@mpgxsvcd The Stream Parser log file of the HBR test you uploaded shows no evidence of the encoder entering Fallback mode. After the initial GOP, all frames were encoded using high quality quantization tables.The Stream Parser bitrate display you uploaded does not show typical Fallback mode bitrate-limiting behavior or a cyclical bitrate cadence. Where that type of dip in bitrate may occur is when an auto-focus lens briefly defocuses in AFC mode, producing a blurry image. If you can send me a copy of the original MTS file (via Dropbox or Vimeo) I'd like to examine the frame encoding.

Raw bitrate comparisons are of limited value when evaluating patches that use differently optimized quantization tables. The important criteria with Flow Motion v2 involve how effectively it makes use of its bits, rather than how large its bitrate can climb. In particular, FM2 avoids using overkill techniques that can max out bitrate without providing perceptible improvements in motion picture quality.

@rawmeyn - 80% VMM uses the same settings as 24p but requires (30/24) 1.25 times the bandwidth.

Thanks lpowell--can't wait to try this out when I get back from Germany.

Wowza - just looking at some low light stuff - stunning ! No problems with a Pretec 433x 64 gig card a Sandisk 64 gig 30 + 45MB card so far in all modes.

I tried to do the elemental streams. I have attached the .log file that it made for the 1080p @ 30 FPS recording. I really am not sure how to read or use it. Maybe someone could point me in the right direction?

I actually don't think the blip in the 1080p @ 30 FPS was that big of an issue. I went back and played the file in streamparser and it went through a rapid change of exposure and focus at that point. I don't think that is indicative of how it behaves in normal scenarios.

Overall the bit rates look good and the footage I took with Flow Motion v2.0 didn't have any visible flaws. It looks really good.

I also did a comparison of it and the No adverse Affects settings to see where the bit rate differences were coming from. Those results are attached and below. I was a little surprised to see how close the I frames were for both of the settings. It looks like the B frames are where most of the difference is.

Also note that I used a version of the No adverse affects settings that does not use B-frames for 720p. I simply forgot to check that setting when I loaded it. I have it turned on now.

1080p @ 30 FPS No adverse Affects ├─── I Frame Size (Min/Ave/Max): 170,112 / 432,871 / 461,952 ├─── P Frame Size (Min/Ave/Max): 53,184 / 235,742 / 420,672 └─── B Frame Size (Min/Ave/Max): 20,160 / 71,198 / 134,784

Flow Motion 2.0 ├─── I Frame Size (Min/Ave/Max): 130,560 / 470,701 / 553,344 ├─── P Frame Size (Min/Ave/Max): 72,576 / 72,576 / 72,576 └─── B Frame Size (Min/Ave/Max): 22,464 / 135,736 / 219,648

1080p @ 30 FPS ETC No adverse Affects └─── Video Frame Size (Min/Ave/Max): 37,632 / 140,530 / 364,800 ├─── I Frame Size (Min/Ave/Max): 132,864 / 288,871 / 339,072 ├─── P Frame Size (Min/Ave/Max): 89,280 / 223,791 / 364,800 └─── B Frame Size (Min/Ave/Max): 37,632 / 92,415 / 154,560

Flow Motion 2.0 └─── Video Frame Size (Min/Ave/Max): 36,864 / 243,366 / 405,120 ├─── I Frame Size (Min/Ave/Max): 91,392 / 380,151 / 405,120 ├─── P Frame Size (Min/Ave/Max): 48,768 / 48,768 / 48,768 └─── B Frame Size (Min/Ave/Max): 36,864 / 175,868 / 254,016

720p No adverse Affects └─── Video Frame Size (Min/Ave/Max): 11,904 / 70,332 / 335,808 ├─── I Frame Size (Min/Ave/Max): 120,000 / 305,347 / 335,808 ├─── P Frame Size (Min/Ave/Max): 11,904 / 62,282 / 121,536 └─── B Frame Size (Min/Ave/Max): 0 / 0 / 0

Flow Motion 2.0 └─── Video Frame Size (Min/Ave/Max): 9,792 / 145,704 / 394,560 ├─── I Frame Size (Min/Ave/Max): 94,464 / 367,384 / 394,560 ├─── P Frame Size (Min/Ave/Max): 21,120 / 139,660 / 246,912 └─── B Frame Size (Min/Ave/Max): 9,792 / 92,426 / 131,520

1080p @ 24 FPS No adverse Affects └─── Video Frame Size (Min/Ave/Max): 21,696 / 226,456 / 798,912 ├─── I Frame Size (Min/Ave/Max): 211,776 / 730,476 / 798,912 ├─── P Frame Size (Min/Ave/Max): 43,968 / 341,631 / 547,392 └─── B Frame Size (Min/Ave/Max): 21,696 / 120,280 / 165,504

Flow Motion 2.0 └─── Video Frame Size (Min/Ave/Max): 88,512 / 483,307 / 869,184 ├─── I Frame Size (Min/Ave/Max): 287,616 / 832,219 / 869,184 ├─── P Frame Size (Min/Ave/Max): 227,136 / 227,136 / 227,136 └─── B Frame Size (Min/Ave/Max): 88,512 / 310,657 / 372,864

1080p @ 24 FPS ETC No adverse Affects └─── Video Frame Size (Min/Ave/Max): 77,952 / 226,825 / 583,488 ├─── I Frame Size (Min/Ave/Max): 287,616 / 531,500 / 583,488 ├─── P Frame Size (Min/Ave/Max): 126,528 / 333,531 / 402,624 └─── B Frame Size (Min/Ave/Max): 77,952 / 148,861 / 213,120

Flow Motion 2.0 └─── Video Frame Size (Min/Ave/Max): 231,936 / 502,954 / 734,400 ├─── I Frame Size (Min/Ave/Max): 280,896 / 693,956 / 734,400 ├─── P Frame Size (Min/Ave/Max): 312,000 / 312,000 / 312,000 └─── B Frame Size (Min/Ave/Max): 231,936 / 409,365 / 587,712

mm, i must ad this.
shooting in stilmode, and taking pictures when filming go's without probs in pal.
BUT.... when you playback the file on camera, the file freeze on the moment where the first picture was taken.
Just pressing the stop button, brings the camera back to live.
The file is further oké, when editing in post.

@jrd In 24H video mode, Flow Motion v2 will automatically shift into 140Mbps turbo mode whenever a peak bitrate higher than 100Mbps is required. This occurs most often when using 80% Slow-Motion 24H mode (which uses 25% more bandwidth), but can also occur when shooting well-lit, highly detailed subjects in regular 24H video mode.

In my testing, I found that 24H peak bitrates higher than 140Mbps would not record reliably on Class 10 SD cards and offered no significant improvement in perceptible image quality. With the 3-frame GOP used in Flow Motion v2, the I-frames produced by a 140Mbps peak bitrate are comparable to an all-Intra encoder running at 280Mbps.

@peternap I've tested Flow Motion v2 in low-light underexposed by as much as two stops. In those conditions it maintained optimal image quality, typically with average bitrates in the 40-50Mbps range.

@LPowell

It's a frequently asked question, I know, and you're bound to get is sooner or later, so here goes: is there anything to be gained -- or lost, other than card space and reliability-- if the user raises the bit rate for 24p? And would that change actually lead to higher frame sizes/less compression? Thanks.

Everyone is faster than I am. I spent the last hour testing it against Sanity 5 with different lenses in 720-60. It flat kicks butt. No errors although I don't get them with sanity either.

@mpgxsvcd Thanks for your reliability testing of Flow Motion in 720p60 and HBR 30p modes. Please post an example of a Stream Parser chart that shows the encoder entering Fallback mode, along with a frame grab if possible. "Fallback Mode" is not merely a temporary dip in bitrate of the type seen in the Stream Parser chart 00008 that you uploaded above. Stream Parser specifically identifies Fallback Mode malfunctions in its elementary stream reports.

One point I haven't mentioned is that it's important to use the camera to format the SD card each before shooting a series high bitrate test footage, in order to guard against SD card memory fragmentation.

In evaluating Stream Parser bitrate charts, bear in mind that bitrate is only an indirect measure of motion picture quality and can vary dramatically as the subject matter or camera position changes. Without reference to a relevant frame grab (or MTS file) of the original video, there's no way to evaluate the context of a Stream Parser bitrate chart. A more direct gauge of encoding quality can be obtained by using Stream Parser to generate an elementary stream file of the video, which will list the base DC quantization factors used to encode each frame.

@duartix

Isn't the point of these settings to prevent it going into fallback? I thought that was what the optimized quantizers were for?

Is NTSC HBR spanning supported?

@mpgxsvcd what setting is this? 1080p @ 30 FPS , mjpeg hd?