Crisp is not the word that I would use. Because of the extremely shallow DOF the baby's arms are out of focus half the time. It's a very nice video but cannot be held up as a demonstration of a 'perfect' 30p>24p workflow. It's a series of static tableaux with very little movement. Each shot is beautifully framed & would make a great still photograph but the DOF is very shallow so half the screen at least is blurred. He doesn't even pull focus in any of the shots. The sort of shots that any frame-rate conversion struggles with are not present e.g. if the baby & background had been all in focus & there were a lamp post or similar behind.

I am curious how those who can detect dropped frame feel about this conversion:

http://www.dvinfo.net/conf/attachmen...stepmethod.wmv

Chris, there is definitely no feel of dropped frames. But this conversion has the opposite problem: many or most frames look like blended frames. Played full speed, movements don't look so much blurred as lost, blended with the background. It doesn't look real.

Having said that, the video looks as good as anything I've been able to achieve in 24p with the 30p 5D2 footage.

Nigel makes a great point about the shallow DOF in the Cherry Blossom video, but I'd love for him to look this time at the pottery wheel shot. Looks like a pretty high shutter speed and to me every frame looks like a real frame, not an interpolated one. Nigel, is there something specific in this shot you could point to that is a giveaway that it's 30p footage?

The reason I asked the question is this. I think shutter speed is an issue in the conversion process. That film was shot at 30 fps, and then rendered to this 24p version by a two step process similar to the one Phil Bloom described, only in Vegas. Now this may be an extreme example, it seems to me that motion blur is good to have in a 30p to 24p conversion. Maybe shutter speed no higher that 1/50th should be used.

No, it looks fine. My point is that none of this footage will show up problems with the 30p>25p conversion.

From my own experiments & those of others I know the sort of shot that causes problems with a 30p>24/25p conversion it is when something like an arm or leg moves fast in front of another object like a lamp post or another person when both are in focus. One or other object will get distorted. There are no shots like that in Philip's film.

There is no stuttering or other symptoms of dropped frames but I see extreme motion blur in this clip. I agree, it doesn't look real.

This depends on the technique. If you are shooting a waterfall, you can use a long shutter and convert with frame blending. On the other hand, some of the most advanced algorithms do well on hard edges, but poorly on motion blur. There's no one answer. It really depends on the conversion technique used.

Also, if you use lots of motion blur (say 1/40 or 1/30), then the end result won't have a normal 24p 180 degree look.

I think 1/50 is good for when using frame blending. Frame drop is simply problematic. Advanced algorithms might be best at 1/80 - unless that puts you in a flicker situation.