I might have an opportunity to film a similar situation as the water with ripples next to some bushes tomorrow. I also found out that the photo file naming is normal. I have been playing with the color space, and the camera changes the first letter for each color space; thus sorry for the false alarm. It takes a long time to get to know this camera; I am constantly learning something new about it... Al
75Mbps GH1 Peak Performance Patch
- Thread starter Lpowell
- Start date
Unlike the GH1, which the Peak Performance Patch is designed for, the GF1 is not capable of recording at 60 fps. When the GF1's AVCHD-Lite encoder records videos with a 60p timeline, the sensor is actually being scanned at only 30 fps and each frame in the video is doubled. To make best use of the GF1, I recommend my Fast Action 3-Frame GOP Patch, which replaces the 720p60 SH video mode with a high-bitrate 720p30 patch:
http://www.dvxuser.com/V6/showthread.php?222791-50Mbps-AVCHD-Fast-Action-3-Frame-GOP-Patch
cloudcitycollective
Member
Thank you for the pointer! So you do recommend AVCHD over MJPEG for the GF1?
And thanks to everyone who's tested the Peak Performance Patch and posted their results here. The stability issues you've helped uncover motivated me to conduct more lengthy and intensive test cases than previous patches have required. As a result, I've made a number of modifications to improve the reliability of both NTSC and PAL versions of the patch:
* I fixed the MJPEG recording errors by increasing the frame height slightly to produce a 1440x816 video frame. It turns out the trick to making 4:2:2 color depth work properly is to set a frame height that is exactly divisible by 8.
* I had to back off on using a 10-frame GOP in 1920x1080 FHD mode, and have increased it to 15 frames in both 24p and 25p video modes.
* I was able to tighten up the stability of both 50p and 60p videos in 1280x720 SH mode.
This is not to say I think the patch is now flawless or bulletproof - there are too many operational variances involved to exhaustively test every possible combination. When I need the utmost reliability for non-repeatable shots, I'll stick with the Reliable In-Camera Playback Patch. But for the highest image quality the camera is capable of consistently producing, this Peak Performance Patch is what I'll be using.
For the latest version of the downloadable PTool INI file, check out the first post in this thread:
http://www.dvxuser.com/V6/showthrea...rmance-Patch&p=2222557&viewfull=1#post2222557[/QUOTE]
______________
Thank you again lpowell for your continued work creating new and wonderful patches to use with the GH13.
I have been using the 50 Mbps AVCHD 729p30/25 Fast Action 3 frame GOP patch with the 60 Mbps MJPEG 720p30 in 4:2:2 color.
That patch has worked very well for me with no problems encountered using a SanDisk Extreme 8 MB class 10 card.
I've been following this thread on the new Peak Performance Patch with great interest. On 1/20/11 I downloaded the zip file, extracted the ini file and printed it to better understand what settings were involved. I then manually entered the settings in PTool ver 3.51 and created a bin file for updating the firmware.
Today I downloaded the updated version 3 of the Peak Performance Patch and printed out the ini file to compare the previous version to the newer version. The new version changes the 1080p25 GOP Size to 15 but there is no command line for changing 1080p24 to GOP size 15. Is this an oversight and should I add that when I create the new bin file?
Is the increased stability in 1280x720 SH mode due to the increase in Constant for 720p from 466560 to 518400? Also curious about the Preset bitrate being increased to 1107 in this patch. That is a much larger number that the 180 in all your previous patches.
Am fascinated by all this and trying to understand. Had never shot video before the GH1 entered my life! Now I'm hooked and planning to have several of your patches available on different memory cards to easily make firmware changes.
* I fixed the MJPEG recording errors by increasing the frame height slightly to produce a 1440x816 video frame. It turns out the trick to making 4:2:2 color depth work properly is to set a frame height that is exactly divisible by 8.
* I had to back off on using a 10-frame GOP in 1920x1080 FHD mode, and have increased it to 15 frames in both 24p and 25p video modes.
* I was able to tighten up the stability of both 50p and 60p videos in 1280x720 SH mode.
This is not to say I think the patch is now flawless or bulletproof - there are too many operational variances involved to exhaustively test every possible combination. When I need the utmost reliability for non-repeatable shots, I'll stick with the Reliable In-Camera Playback Patch. But for the highest image quality the camera is capable of consistently producing, this Peak Performance Patch is what I'll be using.
For the latest version of the downloadable PTool INI file, check out the first post in this thread:
http://www.dvxuser.com/V6/showthrea...rmance-Patch&p=2222557&viewfull=1#post2222557[/QUOTE]
______________
Thank you again lpowell for your continued work creating new and wonderful patches to use with the GH13.
I have been using the 50 Mbps AVCHD 729p30/25 Fast Action 3 frame GOP patch with the 60 Mbps MJPEG 720p30 in 4:2:2 color.
That patch has worked very well for me with no problems encountered using a SanDisk Extreme 8 MB class 10 card.
I've been following this thread on the new Peak Performance Patch with great interest. On 1/20/11 I downloaded the zip file, extracted the ini file and printed it to better understand what settings were involved. I then manually entered the settings in PTool ver 3.51 and created a bin file for updating the firmware.
Today I downloaded the updated version 3 of the Peak Performance Patch and printed out the ini file to compare the previous version to the newer version. The new version changes the 1080p25 GOP Size to 15 but there is no command line for changing 1080p24 to GOP size 15. Is this an oversight and should I add that when I create the new bin file?
Is the increased stability in 1280x720 SH mode due to the increase in Constant for 720p from 466560 to 518400? Also curious about the Preset bitrate being increased to 1107 in this patch. That is a much larger number that the 180 in all your previous patches.
Am fascinated by all this and trying to understand. Had never shot video before the GH1 entered my life! Now I'm hooked and planning to have several of your patches available on different memory cards to easily make firmware changes.
Yes, the patch values must look pretty arcane without any explanation. Here are some notes on how it all works:
* NTSC 24p and PAL 25p are locked together with the same patch settings (except for GOP length).
* NTSC 60p and PAL 50p are locked together with the same patch settings (except for GOP length).
* 24p and 25p are interlaced, 50p and 60p are non-interlaced.
* 24p requires a GOP-length exactly divisible by 5 (due to telecining).
* I could not fully stabilize a 24p GOP-length shorter than the default 15-frames, so I left it unchecked.
* 25p actually needed a longer GOP-length to stabilize it at a 75Mbps peak bitrate.
* Video Bitrate of 54000000 produces 50Mbps target bitrate limit.
* Overall Bitrate of 72000000 produces 75Mbps peak bitrate limit.
* Preset Bitrate was increased to help stabilize 24/25p at the 75Mbps peak bitrate.
* Preset Bitrate 2 needed no adjustment to stabilize 50/60p.
* Constant for 1080p helps balance 24/25p average/peak bitrates.
* Constant for 720p helps balance 50/60p average/peak bitrates.
* NTSC 24p and PAL 25p are locked together with the same patch settings (except for GOP length).
* NTSC 60p and PAL 50p are locked together with the same patch settings (except for GOP length).
* 24p and 25p are interlaced, 50p and 60p are non-interlaced.
* 24p requires a GOP-length exactly divisible by 5 (due to telecining).
* I could not fully stabilize a 24p GOP-length shorter than the default 15-frames, so I left it unchecked.
* 25p actually needed a longer GOP-length to stabilize it at a 75Mbps peak bitrate.
* Video Bitrate of 54000000 produces 50Mbps target bitrate limit.
* Overall Bitrate of 72000000 produces 75Mbps peak bitrate limit.
* Preset Bitrate was increased to help stabilize 24/25p at the 75Mbps peak bitrate.
* Preset Bitrate 2 needed no adjustment to stabilize 50/60p.
* Constant for 1080p helps balance 24/25p average/peak bitrates.
* Constant for 720p helps balance 50/60p average/peak bitrates.
Last edited:
I tested v3 Peak performance patch last night and this morning near the beach by Morro Bay. Lots of video clips of ocean waves, sunset, dunes, calm waters and ocean waves by the shore. Only one type failed consistently near the beginning just around 1 second into recording. All successful videos were stunning, FHD, SH, and MJPEG H. The consistent failure is in the shade of busy shrubbery. Even with this consistent failure, I am very happy with your latest hack lpowell. Here is a 1 second video of the scene that fails consistently
Darren Levine
Veteran
great stuff, shows plenty of promise
Thanks, Al, your results sound quite similar to mine. I assume the failed video is in 1080p24 mode? Startup failures of this type are difficult to analyze as there is no evidence in the video file of what caused the recording to stop. I suspect the additional processing needed to apply interlaced telecine is what makes 24p more susceptible to this failure than the other video modes.
I've found there are exposure adjustments you can use to safely handle the type of scene shown in your embedded video. The hazardous areas for the codec are not the brightly exposed leaves. What spikes the bitrate is the fine detail in the darker areas of the shrubbery, which the unhacked encoder would have routinely smeared into mud. Increasing the contrast via ISO and Film Mode settings would crush the less important details into black, freeing up more bitrate for the leafy motion vectors. In overly busy scenes like this, high contrast often produces better visual results than high detail.
It's funny that you can see some artefacts at the (very) bottom of the picture. I've seen similar (and even worse) artefacts on some of my shots (always at the bottom). It looks like the encoder had not enough time to code the entire frame in allocated time. However, it's probably not true. Maybe it's just related to video player??
To tell you the truth I am embarrassed to say I don't know. I do not have the check mark in Ptools for 24p. I used your v3 hack exactly except for un-checking the PAL/NTSC and the all languages. To get 24p, don't I need to check the 24p options in Ptools? Al
Sorry, by "1080p24" I was only asking if the video was taken in the FHD 24fps video mode. To refer to the non-interlaced 24p hack (which the Peak Performance Patch does not use), I'd typically call it "Native 24p".
I've become confused reading all of the replies. I want to make sure I clearly understand what you're saying, lpowell. This new patch (settings file H) includes all three of the patches you list in our initial post, and these patches should deliver the highest quality, reliable images that you currently believe the AVCHD encoder can handle? So it will create high-quality images on fast-motion subjects in both 720p60 AVCHD and 1080p24 AVCHD? And we can try the 24p native at our own risk? Also, are all class 10 SD cards created equal, or are some faster than others? I have Transcend Class 10 cards, but I don't know the actual speed in MB/s of the cards.
The Peak Performance Patch increases the peak bitrate of FHD, SH, and HD video modes to just under 75Mbps, while limiting average bitrates to 50Mbps for AVCHD and 60Mbps for MJPEG. Although my testing has shown the camera to work with a high degree of reliability at these bitrates, the safety margin at 75Mbps is inherently narrower than my Reliable In-Camera Playback Patch, which sets a 45Mbps peak bitrate. I use the latter patch when I need the utmost reliability and I use the Peak Performance patch when I want the highest practical image quality. If the reliability risk of the Native 24p patch seems acceptable, you may want to try it yourself, but I have not attempted to optimize any of these patches to work well with it.
All Class 10 SD cards must sustain a minimum write speed of at least 80Mbps. That should make any of them fast enough to handle the demands of the 75Mbps Peak Performance Patch.
All Class 10 SD cards must sustain a minimum write speed of at least 80Mbps. That should make any of them fast enough to handle the demands of the 75Mbps Peak Performance Patch.
Last edited:
The native 24p patch doesn't affect the reliability of SH mode, right?
LPowell, you mentioned a version upcoming for the GF1. Any news on this or do you feel the patch is suitable as-is, minus the 1080p entries etc, for a GF1 with a good Class 10 card?
The reason the Peak Performance Patch is not really suitable for the GF1 or G2 is because their 720p SH mode can only record AVCHD-Lite videos. Since these cameras cannot scan the image sensor faster than 30fps, the 720p50/60 mode in this patch does not produce optimal results. For GF1 and G2 cameras, I'd recommend my Fast Action 3-Frame GOP Patch:
http://www.dvxuser.com/V6/showthread.php?222791-50Mbps-AVCHD-Fast-Action-3-Frame-GOP-Patch