Hey Lpowell, I'm interested in more detail depth about what is changed on this patch. I read explanation but I don't feel it answered any technical questions. Can you go into the details about what you did and why you chose the settings you did?

First, I tested the new VGA mode by setting it up with a 960x720 frame size and the HD mode at 1280x720. I used the same 60Mbps 4:2:0 quality settings on both modes. Then I shot videos and compared the resolution quality of frame grabs magnified in Photoshop. This convinced me that VGA mode produces comparable quality to HD mode at the same resolution.

In order to produce iPad-compatible 960x720 videos in VGA mode, I used my Reliable In-Camera Playback Patch MJPEG quality settings to reduce the bitrate to around 30Mbps. These are the maximum MJPEG quality specs the iPad is documented to accept, though I haven't obtained one yet to test.

For the 2X anamorphic 1920x720 VGA mode, I retained the 75Mbps 4:2:0 quality settings used in my 1920x810 HD mode anamorphic patch. Both modes appear to produce comparable quality results. A 1.5X anamorphic patch can easily be produced by setting the frame size of the HD mode to 1920x720. I published optimized quality settings for this 1.5X anamorphic HD mode in my Blackout-Powell Native 24p Patch.

PTool 3.55d also provides two new video and audio buffer size patches. I tested various settings in both NTSC and PAL 1080p modes, along with verifying that the new patches didn't degrade 720p stability. I found that increased buffer sizes improved the reliability margins of the 1080p modes, and allowed me to complete shots of certain subjects that caused the version 1 Max Latitude Patch to fail within two seconds at 24p. However, I also discovered that excessively large buffer sizes can produce sporatic empty P-frames in certain circumstances. I tested various combinations of video and audio buffer sizes until I found one that improved stability without the producing empty P-frames.

The video and audio buffer size patches improve NTSC 24p reliability, and increasing the 24p GOP-size from 12 to 15 has helped as well. However, it's still possible for 24p recordings to fail under extreme stress. While I've also done some testing with Native 24p off, the results have been inconclusive. Those who don't mind interlaced pulldown may be interested to try turning off Native 24p and see if it improves reliability.

The PAL Native 25p mode is definitely more resilient than NTSC Native 24p, though the 25p average bitrate trends lower. I believe this Native 25p patch is the most reliable one that I've tested. The NTSC 720p30 and PAL 720p25 modes are even more stable, with virtually zero failures in my v2 testing.

I'm convinced that the vast majority of failures I've seen in v2 testing are actual SD card write-speed failures rather than encoder malfunctions. Class 6 SD cards are definitely not fast enough for this patch, and not all Class 10 cards will make the grade. In practice, bitrates over 100Mbps only occur briefly, but that does technically exceed the Class 10 minimum write-speed spec. Also, be sure to format the SD card in-camera before important shoots, to guard against memory fragmentation.

@LPowell What's your workflow to play mjpeg video file on iPad? Can iPad import the video files from SDHC card via Camera Connector Kit?

Stonebat, I haven't been able to borrow an iPad to test the VGA 960x720 video files. If anyone has the means to do so, I'd be very interested to hear of their results. Apple doesn't appear to have released public documentation on exactly how the iPad video import process works. An imported MJPEG file is no doubt converted into a more compressed format, most likely H.264. I suspect 960x720 is the native iPad video frame size, chosen to fit conveniently within the user interface of the iPad's 1024x768 screen.

I sold iPad1. There was no automatic way to transfer video file. I had to drag a video file into iTunes desktop app and sync to iPad. Nevertheless this is a good feature to impress Apple friends :)

Prolly you know already, but here's the spec.

Video formats supported: H.264 video up to 720p, 30 frames per second, Main Profile level 3.1 with AAC-LC audio up to 160 Kbps, 48kHz, stereo audio in .m4v, .mp4, and .mov file formats; MPEG-4 video, up to 2.5 Mbps, 640 by 480 pixels, 30 frames per second, Simple Profile with AAC-LC audio up to 160 Kbps per channel, 48kHz, stereo audio in .m4v, .mp4, and .mov file formats; Motion JPEG (M-JPEG) up to 35 Mbps, 1280 by 720 pixels, 30 frames per second, audio in ulaw, PCM stereo audio in .avi file format

Vitaliy, would there be a way to make the resolution beyond 1280x720 better for MJPEG?

I find there's more detail than with AVCHD, but the picture is quite harsh at 1920. I assume this is due to the codec and cannot be improved?
Just thinking out loud : a 4:3 mode recording AVCHD is probably even a stupid idea to think of?

uliman
This questions has been asked dozens and dozens of times. Just google for the answer.

I'm curious, can I set this patch to have a much lower bitrate (I honestly don't need it) but still use the Anamorphic crop? I can't seem to find an answer to this...

I often use the Reliable In-Camera Playback Patch MJPEG settings with an anamorphic frame size of 1920x810. Although the screen glitches a bit at the start of playback, the videos are recorded properly and can be reviewed in-camera.

The MJPEG patches only affect the MJPEG video modes, and the AVCHD patches only affect the AVCHD video modes. Thus, you can use the MJPEG patch settings from one INI file with the AVCHD settings from a different INI file.

In certain cases, the MJPEG frame size patches can be customized without changing other MJPEG patch settings. When using a patch with a large frame size, you can always reduce the frame size, and that will not cause any reliability problems. However, if you increase a patches' frame size (or activate 4:2:2 color), that will increase the bitrate and may cause the MJPEG encoder to fail under stressful conditions. (For best results, horizontal and vertical frame sizes should both be divisible by 2).

Increasing the frame size of the Reliable In-Camera Playback Patch will increase its peak bitrate as well. However, I have tested this modification at 1920x810 and concluded that it records reliably (though it does suffer from an in-canera cosmetic playback glitch).

Before using a new SD card, be sure to use the camera to format it. With the100Mbps Max Latitude Patch, you may find that extremely bright, highly-detailed scenes can be shot more reliably by reducing the exposure level. The high peak bitrate of this patch gives it the ability to capture high quality image details even in underexposed areas of a shot.

I know this doesn't have anything to do with the speed limit error you were getting later on, but as far as filming 15m in, wouldn't the MJPEG mode at 100mbps stop recording no matter what a little over 2 1/2 minutes in due to the 2GB cap in the file size... or has this been worked around and I missed it. Last I recall (unless I recall incorrectly) only the AVCHD mode did clip spanning, where if the AVCHD clip hits the max file size it just starts a new one right away, while the MJPEG just stops until you hit record again. At 100mbps you would hit 2GB in just over 2 1/2 minutes.