1080p/720p Native Recording

[gpp]

The Panasonic Manual for the HMC150 indicates that the camera does Native Recording at 1080p/24 and 720p/24. My memory is that some other products (like the HVX200) also record natively at 30p. Does the HMC150 NOT record Natively at 30p? Does it matter? By that I mean this: Panasonic introduced Native recording (as I recall) to extend the record times on P2 cards, by recording only one frame rather than two at a time. Since the HMC150 is recording at a (relatively) constant data rate, this advantage would seem irrelevant. However, does recording Natively on the 150 offer another advantage: allowing the compressor to work on fewer pixels (one frame rather than two) and thus more efficiently? The point of my question is, IF Native recording with the 150 offers the advantage described, and IF Native recording is only available on the 150 at 24p only, I would be inclined to record at 24p (rather than 30p); IF NOT, I might choose to shoot at 30p. Barry? Anyone else? Thanks.

 

[Barry_Green]

The data rate is fixed, yes... but it doesn't mean you get the benefits that you think you're getting.

Recording fewer frames makes you think that you're getting more bits per frame, and yes, that's probably true. However, it doesn't necessarily work that way. Because a long-GoP codec doesn't really encode the frames, it encodes the differences between the frames, and when you have more frames per second, the differences between the frames are smaller than when you have fewer frames per second!

So, the big benefit to "native" recording in 1080/24p is that you don't have any of the goofy pulldown nonsense to deal with. It's not about space savings or more-efficient compression, it's about the stupid pulldown. And 720/24p is the same way.

In 30p, it's actually recorded within the prevailing broadcast standard (so 1080/30p is recorded as sliced into fields in 1080/60i; 720/30p is recorded as 30 progressive frames and 30 duplicate frames in a 720/60p stream). Neither makes a hill of beans' worth of difference as far as the final video quality goes.

Progressive recording is more efficient than interlaced recording; 1080/30p is recorded as embedded within a 60i data stream but using progressive encoding (did that just blow your brain? It hurts mine...)

So -- don't worry about it. Choose the mode you want to use based on the feel it gives your footage, not over a few pixels here or there because the AVC compression is going to maximize whatever it's given to work with.

 

[Cranky]

Progressive recording is more efficient than interlaced recording; 1080/30p is recorded as embedded within a 60i data stream but using progressive encoding (did that just blow your brain? It hurts mine...)

gpp: http://en.wikipedia.org/wiki/Progressive_segmented_Frame

I think that 1080PsF30 is simpler to work with, just weave fields by setting "progressive" flag and you are done. With 720p30-in-60p one has to remove extra frames, which may or may not be simple.

 

[Barry_Green]

But you don't have to remove anything, the NLE automatically does it. It knows when it sees a 720/30p stream that there are duplicate unnecessary frame flags, so it ignores them.

 

[Cranky]

But you don't have to remove anything, the NLE automatically does it. It knows when it sees a 720/30p stream that there are duplicate unnecessary frame flags, so it ignores them.

AFAIK, Vegas keeps them by default, and you have to set a property on an event... undersample rate I believe, to avoid frame blending in 30p timeline. I found this tip here on DVXInfo when I was playing with 60p clips from the HMC150.

 

[gpp]

BARRY SAID: Because a long-GoP codec doesn't really encode the frames, it encodes the differences between the frames, and when you have more frames per second, the differences between the frames are smaller than when you have fewer frames per second!

Maybe I'm being picky, but isn't it the same in the end? To put it simply, doesn't more frames+fewer changes = less frames+more changes?

BARRY SAID: So, the big benefit to "native" recording in 1080/24p is that you don't have any of the goofy pulldown nonsense to deal with. It's not about space savings or more-efficient compression, it's about the stupid pulldown. And 720/24p is the same way.

I'm confused: I thought shooting at 24 CREATED pulldown issues, when converted to NTSC video. I thought that was a reason to shoot at 30, to AVOID pulldown.

 

[Barry_Green]

Maybe I'm being picky, but isn't it the same in the end? To put it simply, doesn't more frames+fewer changes = less frames+more changes?

Yes, that's what I was saying. It nets out to about the same, whereas the O.P. was thinking that there'd be a big savings to going with pN mode.

I'm confused: I thought shooting at 24 CREATED pulldown issues, when converted to NTSC video. I thought that was a reason to shoot at 30, to AVOID pulldown.

Not on an HMC150. Record native 24p, edit native 24p, and master it out as native 24p. Forget all about pulldown, that's old school. If your DVD player decides your footage needs pulldown, it'll add it. With a 150 you don't have to think about it or ever even know about it. That's what makes 24pN mode so valuable.

 

[Cranky]

I'm confused: I thought shooting at 24 CREATED pulldown issues, when converted to NTSC video. I thought that was a reason to shoot at 30, to AVOID pulldown.

Pulldown is added for broadcast, which is either 60i or 60p (though I've heard that some satellite channels are available in native 24p, I need to double check). How do you think 30p avoids pulldown with 60i or 60p broadcast? It does not. It uses pulldown as well, but not 2-3 pulldown as for 24p, it uses 2-2 pulldown.

But as long as you don't work on a TV station you should not care, because the camera can shoot native progressive, you can edit native progressive, you can output native progressive, and a DVD or a Blu-ray player can play native progressive.

 

[ullanta]

This also depends on how smart the codec is... and it seems to me it's pretty smart.

The P2 cams record with a constant number of bits per frame in each recording mode; there it's clear that native mode saves a lot of space, but doesn't change quality at all.

The HMC has variable-bit-rate recording with long-GOP interframe references. And if the encoder is smart enough, this means that there's really no difference between native and non-native.

Think of it this way: as the Green Barry said, the data stream primarily consists of DIFFERENCES between frames. And I'm pretty sure that the codec can indicate "no difference" very efficiently (something like a "duplicate frame" flag that takes very few bits). So 30p over 60p (in 720p) will result in a stream something like: (thousands of bytes for frame 0) (10 byte duplicate frame flag for frame 1) (thousands of bytes for frame 2) (10 byte duplicate frame flag for frame 3), etc. Extremely little data is wasted on the duplicate frames.

24p needs a more complicated scheme because in 24p over 60i, some frames get blended in the stream, which is not as easy to specify with a simple flag (and still have smooth playback of the 60i stream).. so it's worth the extra effort to have the codec support native 24p.

So, if the HMC encoder is really smart, at 30p it will allocate about twice as much data to each frame (compared to 60p), with some improvement in image quality.... but how big a factor this ends up being is still under investigation. Meaning, noone has said clearly that there's a great noticeable difference... so as has been said, it shouldn't be your main consideration in choosing a frame rate!

-B

 

[rambooc1]

AFAIK, Vegas keeps them by default, and you have to set a property on an event... undersample rate I believe, to avoid frame blending in 30p timeline. I found this tip here on DVXInfo when I was playing with 60p clips from the HMC150.

Cranky can you elaborate on that .. the bit about how Vegas handles 720p 60/30fps footage.

Thanks.