Personally, I'd probably do 3D motion blur or the mirrored camera in practice. But for the sake of completeness of the thread, it is possible to make a shader that renders reflected motion vectors for a static reflector.The trick is to compute the vector not from the actual intersection, but from the intersection reflected over the plane of the previous intersection.
Of course, there are all sorts of gotchas, like the fact that a moving reflector may completely change the motion path in camera space. And you it can get complicated dealing with "nested" aovs like this. So I don't recommend going down this path. On Saturday, September 28, 2013, Elias Ericsson Rydberg wrote: > That seems about right, I think I was using 3delight at the time. A > renderer that likes doing cheap tricks. > > 28 sep 2013 kl. 23:26 skrev "Nathan Rusch" > <[email protected]<javascript:_e({}, 'cvml', > '[email protected]');> > >: > > In broad terms, moving reflective surfaces and motion blur generally > aren’t a problem for raytracing renderers, because there aren’t relying on > tricks or workarounds. > > -Nathan > > > *From:* Elias Ericsson Rydberg > *Sent:* Saturday, September 28, 2013 1:36 PM > *To:* Nuke user discussion > *Subject:* Re: [Nuke-users] motion blur in reflections > > I'll second Nathan. > > Doing motion blur in 3D seems like your best bet. I know that motion blur > in reflections in some renders can be tricky. A great example is a spinning > chrome sphere, which shouldn't make the reflections become blurry. Sure the > surface being shaded is moving but should not affect the reflections in any > way. Possibly it was 3delight where you had to be cautious about this, > however, doing it in 3D will save you a very intricate comp. Specially if > there's a lot of different moving parts that are reflector. Eg. the > internal gears of a watch. > > /Elias > > 28 sep 2013 kl. 21:39 skrev "Nathan Rusch" <[email protected]>: > > I would say definitely just render your 3D with motion blur. Given the > nature of your scene, the time it will take you to get a comp solution > working that’s even close to “good enough” will more than cancel out the > time you could save by avoiding it in 3D, and doing it right will always > look better. > > -Nathan > > > *From:* Ryan O'Phelan > *Sent:* Saturday, September 28, 2013 7:55 AM > *To:* Nuke user discussion > *Subject:* Re: [Nuke-users] motion blur in reflections > > > The mirrored camera is the best choice because you can maintain a true > pass system in the reflection(you will have diffuse, gi, etc.) , which > you can't do otherwise. > > If you are feeling adventurous, you could try projecting a velocity pass > on your geo, from a previous render, and then rendering the reflection. > Your vertical vector will be inverted, but that's easy to fix in comp. > > Personally, I would stick to motion blur in camera. Lazy, I know. > > Ryan > On Sep 28, 2013 2:50 AM, "matt estela" <[email protected]> wrote: > > I'm no v-ray expert either, but knowing that vray can be treated as a > pure path-tracer like arnold, and getting motion blurred reflections from a > pathtracer is down to spreading your reflection samples over time, I'm sure > v-ray could do this happily, albeit slowly. Did a quick google, found some > images that imply it can do it fine: > > http://www.cg-blog.com/index.php/2013/04/11/vray-motion-blur.htm > > (the first overly-streaky motion blurred i > >
_______________________________________________ Nuke-users mailing list [email protected], http://forums.thefoundry.co.uk/ http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users
