50Mbps AVCHD Fast Action 3-Frame GOP Patch

I dont know whether the GOP-length can be reduced for the 720p60 video mode as well. The new PTools patches I used to fine-tune the 3-frame GOP of the 720p30 video mode were not available back when I developed the High Reliability patch. While it may be possible to improve the 720p60 mode with the new patches, it would require a new round of testing to verify that its reliability was not diminished.
 
I used sandisk ultra 8G, 15M/s write speed,I used new pathces on the 30P mode that good,But i
change the FPS to 25P,But gh1 can't works. why?
 
lpowell I am indeed getting different results in different conditions. Yesterday night's tests were poorly lit. Yet still, what strikes me is the difference between how mjpeg is handling low light versus the avc.

It's astonishing. I may be just feeling this and believing it, but the noise looks different even.

Thanks so much, again sorry if my post was confusing to any results, yet yesterday's tests were really poor. Especially compared to the mjpeg.

Daytime brought some relief.

I will post later on tonight after I look at the footage in detail and do some more comparisons between settings.

No write fails except once on Fast Action Settings at 720 shutter 100. SH Mode
 
ive been using the hack for a week. it failed to write only once. pancake + AFS 720/30p avchd recording. i switched to MF and it worked fine. i used sandisk class10 16gb. i thought about switching to the 40Mbps 720/60p reliable setting but kept the 30p setting. more key frames might look closer to native 24p film recording. if needed, i can transcode from 30p to 24p. 2x slow motion... oh well... fallback to the 720/60p hack.
 
I am getting the same very low bitrate that two other people have reported, using a CoreMicro 16gig SDHC card, was getting up to 30MBPS with avhcd in the reliability patch, no problem, with this one, it rarely gets above 5 or 6. Not sure why. I will try the mjpeg and report back, but anyway, I wanted to say thanks for the patches, the reliability patch seems great for me so far:)
EDIT: getting around 50MBPS with the MJPEG setting, but at the risk of sounding like a total noob, why is it only saying 3 mins 45 secs on the timer countdown of available footage(as opposed to 2 hrs or so on the avhcd setting) when a 15 second video file is only 95mb and the card is 16 gigs?
FURTHER EDIT: Tried identical settings and shoots with the reliability patch compared to this one. Although bitrate is improved from my initial impressions, it still seems lower than the reliability patch bitrate...not sure why...and its very tempremental in Low light....not getting write failures, just very poor bitrate and framerate....any failures I am putting down to the card, although it is a class 6, I think its probably a relatively cheap card. Will keep testing, but for now, I think I am happy with the reliability patch until I get a very good card to test properly.
 
Last edited:
Nice patch! I'm getting excellent results in good light, but frequent write failures in low light.

I am toying with the idea of raising the gop length from 3 to 5 and reducing the bitrate from 51mbit to 45mbit to make the patch more reliable. Is it safe to do this? i.e. it won't brick my camera will it?
 
why is it only saying 3 mins 45 secs on the timer countdown of available footage(as opposed to 2 hrs or so on the avhcd setting) when a 15 second video file is only 95mb and the card is 16 gigs?
2GB per file limit for mjpeg. i carry 2 * 16GB cards to cover at least 45 mins footage if all in mjpeg. that might sound bizarre, but one fully charged battery can do it. for longer recording, definitely avchd. avchd mode kept around 20Mbps and some scenes with good lighting and sharp focus yielded 55Mbps. i'd use avchd in good lighting. otherwise mjpeg. if all fails, fallback to the reliable 40Mbps.
 
Last edited:
I am toying with the idea of raising the gop length from 3 to 5 and reducing the bitrate from 51mbit to 45mbit to make the patch more reliable. Is it safe to do this? i.e. it won't brick my camera will it?
I haven't seen any danger in increasing GOP-length or decreasing FHD/SH, H, or L bitrates in the Fast Action patch. The values I recommend are tweaked to the optimal stable settings, but there should be no harm in easing up on them if you like.
 
just had my first fail with this patch.
was working fine with the lumix 14-140mm and 20mm lenses, but finally put on my Canon FD f1.4 50mm with the camera set to shoot w/o lens and it failed immediately.
avchd 30p (SH) was fine though and works with any lens.
is there something about using a fully manual lens that is actually harder for the camera/codec to deal with??
 
just had my first fail with this patch.
was working fine with the lumix 14-140mm and 20mm lenses, but finally put on my Canon FD f1.4 50mm with the camera set to shoot w/o lens and it failed immediately.
avchd 30p (SH) was fine though and works with any lens.
is there something about using a fully manual lens that is actually harder for the camera/codec to deal with??
I have used this patch with a variety of manual lenses, including the Canon fd f1.4 50mm. Have never had a fail.
 
failed again yesterday. avchd 30p mode. nikkor 50 1.4 ais. little low light. quickly switched to mjpeg. all worked fine.
 
very odd.
i can't record at all in mjpeg with manual lenses.
stonebat can't record in avchd with manual lenses.
others have no problems at all.

if my Transcend class 10 16gb card can shoot both patched codecs with the Lumix glass, why can't it handle primes? i may try reverting the mjpeg to Reliable while leaving the avchd with the Fast Action patch.
 
I'm sorry to hear that you and others are having reliability issues with this patch in widely differing recording situations. There is an inherently lower safety margin when recording at over 45Mbps than with my Reliable In-Camera Playback Patch.

Since I haven't yet been able to reproduce the problems you're seeing on my GF1 and two GH1 cameras, it's difficult for me to speculate on the causes. However, the most variable hardware factor that I've noticed is the wide range of write-speed performance among the various brands and types of SD cards. I suspect there may even be differences in individual SD cards from the same manufacturer. You can see how difficult it is to pin down a minimum performance spec for these devices in the following chart:

http://media.bestofmicro.com/F/F/178107/original/sdhc-transfer-write.png
 
any ideas why i'd experience a difference in card performance between auto and manual lenses?? i can't come up with any logic to explain this.
 
GF13 user here. No problems at all with the patch, whether shooting legacy or native glass. Transcend Class 10 16GB SD, but my Transcend Class 6 16GB SD works just as well.
 
oddness.
i've got it working with my primes now out in bright daylight.
the only difference is that it isn't as lowlight as the failed test, but in that test the auto lenses worked and the primes didn't.
 
how about this? use this 30p hack to take 30pN clips at 60th shutter speed. during post-processing, "conform" the clips to 24p. i.e. slow down 30p to 24p without dropping any frames. wouldnt it give more dreamy look without too much jitters?
 
um, why would you want to slow down your clips by 20%??
if you're shooting 30p and want dreamy motion blur, use 1/30th shutter.
if you want to shoot 60p for 24p slowmo, then shoot 60p and use 1/120th shutter....or for more blur use 1/60th.
any jitters are a result of improper pulldown of 24p from 60i or a high shutter speed.
in general, if you're shooting 30p use 1/30th, 24pN use 1/24th, 60p 1/60th and 24p in 60i use 1/60th or 1/120th.
 
180 degree shutter rule. for 30pN, 60th shutter speed. anything faster than that, too smeary look. at 30th shutter speed, closer to home video feel. not enough motion blur.

why slow down to 24p conformation (not conversion)? thats 25% slowdown without losing any original frames. that sounds fine with me when using slow background music. yeap slow it down without any smearing.

for fast action and/or 25% slowdown, 720/30pN hack.
for fast action and/or 2.5x slowdown, 720/60pN hack. <== taken at 120th. conform to 24p.
for slow action and no slowdown, 1080/24pN hack.

for fast action, i'd prefer 30pN hack at 30th or 60th if 2.5x slowdown is not needed. one stop faster than 60pN hack at 60th or 120th. i wish there is a way to support both 30pN and 60pN in one firmware to use 60pN exclusively for 2.5x slowdown. x-mas wish?

edit: i tried 30pN to 24p conformation. from pp cs5, select multiple clips, right click, modify, interpret footage, set the frame rate to be 23.976. very nice. just enough motion blur. more dreamy. this 30pN hack is a keeper for sure. definitely 24pN, 30pN, and 60pN hacks r a wonderful set.
 
Last edited:
so it seems that the 60mbps mjpeg has a tendency to fail in low light static (no movement) shots, especially when using primes and when wide open (f1.4).

lpowell:
do you know what the highest mjpeg bitrate is that is playable in-camera? i'm wondering if i can make a version of your patch that is somewhere between reliable and fast-action.
 
Back
Top