Personal View site logo
100Mbps Flow Motion v1.11 Failsafe Patch with HBR 25p & 50p modes
  • This patch has been upgraded and superseded by Flow Motion v2

    For more info, see the following link:

    GH2 100Mbps Flow Motion v1.11 Failsafe Patch with HBR 25p & 50p modes

    Released below is a minor revision to Flow Motion v1.11, tuned and tested for failsafe operation in all PAL video modes, including ETC zoom mode. In addition, it also supports failsafe operation in all NTSC modes except for ETC mode in HBR 30p and FSH 60i. As an alternative to using ETC in HBR 30p mode, I recommend using ETC with either 80% Slo-mo 24H mode, or MJPEG HD mode, both of which produce high quality results at 1080p30. To download the Flow Motion v1.11 Patch INI file, select the last attachment at the bottom of this post.

    In short, this is the 100Mbps GH2 version of my Reliable In-Camera Playback Patch.

    Congratulations once again to @Vitaliy Kiselev for cracking Panasonic's recent firmware v1.1 update for the GH2! PTool v3.64 now provides support not only for the new HBR 1080p25 and 1080p30 video modes, but for custom Quantizer Scaling Tables as well. Custom tables enable patch developers to tailor the compression properties and image quality of each of the AVCHD 1080p, 1080i, and 720p video resolutions. In addition, @cbrandin's invaluable Stream Parser and Scaling Table development tools have made it possible to quickly take advantage of these new PTool features, providing an unprecedented level of control over the GH2's video recording functions. Combined with the capabilities of Panasonic's v1.1 firmware, this has resulted in numerous improvements to the reliability and motion picture quality of Flow Motion v1.1. Here is a concise list of its features and specifications:

    • 100Mbps AVCHD in 24H 1080p24 and SH 720p50/60 at 8-10 keyframes/sec.

    • 75Mbps AVCHD in HBR 1080p25/30 and FSH 1080i50/60 at 8-10 keyframes/sec.

    • 50Mbps AVCHD in 24L 1080p24, FH 1080i50/60, and H 720p50/60 video modes

    • 100Mbps MJPEG 1080p Patch maintains high bitrates regardless of light levels.

    • Full support for all NTSC and PAL video modes.

    • In-camera playback of all AVCHD video files. (MJPEG HD videos not playable in-camera.)

    • 4GB AVCHD file-spanning in all video modes with 95MB/sec UHS-1 SD cards.

    • File-spanning in 24L, FH, and H modes supported on all Class 10 SD cards.

    • ETC zoom mode with failsafe operation in all modes except NTSC HBR 30p and FSH 60i.

    • 80% Slo-mo 24H video mode with ETC and file spanning in 24L mode.

    • Combines optimal I-frame size with compact, 3-frame GOP file size (GOP-6 in 720p modes).

    • Customized Scaling Tables produce P and B-frames that match I-frame quality.

    As with any complex video camera, there are a number of considerations and limitations to bear in mind:

    • Listed bitrates are the maximum seen in practice, filming well-lit, sharply-focused, highly detailed scenes. Average shots will produce average bitrates. In low-light conditions, I recommend trying MJPEG HD 1080p30 mode for its ability to maintain high bitrates in near dark conditions.

    • If shutter speed is set longer than the frame rate (e.g. slower than 1/60 at 60p), low-quality video files may be produced. High shutter speeds will reduce motion blur, which may force the encoder to operate at higher bitrates.

    • ETC zoom mode increases the bitrate and puts significant stress on the encoder. With Flow Motion v1.11, Failsafe ETC mode is now supported in all PAL video modes, and in all NTSC video modes except HBR 30p and FSH 60i modes. ETC actually does work in all modes, but I've found that with Panasonic distortion-correction lenses, it's not 100% failsafe in HBR and FSH modes under the most extreme conditions. With manual-focus vintage lenses, it has so far proven to be failsafe in all modes.

    • Flow Motion v1.1 has been tested extensively at ISO 3200. It is not tuned to work reliably with PTool's extended ISO patch, but higher ISO can be used reliably in underexposed, low-light conditions.

    • Panasonic Lumix distortion-correction lenses are now fully supported in Flow Motion v1.11. For smoothest results, I recommend using these lenses with a mild diffusion filter to reduce their over-sharpened edge artifacts. My favorite is the Tiffen Black Diffusion FX3.

    • The 24L, FH, and H video modes make use of a special "Trick Mode" that ensures reliable file-spanning at 50Mbps. It does this by setting the video file's transport stream bitrate to a much lower level than the video stream's bitrate. These files play back correctly in video editors with no need for post-processing. However, MediaInfo and Windows will report incorrect duration and bitrate for these files when examined on the desktop.

    • I recommend using the camera to format your SD card at the start of each shoot, to guard against SD card memory fragmentation.

  • 574 Replies sorted by
  • I'm not a big fan of high bitrate settings, but I will give it a try this time. Glad to see Lee back in action :)

  • @Davidz Sorry to hear of your spine problem, I recently had a lumbar fusion myself, and have been steadily recovering my stamina over the past two months. As @matthere pointed out, I've released Flow Motion v2, which supercedes v1.11 with a number of significant improvements. I hope you find it meets your needs.

  • @LPowell To keep this brief, while we're all waiting for V 2.0: 1) I have no idea why, but I am again shooting Flow Motion v1.1 at 1080 60i without any crashes, and with great results! and: 2) after again attempting to shoot with many very high data rate hacks, such as Driftwood's VY Canis (can that be right, with a 64 GB card, a total recording time of 3 minutes 57 seconds???!!!) in what I've been using exclusively, Creative Movie Mode, I tried them all out in HDR, with nary a crash! I had never tried HDR until I read your post (actually, I spent over a day carefully reading ALL twelve pages of this thread) indicating that HDR was 1080 60i - I also note your attitude (is disgust too strong a word?) about interlaced. I have been shooting in 1080 60i for the resulting 1080 detail at 30fps.

    @towi I have done very preliminary testing of your latest, modified Sanity v3, and the results certainly look very good.

    All of my detailed testing of modified Sanity v3 and many other hacks will have to await what I hope is much greater recovery from my latest fractured vertebrate, the pain of which has now become unbearable.

  • The Intra 'setting' used for the HBR recording is simply frame limited.

  • @mozes What Stream Parser refers to as "skipped" macroblocks are instances where no data is encoded for a particular macroblock. What happens in skipped macroblocks is that a section of the image from the previous frame remains visible in those macroblocks. This can occur in benign cases where an area is completely blown out to white, with literally no image details to encode. In pathological cases, however, it can occur when the encoder is either starved of bitrate, or fails to accurately predict the bitrate requirements of the frame.

  • I am sorry to go off topic, but i hope you can help me with this.
    its shot in 3200 iso
    What are meant with skipped?...

    ps. screens are not from FM .....

    658 x 916 - 77K
    746 x 412 - 53K
  • Thank you so much for explaining this @LPowell

  • @LongJohnSilver No problem, I just wanted to make sure no one misinterpreted your intention.

  • @Lpowell just to clarify: I never told this was shoot in FM. Actually I clearly wrote it was shot all-intra.

  • Just to clarify, the video above was not shot in Flow Motion; it is an HBR 1080p30 file using an all-Intra patch. If you download the original MTS file and examine it in Stream Parser, you'll see something like this:


    This is an example of what is sometimes euphemistically called a "CBR patch", as it certainly does display a constant bitrate. The reason for that, however, is because after the initial I-frame, the encoder immediately collapses into a crippled, low-quality Fallback Mode with extremely unbalanced macroblock quantization levels. An examination of the detailed Stream Parser report reveals QP factors ranging from 18-43 within each frame, which indicates pathological encoder operation. In addition, the reported DC value of 2 indicates the quantizer Scaling Tables are being flattened into gross amounts of overkill. (As cbrandin has explained, once DC value drops to 4 or lower, the encoder is forced to waste bitrate on excessively precise data that produces no visible improvement in image detail.)


    For those who have Elecard Streameye, a detailed examination of individual macroblocks shows what is amiss with this patch in HBR mode. As in all HBR videos, the encoder divides each I-frame into an interleaved pair of fields, an I-field and a P-field (which is derived from the I-field). In order to preserve optimal static image quality, the P-field should be encoded at the same average quantization level as the I-field. In this case, the P-field (blue lined) is encoded with a QP range of 20-26 (good quality), while the I-field (red lined) is encoded over a QP range of around 32-48 (poor quality). What's also disturbing is that the quantization levels get coarser toward the bottom half of the frame, indicating that the encoder overestimated its available bitrate, and was forced to reduce the bitrate with coarser quantization before it was able to complete the frame.


    While this dysfunctional behavior is not an inherent drawback of all I-frame patches, it highlights the compromises that 24p Intra-frame patches impose on HBR video mode on the GH2. This dilemma is due to the unfortunate fact that a number of patch settings are shared between 24H and HBR mode, in particular the Intra-frame GOP length of 1080p24 and 1080i50 modes. This is one of the reasons I concluded that the 3-frame GOP used in Flow Motion produces optimal motion picture quality in the broadest range of GH2 video modes.

    Stream Parser Report.jpg
    990 x 490 - 231K
    Stream Parser Details.jpg
    835 x 390 - 173K
    StreamEye Macroblocks.jpg
    960 x 540 - 763K
  • I had bad experience with the HBR 30p and intra frame settings. it's not the settings itself which gave me wonderful results in 24p but the limitation of the current HBR implementation. Look at this, password is crap and you can dwnld the original mts

  • @LPowell

    Man... I am really looking forward to testing FM2!!!

  • @Davidz

    Sanity 3 is different from Sanity 4 or 5. I've used Sanity 3 for quite a long time for 24H and FSH/50i modes. It never faild. It always delivered extremely good looking footage. Really!

    If Sanity 3 (the updated version for the latest PTools version I've attached above) works for you, you may also try this modification of Sanity 3: . In my book it looks better (somewhat sharper & nicer noise) than the original Sanity 3 patch. This setting survived several torture tests, however I can't yet promise it's stable under all cirmumstances... Above all - I'm shooting in PAL mode so I can't speak for NTSC modes (60i/30p...), but I assume it will be good for NTSC as well...

  • @LPowell , WOW, that looks great, much less (almost no) macro blocking , can't wait to test it out. Hope a HBR mode get also some improvement ...

  • @feha Flow Motion v2.0 is in the final testing phase, and I'm preparing sample footage. This takes a while, because any change in settings requires another cycle of reliability testing. (Some of the test subjects are located several miles away.)

    Here's a preview of the motion picture quality issues I've been optimizing in FM2. This is a shot of a shaded stream of water with a moderate distribution of specular highlights. This type of subject is deceptively brutal, requiring the encoder to track countless independent motion vectors in each frame. On the top is a 200% crop of a 24H mode frame grab from the unhacked GH2 AVCHD encoder. You can clearly see macroblock encoding artifacts due to the stock firmware's inadequate bitrate. Below it is the same stream shot with Flow Motion v2.0 in 24H mode:


    Creek GH2vsFM2-200.jpg
    1024 x 868 - 405K
  • @LPowell , any news on Flow Motion 2.0 ?

  • @LPowell May 20 ...sunrise ?

  • Gentlemen:

    Thank you sooooo much for your helpful responses. Regret the delay in getting back, but I've been involved in some long overdue bookkeeping (ugh), and simply lying down, just beginning recovery from a lumbar vertebral compression fracture of a week ago (ouch).

    @towi - thank you for the PTools Sanity V3.1 attatchment. After I've had a chance to throughly test out four hacks I've already downloaded to cards ready to install in the GH2 (including Cake, that I've heard so much about, and a couple of Sednas) I'll give the Sanity 3.1 a run through. Incidentally, I had already earlier tried out the Sanity 4.1, and along with many other hacks, it was an immediate crash.

    @LPowell - I, as many others, have been watching and waiting for the v2.0. I find your comments about the demands of 1080 60i FSH very interesting and informative, in view of the frequent and almost insurmountable problems I've had in getting 1080 60i FSH to work at all with any of the hacks for this camera.

    My associates, who all shoot in 24p have found my issues a total puzzle - many of them are using the GH2 as a primary, secondary, or tertiary camera, with hardly nary a problem. And incidentally, when I ran my last tests with the Flow Motion, I was very pleased with everything that I successfully took in 1080 60i FSH, but I very quickly gave up on the HDR setting, since though the shots were often as sharp as the 1080 60i FSH, the HDR exhibited some very serious banding problems, the likes of which I've never seen before from any of my other cameras or any GH2 hacks set to 1080 60i FSH!

    If it's of interest to any of you others who are attempting to shoot 1080 at any data rates, I'll post here some summary results of what has worked (if anything), and what hasn't.

  • @LPowell i am curiosity how you have fixed those p-frames.

  • Sounds very nice, Lee! :)

  • @LPowell , i can't wait to test FM 2.0 :-)

  • @Davidz Flow Motion v2.0 is almost ready for release, and I think you'll find it a significant improvement in both reliability and motion picture quality. I've been subjecting it to a variety of brutal test cases, and have found that in some circumstances the bitrate demands of FSH 1800i60 mode are even higher than HBR mode. In addition to its high data rate (due to operating at 30fps), the internal P-field that is embedded in each FSH I-frame requires finer quantization to maintain image quality comparable to I-frame macroblocks. Balancing these requirements with failsafe reliability in the complete range of video modes has turned out to be one of the most challenging aspects of optimizing the GH2 for operation at 100Mbps.

  • @Davidz

    for 60i & 50i I also opt for Sanity. Me personally for Sanity v3.1 rather than V4 or V5 (I use V3.1 when I have to shoot 50i) ...

    Sanity v3.1:

    Sanity 4 & 5:

    edit - attached Sanity V3.1 for PTools 3.65d (latest version)

    Sanity V3 PTools