Hi, I have this interior image in which i try use GI+FG. I have this problem with MR sun and seems to be from its photon radius. It lets light inside the corners of the room. I had to decrease the radius to unrealistic value comparing to the room size because before light was all over the inside edges of the room. Also some white dots appear on the left side of the image, like reflections, and I ve tried increasing a lot of things but didnt seem to disappear. The material has 0 reflectivity. I use one skylight with its photon multiplier 50 and MR sun. I m very new to GI so i might do everything wrong..
Help me out here, thanx :)

3ds max 2010 x64 sp1 | Mental Ray

i7 920 @ 2.67GHz / ASUS P6T deluxe / 12GB RAM Corsair dominator DDR3 / ATI Radeon Sapphire 6950 HD 2Gb ram / Velociraptor 150GB 10000 rpm / OCZ vertex 2 SSD 60GB

	Attachment

	Attachment

I had to decrease the radius to unrealistic value comparing to the room size…

What do you consider “unrealistic”?  For a small room like this, I’d use 3” - 6” (7.5 cm - 15 cm) for the photon radius. 

One trick I’ve used several times to control photon bleed is to build a plane outside that completely blocks the photons from hitting the wall and coming through.

3DS Max Design 2011 64-bit - Advantage Pack
Dell Precision T5500, Dual Six Core Xeon X5650 @ 2.67GHz, nVidia Quadro 5000, 24 GB RAM, Win 7 Enterprise 64-bit
Minneapolis, MN, USA

Thanks Jeff. Im sorry my english are bad and many times I cant explain proper what I need to say. Yes, I understand that photon radius is different than sun’s photon target. One last question: In this case where sun doesnt hit the opening window and hits the west wall, and room is mainly indirect illuminated, how much approximately do I set the sun’s photon target radius?
Thanks again :)

3ds max 2010 x64 sp1 | Mental Ray

i7 920 @ 2.67GHz / ASUS P6T deluxe / 12GB RAM Corsair dominator DDR3 / ATI Radeon Sapphire 6950 HD 2Gb ram / Velociraptor 150GB 10000 rpm / OCZ vertex 2 SSD 60GB

If the sun doesn’t hit an open window the target photon radius actually has no sense Thanulee. The sun photons now just hit the wall and bounce somewhere in the Universe, you are using only indirect light through the Portal lights. It could be set to “0”.
If you want to have direct sun photons into the room position the sun directly over the window and make the target radius a bit bigger than the opening, that is the normal workflow. It is not bad to create a plane object beneath the room, as well. It helps to catch the bounced sun photons and this way make the rendering faster.
What about photon radius; let MR evaluate it alone leaving the maximum sample radius box unchecked. This done take a look into the Render Message dialog and see what value has the computed “globil radius”, it will be 1/10 of the overall radius of your scene.
This is usually a good start point for this particular case; you will go down for the “beauty pass”.
But first of all you need enough photons “stored” on the geometry; 40000 per light is usually not enough; you are having just three light in the scene. Try with 500000 as a beginning. Remember having enough photons stored you can play with all the other setting as much as you need without damaging the solution. (No need to render another photon map.) And keep in mind there are no “magic settings” you will get the right ones through testing.

ivan

… What about max num. photons per sample Ivan? Do i mess up with it?

The Maximum Sampling radius value is actually the effective size of each photon.
The Maximum Num. Photon per Sample defines the number of photons collected in a particular sample.
In an area of dense photon coverage MR will collect photons in an ever-increasing radius until the “maximum num. photons per sample” value is reached. That could happen actually long before the defined “maximum sampling radius”, it depends from the particular case.
On the other side in regions of low photons density MR will stop gathering photons when
the “maximum sampling radius” is reached and this could happen long before the “maximum num. photons per sample” value. The process is an adaptive one; I hope my explanation is clear enough.
In one way or another the Samples (effective photons) must be overlapping, or you need enough of them stored into the scene, keep that always in mind.
It is clear that when MR has to gather more photons per sample it needs more render time isn’t it?
You need to decide what you are after; a fast visually acceptable result, or physically correct one and in the same time really beautiful image. One cannot get the two of them in one solution. Quality has a price you know.

What about the tutorial, it is an useful and very instructive one but in my humble opinion following a “physically correct” workflow is a better way to go; actually that is the philosophy of Mental Ray.

ivan