HDRI setup and placed a disc in the scene to occlude the sun. The chrome sphere came in handy for getting the disc in the right place. It took about 8 quick renders before I had it.




And, the top view for reference. The disc is black because I disabled send and receive GI in its VRay properties.




I made the disc invisible to reflection in its Max properties and added a direct light just under it. The light has a multiplier of 1.5 and a light yellow color (255,254,22. And, yes, the light does show up in the reflection. It seems that both the direct light and the sun in the HDR reflection map are adding to the specular in this rendering. If I still had access to HDR shop I would remove the sun from the map and re-render. Though, the sun does produce a nicer reflection than the disc that we get from the direct light. So, I'll get rid of the light instead.




Here I unchecked "Specular" under the "Advanced Effects" parameter of the direct light and, while I was at it I enabled caustics. The shadow of the glass sphere in the previous image just wasn't right. The direct light has a value of 4000 for its Caustic subdivs in it's VRay light properties (select the light, right click in the viewport and choose "Vray properties"). For the caustics I used a search distance of 0.4', a max density of 0.04', and max photons=0. (For reference, my spheres have a radius of 1.5'.)

Well, I must say that this is much nicer than the convolved solution. I miss some of the subtleties inherent to pure HDRI lighting, but the ability to use caustics, the shorter render times, and the crisp shadows are hard to argue with. Thanks for pushing me in this direction cpnichols!

-Jeremy

[cpnichols]

Well done... good examples and very well illustrated.

[flipside]

There's another way to get nice caustics without using the hdri as lightsource.

Put the full res hdri only in the reflection slot. Put a light blue in the skylight color box.

Just like you placed the direct light, place a spherical vraylight in that position (small radius and far away from the scene so it will cast directional shadows). You'll need a very high multiplier for the light to reach the scene. Make it yellowish.

With this approach you have full control over the lighting color (instead of relying on the hdri map). Rendertimes will be faster because hdri lighting is always slow. You don't need to block the sun etc...

Plus, a vraylight creates GI caustics too (a max light does not, that's why you leave your hdri in the skylight slot in your example).

Another thing to realize, is that when you would take a picture of your scene, with an exposure time that makes the greeble parts look well lit, the reflections in the chrome ball will become pure white almost (because then these parts are over exposed now).

Here's what I came up with:

http://users.pandora.be/stor1/vray/sun01.jpg


Note that the caustics actually look very wrong (also in your examples). The sun would create a very tiny bright spot under the sphere in real life. Here is gets bigger, because of undersampling in the irradiance map, interpolating between nearbye samples.

If you render with QMC GI for first andsecond bounce, you can see how it 'should' look like. To have a similar result with IR map, you will need very high quality settings.

qmc gi example (region render around sphere only)

http://users.pandora.be/stor1/vray/sun02.jpg


You can see the small bright spot, fading out quickly to the sides. This is nearly impossible with IR map. Only very high quality settings will do, but that defeats the purpose of the ir map of being faster than qmc gi...

sorry for going off topic lukx!!!

[analogueWay]

Flipside, I had noticed that my reflection seemed to be too dark, but thanks for pointing it out. I placed the same HDRI (with the same exposure) in both slots, and that is what I got.

In my scene, since the sun was so intense against a clear blue sky, I was actually "stopping down" the exposure to get the greeble surface to not be blown out. Thus, the sky got dark. Most likely, this is just an issue with the HDR image that I used.

The renderings with the direct light were done to match the original all-HDRI renderings.

You also had some good comments on the caustics. This is a good reminder that irradiance maps are just a fast fake of real lighting (but, a *really* good, fast fake).

One of my favorite things about computer graphics is figuring out how to get close to realism with minimal compromises in performance (or maximum accumulation of hardware and software performance ).

-Jeremy