shader for this Motion vector?

[kensonuken]

http://www.alamaison.fr/3d/lm_2DMV/lm_2DMV_ref.htm

I tried everywhere but couldnt find the renderman shader for this effect..

[Jason]

http://www.alamaison.fr/3d/lm_2DMV/lm_2DMV_ref.htm

I tried everywhere but couldnt find the renderman shader for this effect..

You want the Renderman shader to translate to Mantra? Or the Renderman shader to use with Renderman?

Mantra info, actually your own post: http://forums.odforce.net/index.php?showtopic=7656

[kensonuken]

but is that shader used in production? is it good enough?

[Jason]

but is that shader used in production? is it good enough?

Have you tried it out? It'll work fine. Ours is just slightly different, mostly due to features and quirks for our compositing software. This shader should provide you good data, but you'll have to be aware of whether your vector-blur operation wants unpremultiplied vectors or not as this will affect your ROP's pixel filter settings. This is something you'll have to confirm for yourself.

We've also put some of this functionality into a VOP too, so it can be easily added to an existing VOPnet.

[kensonuken]

whether your vector-blur operation wants unpremultiplied vectors or not as this will affect your ROP's pixel filter settings. This is something you'll have to confirm for yourself.

I didnt get that how does it affect the pixel filter settings...

I am putting arround 2.5 x 2.5 sinc

can u explain in detail with some example...

[Jason]

You should either probably be using a "minmax/min" pixel filter and "closest" sample filter - for unfiltered, closest sample information. Either that or probably use the default "gaussian 2 2" and "opacity" filtering and then unpremultiply the result in composite before you use it.

Look around and read up on Pixel- and Sample-Filters - and read up what your vector-blur operation expects. I don't have time to put an example together; there is plenty of information out there and you can always experiment a bit yourself; don't just trust any old solution. In any case you should be thoroughly testing the limitations of the vector-blur approach.. there are some cases where it works, and some cases where it breaks down. And those cases vary a little depending on the combination of software you're using. Put together a test-suite and compare the results to highly sampled true 3D motion-blur.