Fast motion blur / DOF

Hi Vlado and dev team.

On the same subject of “fast”, I’d like to request : Fast motion blur and DOF.
Basically what we are looking for is a not so physically accurate but reasonably fast solution to motion blur and DOF. More so to motion blur then anything else. True 3d motion blur is great in vray, but in some cases it can be both slow and memory heavy, when you need to have more then 2 geo samples.
I am wondering if there are ways to achieve a faster solution. Something like what mental ray offers with rapid motion blur. A few options for that would be great.

Thanks.

You could sort of do that if you render the static image, and then use something like camera mapping to project it on the motion-blurred image.

Best regards,
Vlado

well…that is quite a bit of work if you need to do passes and work with long complicated sequences. So that’s why we are asking for a simpler solution.

No simpler solution for now, but a lot of that can be automated. I’ll make a note to think about it though.

Best regards,
Vlado

The only times I can afford real motionblur is when I am using ambient occluion to fake my lighting ( rendertimes saved there can go to the motion blur), which brings us back to this thread: http://www.chaosgroup.com/forums/vbulletin/showthread.php?t=45161

:smile:

Hello,

well for real fast Moblur and DOF, i would do it in post. If you have a package like Nuke it works really great and is quite easy and fast.

As for fast AO, i second the need for such a shader :wink:

Not quite. Post blur is a solution…some what. Yet still it requires a fair bit of work. Think of layering hundreds of passes.
What I was thinking about was something along the lines of 2.5d blur where objects trajectories can be blurred. Something like the final DOF/Mblur.

http://www.cebas.com/products/feature.php?UD=10-7888-33-788&PID=48&FID=433

Well if you have the Z depth and velocity, it’s quite easy enough to get some “fast” dof and moblur no?

or could you export the camera to a post package like Nuke, and derive the camera’s transform data from that import, then use this to create a vector blur node or a motion blur node (or whatever it’s called can’t remember off the top of my head) and pass that data around to whatever nodes (or layers) you want?

did you guys read my post carefuly?
When dealing with hundreds of passes with overlaping geometries of all kind, this gets really complicated.