Any reason to shoot in 60p instead of 24p?

Dave Haynie · March 25th, 2011, 11:54 AM

Well, you do have twice the information coming in... you just toss out more of it. And yeah, that may be quite acceptable.

One thing to keep in mind: 60p is by far the favorite for fast motion, action, sports, etc. So much so that it was the ESPN/ABC (maybe Disney, too, I'm not sure when Disney bought ABC) lobby that got 720/60p into the "Grand Alliance" for HDTV.

So for one, the GOP rate is the same... so if your frame rate is 2.5x faster, you have 2.5x as many I-Frames. And those, you really want as uncompressed as possible. This is where 24p can produce better video: you have 2.5x the bit budget for each I-Frame.

But then consider P and B frames. For those not familar with AVC and other MPEG algorithms, it works like so. The I-Frame is essentially a JPEG. When the next video frame is captured, the camera uses sophisticated motion estimation algorithms to figure out where things in your first frame have moved, if they've moved at all. This results in a set of very compact motion estimation vectors. You then apply those vectors to the first image, and produce a difference fame... basically, what didn't the motion estimation get correct. This is compressed as part of the P or B frame, but there's not much of a bit budget for it.

So back to 60p... all things being equal, you'll have 1/2.5x as much motion between any frames in 60p as 24p. So there's actually a much better chance the motion vectors work well, and there could be substantially lower differences between frames. This becomes really apparent when MPEG starts to fail. If you ever see really fast motion in your videos, you probably start to see "pixelization" and other garbage. Those aren't real pixels, but emergent blocks... the video is being over compressed, so the blocks created for the DCT (JPEG-like) compression start to be visible. In a professional DVD or Blu-ray mastering process, the mastering engineer would speed up the bitrate in such areas if possible... if not, they'd apply a global low-pass filter (eg, blur), which of course blurs the image, but also reduces the detail, and thus, makes the MPEG work better and hopefully not show off the DCT blocks in fast motion. But I digress.

And then there's interlaced video. The encoder breaks this up into separate fields and encodes each one separately. This means that your 4:2:0 color decimation is really run across four lines rather than two, and of course, rather than 16x16 DCT blocks, you have two interlaced 16x16 blocks... so you're looking over 32 pixels, effectively... much more likely that some compression artifacts show up here. This is also a shortcoming of cameras, like the TM700, that pulldown real 24p into fake 60i for recording.. there's absolutely no good reason for that in modern times, other than perhaps to intentionally crapify it and make you want to spend the extra crash for a more professional model. Honestly, in most video, you don't any single difference all at once, but when you stack a few things: wrong frame rate, interlacing, lower bitrate, etc. it can make a noticeable overall difference.

Phil Lee · March 26th, 2011, 02:07 AM

Hi

Quote:

Well, you do have twice the information coming in... you just toss out more of it. And yeah, that may be quite acceptable.

If you think about it pragmatically, all 60i actually is, is 60p with half the information thrown away immediately by interlacing which is the crudest video compression system dating back to before computers existed.

So the argument more is thrown away with 60p compared to 60i, is really the other way around, and what you end up with is more video information recorded in 60p footage, which is why we like it so much :)

Regards

Phil

Les Wilson · March 26th, 2011, 06:24 AM

Epilog: I shot with the TM-700 in 60p and converted to Prores LT in post. Daylight footage cuts well with the EX1r. I always had a B+W polarizer on the TM-700 which reduces light by about 1 stop and didn't notice any need for an ND. I mostly shot with it one handed and ran it in manual mode ... riding the iris with forefinger and zoom with my thumb. No complaints for a $750 camera although I despise the way that lens ring mode is selected. Sony has done it right on the NX70 albeit at a much higher price point and larger form factor.

March 25th, 2011, 11:54 AM	#16
Dave Haynie Regular Crew Join Date: Mar 2004 Location: Upper Pittsgrove, NJ Posts: 95	Re: Any reason to shoot in 60p instead of 24p? Well, you do have twice the information coming in... you just toss out more of it. And yeah, that may be quite acceptable. One thing to keep in mind: 60p is by far the favorite for fast motion, action, sports, etc. So much so that it was the ESPN/ABC (maybe Disney, too, I'm not sure when Disney bought ABC) lobby that got 720/60p into the "Grand Alliance" for HDTV. So for one, the GOP rate is the same... so if your frame rate is 2.5x faster, you have 2.5x as many I-Frames. And those, you really want as uncompressed as possible. This is where 24p can produce better video: you have 2.5x the bit budget for each I-Frame. But then consider P and B frames. For those not familar with AVC and other MPEG algorithms, it works like so. The I-Frame is essentially a JPEG. When the next video frame is captured, the camera uses sophisticated motion estimation algorithms to figure out where things in your first frame have moved, if they've moved at all. This results in a set of very compact motion estimation vectors. You then apply those vectors to the first image, and produce a difference fame... basically, what didn't the motion estimation get correct. This is compressed as part of the P or B frame, but there's not much of a bit budget for it. So back to 60p... all things being equal, you'll have 1/2.5x as much motion between any frames in 60p as 24p. So there's actually a much better chance the motion vectors work well, and there could be substantially lower differences between frames. This becomes really apparent when MPEG starts to fail. If you ever see really fast motion in your videos, you probably start to see "pixelization" and other garbage. Those aren't real pixels, but emergent blocks... the video is being over compressed, so the blocks created for the DCT (JPEG-like) compression start to be visible. In a professional DVD or Blu-ray mastering process, the mastering engineer would speed up the bitrate in such areas if possible... if not, they'd apply a global low-pass filter (eg, blur), which of course blurs the image, but also reduces the detail, and thus, makes the MPEG work better and hopefully not show off the DCT blocks in fast motion. But I digress. And then there's interlaced video. The encoder breaks this up into separate fields and encodes each one separately. This means that your 4:2:0 color decimation is really run across four lines rather than two, and of course, rather than 16x16 DCT blocks, you have two interlaced 16x16 blocks... so you're looking over 32 pixels, effectively... much more likely that some compression artifacts show up here. This is also a shortcoming of cameras, like the TM700, that pulldown real 24p into fake 60i for recording.. there's absolutely no good reason for that in modern times, other than perhaps to intentionally crapify it and make you want to spend the extra crash for a more professional model. Honestly, in most video, you don't any single difference all at once, but when you stack a few things: wrong frame rate, interlacing, lower bitrate, etc. it can make a noticeable overall difference. __________________ --Dave

March 26th, 2011, 06:24 AM	#18
Les Wilson Inner Circle Join Date: Jan 2004 Location: Boca Raton, FL Posts: 3,014	Re: Any reason to shoot in 60p instead of 24p? Epilog: I shot with the TM-700 in 60p and converted to Prores LT in post. Daylight footage cuts well with the EX1r. I always had a B+W polarizer on the TM-700 which reduces light by about 1 stop and didn't notice any need for an ND. I mostly shot with it one handed and ran it in manual mode ... riding the iris with forefinger and zoom with my thumb. No complaints for a $750 camera although I despise the way that lens ring mode is selected. Sony has done it right on the NX70 albeit at a much higher price point and larger form factor.

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search