|
|
All the news now and in your digital future
jumpy footage
Posted by Eddietor
|
Re: jumpy footage March 01, 2009 07:20AM |
You should be talking about the HPX 170.
Are you shooting 24PN? When you ingest in L&T, do you have "remove duplicate frames" selected?
www.strypesinpost.com
Are you shooting 24PN? When you ingest in L&T, do you have "remove duplicate frames" selected?
www.strypesinpost.com
|
Re: jumpy footage March 01, 2009 10:35AM |
Quote
On the computer monitor you can see the interlacing
There is no interlacing in 24p material. ("There's no crying in baseball!"
What you're seeing is almost certainly a 3:2 pulldown pattern inserted by the camera to get 24p material to fit in a 60i data stream.
Conventional wisdom is you're better off by far shooting in the "PA" mode (which is "advanced pulldown" abbreviated sideways), because Final Cut can remove the pulldown automatically. FCP can't deal with traditional 3:2 pulldown automatically. Compressor can, if you use the "reverse telecine" deinterlacing option when you convert to an intermediate format like ProRes.
|
Re: jumpy footage March 01, 2009 11:01AM |
>FCP can't deal with traditional 3:2 pulldown automatically.
That's because you lose a generation on some frames when you remove the pulldown. 24PA is a better choice as FCP can remove the pulldown on ingest (remove duplicate frames). 24PN is the native frame rate that the footage is stored in..
www.strypesinpost.com
That's because you lose a generation on some frames when you remove the pulldown. 24PA is a better choice as FCP can remove the pulldown on ingest (remove duplicate frames). 24PN is the native frame rate that the footage is stored in..
www.strypesinpost.com
|
Re: jumpy footage March 01, 2009 11:20AM |
Eh, sort of. It's more about efficiency of the first round of compression, inside the camera. The 2:3:3:2 pulldown pattern records each 24p frame as a pair of fields, while 3:2 spreads one out of every four 24p frames across two pairs of fields. Since DV compresses each field pair spatially at a fixed data rate, it tends to compress the jitter frames less efficiently.
For every four frames of 24p footage ? call them A, B, C and D ? the camera splits them into four pairs of fields: AA, BB, CC and DD. It then arranges arranges the fields like so: AA BBB CC DDD. That's where the name comes from: two, three, two, three. (Why it's called "3:2" when it really goes "2:3" is a mystery for the ages.) When those fields are written out to tape in pairs, you end up with the classic pulldown cadence AA BB BC CD DD. Notice that the "C" frame is never written out as a full frame, like A, B and D are; it's only written out as half of two jitter frames. Since DV encodes jitter frames less efficiently, the conventional wisdom is that once the pulldown is removed, every C frame looks slightly worse than the other three-out-of-four frames.
In practice, I've never actually seen it.
Advanced pulldown arranges the fields in a different order. Instead of a 2-3 pattern and AA BB BC CD DD, you get a 2-3-3-2 pattern and AA BB BC CC DD. The plus is that instead of having to rearrange fields to get your 24p back, you can just drop every third frame entirely; that's what Final Cut does when it removes advanced pulldown. The minus is that playing back 2:3:3:2 on a 60i display looks all wrong. It stutters. Which is why 3:2 is still the universal pulldown cadence for showing 24p material on television. You can shoot in 2:3:3:2, but only if you're going to remove the extra frames and edit in 24p, then insert 3:2 for broadcast.
In Final Cut, both 3:2 and 2:3:3:2 look wrong when viewed in the canvas, because computer screens are not televisions. Arguably, 2:3:3:2 looks more wrong, but that might just be my eyes.
I'm pretty confident that the original problem here is related to pulldown, either 3:2 or 2:3:3:2, that should have been removed but wasn't.
For every four frames of 24p footage ? call them A, B, C and D ? the camera splits them into four pairs of fields: AA, BB, CC and DD. It then arranges arranges the fields like so: AA BBB CC DDD. That's where the name comes from: two, three, two, three. (Why it's called "3:2" when it really goes "2:3" is a mystery for the ages.) When those fields are written out to tape in pairs, you end up with the classic pulldown cadence AA BB BC CD DD. Notice that the "C" frame is never written out as a full frame, like A, B and D are; it's only written out as half of two jitter frames. Since DV encodes jitter frames less efficiently, the conventional wisdom is that once the pulldown is removed, every C frame looks slightly worse than the other three-out-of-four frames.
In practice, I've never actually seen it.
Advanced pulldown arranges the fields in a different order. Instead of a 2-3 pattern and AA BB BC CD DD, you get a 2-3-3-2 pattern and AA BB BC CC DD. The plus is that instead of having to rearrange fields to get your 24p back, you can just drop every third frame entirely; that's what Final Cut does when it removes advanced pulldown. The minus is that playing back 2:3:3:2 on a 60i display looks all wrong. It stutters. Which is why 3:2 is still the universal pulldown cadence for showing 24p material on television. You can shoot in 2:3:3:2, but only if you're going to remove the extra frames and edit in 24p, then insert 3:2 for broadcast.
In Final Cut, both 3:2 and 2:3:3:2 look wrong when viewed in the canvas, because computer screens are not televisions. Arguably, 2:3:3:2 looks more wrong, but that might just be my eyes.
I'm pretty confident that the original problem here is related to pulldown, either 3:2 or 2:3:3:2, that should have been removed but wasn't.
|
Re: jumpy footage March 01, 2009 11:39PM |
I'm back, and it turns out I do have "remove duplicate frames" selected. Do I want that or no?
Here's a screen shot of my hand waving (forgive the non-white balance):
What's going on here?
Here's a screen shot of my hand waving (forgive the non-white balance):
What's going on here?
Sorry, only registered users may post in this forum.
