Halving your FPS and eating it too

S

shaveblog

Well-known member
I'm curious why there doesn't seem to be any discussion about Ptool 3.40d's AVCHD FPS option, which halves the FPS from 59.94 to 29.97 (NTSC) or 50 to 25 (PAL) when shooting AVCHD.

I understand this setting doesn't improve compression, but it would seem to be a good thing in terms of halving the bitrate for a given compression setting to allow for twice the footage on an SDHC card.

Am I missing something here? If I'm log-and-transferring my AVCHD footage to FCP and winding up with 30p ProRes anyway, am I losing something by shooting at 29.97 to begin with?
 
Last edited:
no responses?

no responses?

I was curious about the answer to that too. Where is Barry when we need him? ;)
 
I'm a PAL user. I've tried setting the 720p 50 to25fps option 'on' but the Panasonic PhotoSillyFun name HD software (sic) still showed the files as 50 fps. I'll be able to load them into my editing software after the weekend to see if that's correct...but I do see 25fps for the AVCHD 1080p ones, so I suspect the conversion isn't working. Anyone else checked yet?
David
 
We don't want to shoot at an unusual framerate, we're just curious about an option in Ptool 3.40d which nobody seems to have mentioned yet.

30fps isn't "unusual". It's NTSC standard format 720p video, every GF1 and GH1 shoots it fresh out of the box. Or at least it's what you wind up with after the GF/H1's 720/60i footage is properly deinterlaced and stripped of duplicate frames by whatever video editing program you're using. I use Final Cut Pro and iMovie on the Mac.

What chrisccw and I are curious about is whether the Ptool 3.40d's option to halve the GF1's shooting framerate in AVCHD -- in essence, to change it from shooting 60 frames per second with every other frame a duplicate of the one before it, to shooting 30 frames per second with no duplicates. Seems a win-win, as it effectively halves the bitrate for a given AVCHD compression setting, with no apparent loss in quality. So you can up the bitrate, lower the compression artifacts, and wind up with as much or perhaps even more AVCHD shooting time per SDHC card than the stock settings.

Once again, to the experts here...am I missing something?
 
shaveblog said:
30fps isn't "unusual". It's NTSC standard format 720p video, every GF1 and GH1 shoots it fresh out of the box. Or at least it's what you wind up with after the GF/H1's 720/60i footage is properly deinterlaced and stripped of duplicate frames by whatever video editing program you're using.
Click to expand...
I'm guessing there was sarcasm there, but you never know.
 
I thought it might be sarcastic, but then some poor young kid comes on here, sees that comment, and then believes 30fps is somehow bad, and the world is robbed of what could have been the next Harmony Corrine.

Now that's sarcasm.

Seriously though, nobody has any thoughts on halved AVCHD framerate option in Ptool? I can't believe this is such an arcane topic. I mean, I know all the hot action is about extreme bitrate MPEG but some of us need to get more than 3 mins per 16GB card.
 
Well, its the standard framerate for YouTube. If I know something is destined for YouTube, then 30fps makes sense.
 
I enabled the 720p 59.94->20.97 feature a while ago, but the numbers didn't change much. Bitrates stayed pretty much the same with the exception of the L setting, which would half at shutter speeds of 1/60th+. There exact relationship between the AVCHD settings is a bit of a mystery at this point.
 
Okay, I did some actual testing just now -- shot the Pappas Star Chart with a Canon FD50/1.4 on my GF13, S-mode, shutter locked to 60 with AEL, lens stopped down to f8. Identical lighting, tripod lockdown, the whole nine.

First I shot a minute with lpowell's MPEG settings, which Quicktime Player says was around 35 Mb/s at 30fps. Oddly enough the audio was showing up as 16K as viewed in Quicktime. This is straight off the card, no transcoding.

Next I shot a minute of Star Chart in AVCHD using the "Perfect Project" SH setting, with the firmware's halved framerate feature ticked. And sure enough, when viewing the raw MTS file with Toast Video Player, I got 29.97fps with a bitrate of 18.25 Mb/s. Audio was listed to stock spec at 192kbps/48kHz.

I went into Ptool 3.40d, created a new firmware with all the same settings as above except for the halved AVCHD framerate option unticked, and flashed the GF1 again.

As expected, now the AVCHD SH mode gave me 59.94fps, but the bitrate was unchanged -- still 18.25 Mb/s. So modulr is right, halving the AVCHD SH framerate does not, as the Ptool's tooltips imply, halve the bitrate.

Here's what puzzles me, though: looking at the two clips, both at the same bitrate but one spreading that pool across twice the frames, I'll be damned if I can see a difference. If there is one, then the Star Chart viewed on a 24" IPS monitor at 1920x1200 isn't a good enough test to resolve it.

So what I'm left with is this: there's no reason to use the halved AVCHD bitrate option in Ptool, unless you're shooting in the lowest quality L mode which you're not because you're here.

Unless, as I said at the head, I'm missing something.
 
You shouldn't see a difference between 18mb/s 30p and 18mb/s 60p, when you're shooting a static scene.

Try it on a motion scene and you may see the differences, but not likely on a static scene. Remember that a long-GoP codec only encodes the changes between frames, and on a static scene there is no change between frames, so there's nothing to tax the codec or show differences.
 
Good point, thank you for the feedback. I was trying to set up as identical a shot as possible for the two modes, and I thought the Star Chart was sufficiently codec taxing. But you're right, let me set up a moving shot that's duplicable for testing both modes.

On a side note, at one point I forgot to swap the lowly 2GB Sandisk class 2 card I use for Ptool flashing with the 16GB Transcend class 10 I use for shooting, and damned if the Sandisk didn't keep up with 35Mb MPEG without a hiccup. In camera playback too.

Let me repeat the test with a moving shot and I'll report back. Thanks again for the feedback.
 
At the suggestion of Barry Green, I redid the comparison between Ptool 3.40D's "Perfect Project" AVCHD SH setting with the halved-framerate option ticked and unticked.

This time I placed the Pappas Arts Star Chart on a turntable spinning at 33-1/3 RPM to create a moving subject that was perfectly repeatable for the comparison. The GF1 with Canon FD 50/1.4 was set for f/8 and positioned directly over the turntable platter on a tripod for a full-frame shot of the spinning pattern. Shutter Priority mode, locked to 30fps with AEL trick. One minute of footage was shot for each trial.

The results:

AVCHD SH at 59.94fps -- 12.35 Mb/s bitrate

AVCHD SH at 29.97fps -- 12.35 Mb/s bitrate

MPEG HD at 29.97fps -- 34.5 Mb/s bitrate

I can't tell any difference between the two AVCHD clips. The MPEG clip looks very, very slightly cleaner near the center of the chart where the pattern tightens up. But it's a squinty, pixel-peepy difference, not night and day at all, and I'm not sure it doesn't come down to playback engine differences between Toast Video Player (raw AVCHD playback) and Quicktime (MPEG).

So basically the verdict is do what makes you happy?

That's it?

I thought there'd be more.
 
shaveblog said:
This time I placed the Pappas Arts Star Chart on a turntable spinning at 33-1/3 RPM to create a moving subject that was perfectly repeatable for the comparison.
Click to expand...
Now that's a cool test trick.

The results:
AVCHD SH at 59.94fps -- 12.35 Mb/s bitrate
AVCHD SH at 29.97fps -- 12.35 Mb/s bitrate
MPEG HD at 29.97fps -- 34.5 Mb/s bitrate
So basically the verdict is do what makes you happy?
That's it?
Click to expand...
Don't forget that GF1 has Lite implementation of AVCHD codec, which most likely means that sensor is only capable of 30 fps natively, which are then doubled to 60 fps in AVCHD encoder.
 
You must log in or register to reply here.
Back
Top