Well, I'm not even trying to launch a discussion about Maxwell here.
Let's come back to my initial question because I still think it might
be a good approach to convincingly fake DOF in post.
There must be a way to let mental ray translate reflection ray lengths
into a grayscale image, no? Or am I making no sense?
Ray 1 travelled infinetly far (hit the environment) -> make that
pixel black
Ray 2 travelled 100 scene units (this is the longest ray) -> make
that pixel black also
Ray 3 travelled 80 scene units -> make that pixel a dark gray
Ray 4 travelled 50 scene units (half as far as the longest ray) ->
make that pixel 50% gray
Ray 5 travelled 20 scene units -> make that pixel a light gray
etc. etc.
Gent Krasniqi schrieb:
well
md_aperture gives convincing results if you use enough samples and if
you're willing to accept the rendertimes. Let's not forget that with
maxwell you're looking at ~ 20 hours (on an up-to date system) for the
noise to clear up on reflective surfaces such as in this case.
This is a typical scene where a renderer like Maxwell will turn out
largely 'undisputed' superior renders, because it is what's it
advertised to be.. a simulator. But you'll have to pay in someway with
every renderer, in maxwell's case its rendertime.
Though I think that for fields such as product visualization etc, that
require highly realistic stills,
it probably is the best choice.
On 8/14/06, Christian Götzinger <cg(at)pension-goetzinger.de>
wrote:
Hi,
I didn't try md_aperture, but last time I checked it didn't really give
fully convincing results. I might want to give it another try, though I
think it may be too slow.
As for the reflections in the rings - I saw your image in the CGTalk
thread. If you compare that to ThomasAn's image directly above it
becomes obvious that the reflections in the two rings are not blurred
because of blurry reflections. You can clearly see out-of-focus
highlights that are caused by the depth of field. These are missing in
your 2D solution which is why your version looks flat.
Furthermore, the challenge rules state that both rings are supposed to
have "high gloss reflections" - only the main bodies of the coins are
supposed to have very blurry reflections. But as you have seen, even by
bypassing the rules and making the reflections blurry you cannot
simulate the rich look of out-of-focus highlights.
Gent Krasniqi wrote:
> Did you try md_aperture? There is a spdl somewhere at
cgtalk. It's
> still slow (like 3d dof in 'most' renderers) but has more control
like
> aperture shape etc.
> BTW, alot of the 'sparkle' in the reflection of that gold ring is
> caused by the gold ring not being perfectly reflective (alittle
> roughness) and high gloss. Doing a render of that scene right now
and
> will post a mental ray result here.
Apparently it originates from Maxwell Render. If I recall correctly
however, I've seen similar quality come out of VRay in this thread.
I don't want to majorly dwell on this particular image however, as it
was just supposed to be one example for an area where I can't seem to
get convincing results out of mental ray.
By the way, I did try X-DOF on this scene back in the day, and the
results were absolutely horrible, they screamed "post DOF".
This is especially annoying because X-DOF's Lens Shader seems to
calculate everything correctly (the reflections become way more
defocussed than the ring surfaces), however the X-DOF Output Shader -
which is supposed to wrap everything up - doesn't care about this and
incorrectly adds the final blur based on Z-depth, not on ray length.
Steven Caron wrote:
> hey christian what renderer did this image originate from? just
curious...
>
> steven
---
Unsubscribe? Mail
Majordomo(at)Softimage.COM with the following text in body:
unsubscribe xsi
|
