I am facing an issue and not sure how it should be. I am working on a comp and the director is complaining as he says the motion blur is not being applied, and of course, it is on. At first in my batch, after the motion blur, I was applying lens defocus, therefore the motion blur effect gets killed/reduced. Now I am doing it differently and I am applying first the lens blur and then the motion blur and of course, MB gets better but reduces the lens blur fx.
As per the “logik” I feel is closer to reality first lens then shutter… How would you do it? What’s correct? LB+MB or MB+LB?
Ye agreed, mainly because the other way around has never given me any good results not to mention I’ve no idea how to do it ,
also I think we had a topic about this before? The dream would be having one node doing both at the same time. Surely there are some learnings on how 3D software handles this problem (ofcourse I don’t think anyone renders with DoF)
Typically in these situations if the director isn’t quite happy then just add more motion blur.
Also don’t forget to be doing all this work in linear so that all your hard work is being executed correctly. A very in-tune Director to motion blur and depth of field may be responding to 3d blurs in the focus is happening on Rec709 footage.
Usually I like to try to think about how it would happen if it were real. In this case, the lens would be applying the blur optically before the information really enters the camera. The motion blur will come from the relationship of the camera’s shutter speed to the observed movement of the subject. That motion blur would be applied internally when the information hits the shutter and is recorded to the chip. Given that, the answer should always be lens blur then motion blur. (which is sometimes a pain in the ass to setup in your pipeline… which is how you know its the right way to do things)
I have been promoting the grade at the end of the job on all of our big CG compositing jobs. If that is what you mean.
If I need to colour correct any CG I will do that on my unPremult then multiply before I blur RGB and A seperate. Then I divide back again before I comp.
I’ve just been using the physical bokeh matchbox node in action on a project and it does an awesome job with lens blur - since it’s in action I’m assuming you could have motion blur turned on as well? The only annoying part is that with it hanging off the camera it affects everything in the scene so you can’t be selective. Also it seems to work better on Mac than Linux.
The difference is always gonna really stick out in the highlights. That’s just cameras and exposures and sensors. Any blur, be it Gaussian or defocus is going to effect your highlights. It makes sense in the same way that you color correct before applying blurs (as @GPM was talking about).
And it’s gone through several attempts of various names, but unfortunately Action still does not have a dynamic+robust selective system in its multilayered 3D space. I thought they had hit the holy grail about 6 years ago with Camera FX. Then I saw how limited it was.
It’s still the unicorn. Some features here and there kind of hint at what it could be, but still no dice on a comprehensive solution.
,