New CG Motion blur estimation

[aweidenhammer]

http://web.mit.edu/newsoffice/2011/f...hics-0610.html

dunno if this is just a super-controlled result without hope of scaling for production, or if it is actually killer... but worth a mention!

Andrew

[vlado]

Hehe, let's see the papers first

Best regards,
Vlado

[instinct]

Here it is : http://groups.csail.mit.edu/graphics/tlfr/

It is based on anisotropic reconstruction techniques, but claimed to work in situations that previous technique fails.

Regards,
Thorsten

[aweidenhammer]

Thanks for the link. That certainly sounds great to my non-mathematic brain. Whaddaya think Vlado? Rubbish? Inspiring?

[vlado]

Well, it has the potential to work well for relatively diffuse surfaces. I would be curious to see how it will handle mirror- and glass-like materials and very complex occlusion scenarios (e.g. motion-blurred hair).

Anyways, interesting approach. I'm wondering if the same can't be done in a simpler way though (thinking about our shademap implementation).

Best regards,
Vlado

[Neilg]

'That's not a knife...'