Blackout-Powell Native 24p Patch - GH1 FTW!!!

First I want to thank lPowell for all the work you are doing for the benefit of all. This is my first post after a lot of time reading yours. I've experimented with all the settings and I've reached my own conclusions. Mixing various settings, I use the one here today show if anyone wants to try. The result is exceptional 1080pN/25 with only one failure (read speed) in many months of use and according StreamParser any empty frames. I do not know if it can be useful for someone. 48Mbps average in high detail scenes, without mud, and some peak of 100Mbps !!!!!. I understand that is not 100% reliable, but its quality is exceptional. In 720p/50 I get plenty of empty frames and its not usable, but incredibly ok for 1080pN/25.

Thanks everybody.

Settings:
Prevent version compare=Checked
Third-party battery=Checked
All Interface Languages=Checked
30min limit removal=Checked
PAL<->NTSC Menu=Checked
720p30 width=1440
720p30 height=816
720p30 420->422 color sampling=Checked
E1 Quality=280
E1 Table=4
E2 Quality=250
E2 Table=10
E3 Quality=225
E3 Table=24
E4 Quality=200
E4 Table=48
Native 24p/25p=Checked
Video Bitrate FHD/SH=51000000
Video Bitrate H=29000000
Video Bitrate L=17000000
Overall Bitrate=61000000
720p50 GOP Size=20
720p60 GOP Size=20
1080p25 GOP Size=15
1080p24 GOP Size=15
Preset bitrate=545
Overall Bitrate 2=61000000
Constant for 1080p=5184000
Constant for 720p=518400
Framerate related comparison=12
 
This patch looks v promising. Many thanks lpowell and blackout.

For those without anamorphic adapters, what pixel squeeze factor should one use in post (in a package like AE) to eliminate the horizontal stretch correctly?
 
Thanks, Blackout and lpowell, for your continuous effort. After installing the patch I had 4 card speed errors in FHD 24p with high detailed green foliage after about 15-20 sec in the clip (using the 14-140 kit lens and a 8 gb class 10 sandisk). So it seems that stability is still an issue to some extent. Nevertheless, it is a great feature and I will continue to use the 24p mode. As you mentioned at the beginning of the thread - it is rather predictable at what type of shot the cam will fail. So one can avoid these shots to a certain extent. Great work
 
Hi everyone, I know this question has been posed very often, but I still can't find an exact answer:

What's the advantage of 25pN over just record in 50i and then put it into 25p timeline?
I've read that there's a difference in color, but then in another thread I read there is no difference.

So, what's the advantage of this hack if I'm in PAL-country?
 
"The key to stabilizing the Native 24p patch was finding a way to eliminate the sporatic occurance of empty P-frames in the video stream. Once I figured out that trick, I fine-tuned Blackout's average and maximum bitrates to match. Nothing was more stable than Blackout's original12-frame GOP, which I used in 25, 50, and 60fps modes as well."


so what was the trick?

Now that my GH2 is coming, I can mess with the GH1 settings a little bit again before my next shoot. I'd like to shoot native 24p but still can't handle the R/W speed failures during production.
 
I've been chomping at the bit to put a native 24p patch on my camera.

Does this patch still disallow in-camera playback? That is a near deal-breaker for me as the HDMI out only works for playback and when on set I need to playback takes for the director.
 
For those without anamorphic adapters, what pixel squeeze factor should one use in post to eliminate the horizontal stretch correctly?
When shooting in MJPEG HD mode with a normal, non-anamorphic lens, you'll need to vertically stretch the patch's 1920x720 frame out to the standard FHD 1920x1080 frame size.
 
So is a stable 1920x1080 MJPEG just not possible? Is that the reason for the 1920x720?
A Full HD 1080p frame produces too much data for the MJPEG codec to handle reliably. The 1920x720 MJPEG mode in the Blackout-Powell Native 24p Patch is the widest that will support 4:2:2 color depth at moderately high bitrates of around 50Mbps. It also works natively with 1.5X anamorphic adapters, which will produce widescreen videos that require no stretching in post-production.
 
I don't mind the pull down method with JES, so do you think this is better than your 75Mps patch? I'm just looking for the highest quality image in the end...

Also, this may be a dumb question, but when I import a hacked file into FCP, it never reads the data rate above 13-17Mps. The images look much better, so I know the hack is working, but why doesn't FCP read the correct Mps?
 
No this was not Ocean Footage it was in my backyard.

Ok then if there is more bits going to the 25p raster rather than interlaced, should i reduce the overall bitrate and stop the card write errors. Or are the failures the empty bits as you said, I'm not seeing these in any of my footage in stream parser though.

DrDave, are you shooting in creative movie mode or one of the still mode settings?? ISO 400 and 800 are you shooting at night??? I'm lucky to get 5 seconds before card write error.
I'm in movie mode, full manual. Basically, I just loaded the patch without changing even one box, then used my usual standard film setting with -2 -2 -2 -1. I have the dynamic range on standard, but that doesn't seem to change anything. I shot some in room light and some just outside. I'm wondering if you tried a different card? Or maybe reflash the patch?
 
OK--using the kit 14-45 in AF mode I was able to crash the cam reliably, every time, in 5 seconds by pointing it at a shrub in bright sunlight. Changing the ISO gave the same result.
I swapped out the 14-45 Panny for the Rokkor 50mm and was unable to crash the cam at the shrub of doom. Now, that lens is sharper than the 14-45 at F/8, but that was the difference.
The error I got was "card unable to keep up data" or something like that.

One thing I have noticed is that the remaining time counter speeds up when you are recording complex subjects, so you can kind of judge if you are shooting something borderline. I'll have to test it some more, but if the RTC isn't jumping, I would guess that you are in a safer recording zone.
 
If you try yes you can break it. I have shot many minutes of continuous footage since loading this patch, though after reading your replies I decided to make it fail, so I went outside. shot wide of tree's fine, close of tree's fine, shot a full frame of grass and got it to fail.

using a manual lens, 24p, iso 320, shutter 400, awb (lazy), -2,-2,-2,-2, untouched settings, and a transcend 16gb (not very fast or my fastest card)

Thing is it's much more stable than anything I used before, and I could break almost any patch by shooting grass, but if I had an actor in frame no problem, If I really had problems I'd pop out ye olde reliable patch and load it shoot the shot and move on. Seems needlessly confusing but that's what you have to deal with when working with equipment like this and within the budgets we have at our disposal.

Any high detailed shot since I started using this hack causes problems it's all about finding your own solutions that work. I have found putting my hand in front of the lens for a few seconds of recording can help to get past the failure, and at other times have found that if I mange to record something like a less detailed scene and go back to the high detailed one it works. I know none of this is solid or professional but it is a hacked camera after all. I have found that I had more issues than normal with an electronic lens attached to the camera, especially when image stabilization is on. I would recommend (as it's what I do but I have no proof as to weather or not it actually makes any difference) that you turn off any auto functions or added functions the camera has to do as well as record like intelligent exposure and things like that.

or pick up a sandisk class 10 extreme 30MB/s card, happy testing and I hope you figure out the best setting for you!

have a good one!
 
OK I was able to crash it with my 50mm Rokkor by stopping down (more DOF) and then adjusting the focus until the counter on the remaining recording time went wild. So I think it really is as simple as a "detail threshold".
 
Forgot to mention Powel points out that there are issues in high detail scene in the second post, and advises using a diffusion filter to lower the detail and hopefully avoid lock ups (not quite what he said but his advice on avoiding detail above 1080 was to use this filter). I figure if you use the right filters with the right card and lens/setting combination I think you will find it much harder to crash.

nothings perfect, testing must end at some point, but I understand wanting to know why anything happens so as to avoid or work around whenever potential issues may pop up while out on a gig without having to change settings.

good luck!
 
A Full HD 1080p frame produces too much data for the MJPEG codec to handle reliably. The 1920x720 MJPEG mode in the Blackout-Powell Native 24p Patch is the widest that will support 4:2:2 color depth at moderately high bitrates of around 50Mbps. It also works natively with 1.5X anamorphic adapters, which will produce widescreen videos that require no stretching in post-production.

Thanks, that's pretty much what I figured. I've also heard that even stretched to full 1080p, the MJPEG hack doesn't resolve as much detail as a 1080p frame "should" or normally would.
 
I don't believe the card speed has anything to do with it even though the camera reports a card write error. I am using a sandisk class 10 extreme 30MB/s card. Most people on this forum are shooting shallow DOF footage, I'm not, i shoot mostly deep DOF which appears to trigger the errors. That is in my opinion the reason for most of my fails.

lpowell/Vitaliy, you both should be nominated for a Nobel prize for the 40Mbps AVCHD High Reliable patch, it's unbreakable with any lens, any detail, all day long. I just wish this new one was the same.

Question (Mr Powell)

Is it possible to apply the 25pN magic tweaks of this new patch to the 40Mbps High reliability patch so it is still 100% reliable but in 25/24pN????

Rambo
 
I don't think it is the card speed either--if you look at the remaining time counter, you can see the numbers really speed before the crash, so something in the software has triggered an overrun.
I haven't tested why it kicks in outdoors as opposed to indoors, maybe the shutter speed is an issue, in addition to the higher resolution of bright sunlight.
 
You may be right about card speed having nothing to do with it, seeing as my weather is not really great up here haven't done many outdoor test shots yet so I'll leave this to you fellas.

Thanks for the patch (I was using a much riskier bit-rate with 24p and was doing fine so this is even better), and hopefully we will be able to find and iron out all the kinks.

have fun!
 
Back
Top