Re: Ray Length "Depth Pass" -- DOF in Reflections

[Christian Götzinger <cg(at)pension-goetzinger.de>]

Hi Brett,

I'm not sure I'm understanding you correctly. Why do you bring up glossy reflections? I'm really only interested in depth of field simulation, glossy reflections are working nicely :)

I can show you better examples later if necessary, but please go to:
http://forums.cgsociety.org/showthread.php?t=328799&highlight=jewelry

Look at the image in the very first post. It was either rendered by Maxwell or by VRay, I'm not sure. Pay special attention to the differently shaped highlights in the back of the gold ring. I have not yet found a SINGLE WAY to do this in mental ray. This is how depth of field in reflections should look like but mental ray can't do it. If you fake it with for example Frischluft Lenscare, you'll be amazed at how much worse and how much flatter the depth of field is going to look like. It'll look completely fake because the fact that these reflection rays have travelled a *long way* won't be taken into account.

Like I said, I can give you many more examples of reflective objects where fake techniques like Lenscare won't do if you're going for realism.

I may not have explained what I'm trying to do correctly. Let me try again. Basically I want a grayscale "depth pass" that takes into account only the reflections. And the gray levels should be based on the total distance from the camera to the reflective object to the reflected object. Does that make sense to anyone?

Cheers



B Miller wrote:

Hi Christian,

First of all, the depth of field effect present in the lens is very different from the light scattering effect that happens with glossy reflection. With a glossy reflection, the focus located on the surface of the reflective object, which is not usually the case with a lens.

For lens DOF, we use the frischluft lenscare plugin for after effects. It's beautiful and fast.

For glossy reflection, I would maintain that there's no better substitute for raytracing scattered samples using the reflection diffuse node. Just be sure to tweak the BSP tree and set a reasonable maximum depth and sample count. And run a fresnel coefficient into the reflection power.

my 2c,

Brett

On 8/9/06, Christian Götzinger <cg(at)pension-goetzinger.de> wrote:

Hi List,

I hate the fact that there's no proper way to calculate true depth of
field in mental ray. Mental ray's native DOF shader is absolutely
ridiculous. Way too slow and it doesn't even simulate the aperture shape.
Also, I had to find out recently that even X-DOF is fake because
reflections are not blurred correctly based on ray length - instead the
z buffer is used. For reflective objects this ALWAYS looks totally wrong
and fake.

Now, I was thinking that if you rendered a diffuse pass and a reflection
pass and applied DOF in post independently to both passes then all you'd
need is an additional "depth pass" that's based on ray length. You could
apply DOF to the reflection pass based on this ray length pass and add
the result on top of the diffuse (which receives independent DOF).
Of course this is still not ideal because you have to guess the depth of
field amount for every shot instead of basing it on real world values,
but doing depth of field in post seems like the only solution to getting
good depth of field in reflections when dealing with mental ray.

Is this thinking flawed? Is there any way to do this? I'd appreciate any
help on this because lack of realistic depth of field is driving me crazy.

Cheers,
Chris


---
Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in body:
unsubscribe xsi