Native 24P/25p settings/problems discussion

With 24-over-60i I have increased the bit rate- Trying to find settings that will go the highest.
Click to expand...


I'm right there with ya, Pappas. I have been doing the exact same thing but, instead of the star chart, I've been using this green leafy tree behind my back porch. I finally got a Sandisk Extreme class 10 16gig so, tomorrow I'll be seeing how far I can stretch the bitrates without getting write errors in a span of 3 minutes of filming. That's my cutoff point. If I can record at least 3 minutes with no errors, than those bitrate settings pass. I was using a class 10 Patriot LX card and I'm hoping I'll be able to set the bitrate settings higher with the Sandisk card.
 
i believe the 24pn was very reliable when it wasn't playable in camera. i hope tester can look into this as the previous 24p were rendering files at 47 and seemed stable, however, when i had the camera i didn't point it at a wall of detail to try and make it fail.
 
c.cook88 said:
Pappas... any idea where I can learn about the GOP setting... all i've been able to find is that it stands for "group of pictures"

p.s. thanks for all your input man.... and the new "star" chart trips me right out haha
Click to expand...

Good question- I don't have any links to GOP. 15GOP is what you want IMO. It works for most of the industry and and it gives you the right balance for 24-over-60i.

Here's a quote from Convergent Design : ""For reference, I-Frame compression only considers spatial redundancies (within a given frame), while the more sophisticated MPEG2 Long-GOP considers both spatial and temporal redundancies (frame to frame).Comprehensive testing has shown that Long-GOP is typically 2 to 3 times more efficient than I-Frame-Only. So 180Mbps Long-GOP is roughly equivalent to 400 Mbps I-Frame, thus providing the highest possible quality from the nanoFlash. """

Now if you just take the basic numbers 50Mbits long gop is like 133 Mbps Iframe ( now that's a ballpark and not exact- however even if with it's in a + or even if minus 30, that is still amazing.. This is why I'm a Long GOP convert now. You just get so much more for so much less. It's efficient- and at even higher Mbits- it pays off. Now I could be wrong; and Barry could step in and set the the numbers straight.



rambooc1 said:
Yeah that new Pappas chart is even giving my eyes a workout........... LASIK SURGERY bill coming your way Pappas ....hahaha
Click to expand...

LOL-


BTW- Anyone know what the Mbits allocation per frame or GOP sequence gain we get going to 24PN vs 60i for the same settings? I know when the 5D went from 30P to 24P that the codec went up in mbits. Going from 24-over-60i to 24Frames is 36 less video fields to deal with.


Here is a part of a frame today when I was testing 24-over-60i-.... When shooting 24pn, this does not exists. It was -2 on all, and processed in the latest NeoScene with GH1 fix chroma selected.
Picture51-1.jpg



Pappas
 
Last edited:
Are you talking about the general haze there pappas? I'm an untrained eye! Sorry!

I wouldn't say that's mud?
 
Here's another example, the reds aren't the same since I shot these on different days using a different white balance, but it shows how red edges in 60i have a more jagged look while those in native 24p are rendered more cleanly (but still not totally smooth).

60i:
ixoraGH1closeup60i.jpg


24p:
ixoraGH1closeup24p.jpg
 
interesting. I don't see that red chroma problem with my camera and I'm shooting 24p/60i.

I also did some testing of GOP lengths and I get a range of 12-17 that is somewhat reliable. Outside of that range and either the camera locks up or the video has horrible macroblocking, or both.
 
Svart said:
interesting. I don't see that red chroma problem with my camera and I'm shooting 24p/60i.
Click to expand...

It's 101 basics- Your camera does do it; and all 4:2:0/interlaced 60i/24 in combination make it worse.

If I had your camera I would show you it to you- and then you wouldn't like your camera anymore- Just kidding.


c.cook88 said:
Are you talking about the general haze there pappas? I'm an untrained eye! Sorry!

I wouldn't say that's mud?
Click to expand...

That is a color sampling issue I'm pointing out that is in 24/60i, and greatly reduced when using 24PN.



Pappas
 
Yes, what you're pointing out is exactly what would be expected -- the chroma encoding in 4:2:0 is using interlaced encoding when using 24p-within-60i, and it's using the much-superior progressive encoding when using 24p Native files.
 
Yes Barry, but I would still like to see the difference between 50i and 25p. I suppose that in this case the difference would be noticeably smaller. Namely 24p should be coded into 2 or 3 60i video fields, while 25p is coded into 2 50i fields. Am I right?

By the way, why it happens to reds and not so, for example, for green colour? Namely, the same problem appeared with 4:2:0 or 4:1:1 DV codec (PAL/NTSC) and I was never able to explain it.
 
Last edited:
Curious- in 24PN it has been said that the GH1 automatically goes is into 12 GOP and there is no need to select the gop setting. Is that true?

I'm assuming that the camera goes into 0.5 gop sampling, what ever frame rate it's set for.. Unless modded with the hack.

Pappas
 
I think I'll say what we're all thinking.... PLEASE BRING ME STABLE NATIVE 24P hahaha

Funny how before this hack we weren't so demanding hahaha
 
crunchy said:
Yes Barry, but I would still like to see the difference between 50i and 25p. I suppose that in this case the difference would be noticeably smaller.
Click to expand...
Shouldn't be. Should be pretty much exactly the same.

By the way, why it happens to reds and not so, for example, for green colour?
Click to expand...
Because green is pretty much sampled at full luminance, whereas red and blue are sampled at 1/4 luminance. You'll see this problem on reds and blues even in progressive encoding, but it's much more pronounced on interlaced coding.

Namely, the same problem appeared with 4:2:0 or 4:1:1 DV codec (PAL/NTSC) and I was never able to explain it.
Click to expand...
Look in the "articles" section of the forum for my article on "understanding color sampling" and it should hopefully make it a lot more understandable.
 
Barry_Green said:
Yes, what you're pointing out is exactly what would be expected -- the chroma encoding in 4:2:0 is using interlaced encoding when using 24p-within-60i, and it's using the much-superior progressive encoding when using 24p Native files.
Click to expand...


so you're saying that panasonic coded two different setups for chroma encoding even though they only have a single choice of interlaced output? That's doesn't compute. If you can show how you came to that conclusion that's cool but I don't think pana would have wasted the effort to do it like that yet only use a single choice of output. I could be wrong though.

also, how do we know the strange "red chroma" problem isn't actually a *bug* in their firmware?

PappasArts said:
It's 101 basics- Your camera does do it; and all 4:2:0/interlaced 60i/24 in combination make it worse.

If I had your camera I would show you it to you- and then you wouldn't like your camera anymore- Just kidding.
Click to expand...

I understand the logic but I'm staring at some shots I did while scouting locations today and I have a number of red cars and people wearing red shirts in the different shots and even when I blow the stills up, I'm not seeing anything nearly as bad as:

ixoraGH1closeup60i.jpg



If I have some time, I'll post a still. I don't have time right now though.

EDIT:

vlcsnap-2010-07-05-17h11m43s105.jpg

Darn, it's too small to see right.

screen.png


I don't see the same problems here.
 
Last edited:
Barry_Green said:
Because green is pretty much sampled at full luminance, whereas red and blue are sampled at 1/4 luminance. You'll see this problem on reds and blues even in progressive encoding, but it's much more pronounced on interlaced coding.
Click to expand...

Thanks Barry! Now it's perfectly clear!
 
Svart said:
so you're saying that panasonic coded two different setups for chroma encoding even though they only have a single choice of interlaced output? That's doesn't compute. If you can show how you came to that conclusion that's cool but I don't think pana would have wasted the effort to do it like that yet only use a single choice of output. I could be wrong though.

also, how do we know the strange "red chroma" problem isn't actually a *bug* in their firmware?
Click to expand...

Svart,is right,

It's CINEFORMS NEOSCENE that's screwed up.

I just ran off 10 conversions 5 with Chroma smoothing- and the same file with no Chroma smoothing.

Chroma Smoothing on- it does that constant with with all reds.

If you turn off chroma smoothing it pops in every forth frame.

NeoScene has always been buggy- this confirms its.

What sucks is I just upgraded Friday to the latest neoscene- For what?

Sample coming--

Pappas
 
Last edited:
Shot all weekend with 24 native. Had zero issues and the footage looks amazing. Since it's not broken I'm not gonna tinker with it.
 
Just for e record, I've had no issues with the c settings and native 24p and cineform to turn it into prores for awhile now.

I'm racing to hawaii tomorrow, I'll let you know in two weeks if I saw any trouble with lots of shooting.
 
You must log in or register to reply here.
Back
Top