Personal View site logo
Make sure to join PV on Telegram or Facebook! Perfect to keep up with community on your smartphone.
Please, support PV!
It allows to keep PV going, with more focus towards AI, but keeping be one of the few truly independent places.
GH2 Flow Motion v2 - 100Mbps Fast Action Performance & Reliability for Class 10 SD cards
  • 1036 Replies sorted by
  • @LPowell Well I just came back to my camera and it seems to have filled up the card with no problems in 1080i50 on ISO3200 at 100Mbs :D I understand what you mean but the only 'failure' that happens is when recording at ISO12800, but I'm glad you've pointed out that it's a poorly implemented feature (how did you know this?). So recording at 100Mbps seems to be fine afterall.

    Also to let you know THIS is the card I own and bought from: http://www.amazon.co.uk/SanDisk-Extreme-16GB-SDHC-Card/dp/B005LFT3MA/ref=sr_1_1?s=computers&ie=UTF8&qid=1340579623&sr=1-1

    Regarding the AVCHD encoder, I was pretty sure it did do P-frames, wouldnt a I-P-B frame structure weild better results? And either way, isnt it possible to implement this as well?

  • @GH2_fan Yes, that is a 95MB/sec Class 10 SD card, and it's good to hear that you were able to confirm file-spanning with Flow Motion v2 at 100Mbps. ISO 12800 is an extremely noisy setting that I'd recommend against using in general. Closing the current MTS file and opening a new one in the middle of a high-bitrate recording is a fragile process. It's unfortunate that the GH2 and most other SD card cameras persist in using an archaic file system with a 4GB file-size limit that forces the camera to juggle files in this manner.

    Flow Motion v2 1080 modes are designed to use a 3-frame GOP with very fine B-frames whose image quality is indistinguishable from its I-frames. There would be no value to downgrading its encoding efficiency with P-frames.

  • @LPowell Yes that is the card I have been using the whole time, especially when I was reporting my problems here and indeed it is unfortunate that the file system they persist in using is FAT32 but I would consider this minor, the file-splitting thing is not a biggie for me when it comes to capturing :D

    Not too long ago I have taken my newly recorded Flow Motion v2 1080i50 footage into After Effects, and actually I am able to distinguish the frames from the I-frames and it doesnt look too pretty (just being honest!) This is probably why I favour Intra-frame recordings and went with Driftwood's patches before but was having imense problems with them. This is also why I mentioned of the P-frames, I think it would noticably help the image quality as it contains motion-compensated difference from the I-frame. The B-frames is what I was referring to them not looking too pretty by the way.

    I know this may seem quite abit to ask, but is it possible to make a Intra-Frame Flow Motion where it is able to playback incamera and for it to span successfully in all modes? Or even keeping your GOP3 but the structure at I-P-B?? Right now I'm going to see if the camera will span completely (filling up my SD card) on 24p ISO12800. This should be interesting :D

  • @GH2_fan I took some time to answer your persistent questions in detail, and your response is to take a cheap shot at my work with nothing tangible to even show what you claim "doesn't look too pretty"? Pardon me as I intend to use my time more wisely.

  • @LPowell No no no you might've misunderstood me, obviously I'm not here to offend anyone at all, especially you, I am greatly appreciative of your work, its probably sometimes how I word things or the way they come out (I'm dyslexic, but I often try not using this as an excuse), and not taking any cheap shots at you in any way.

    What I mean by "doesn't look too pretty" was to say that I was able to distinguish the frames from the I-frames from the B-frames, from what I did was zoom in 200% and slowly analyze the frames and noise and saw that get 'washed' away due to the compression, noticably enough that the fine detail (of the noise especially) was not as visible compared to the I-frames.

    This is probably how I should've explained it in the first place. Because of this all as well and trying to get the most from my (our) cameras, that is where I asked about the possibilities of making changes or improvements.

    I also come with some news now (which myself I took time to do to help you and manjy others), I have by now tried recording at ISO12800, 10000, 8000 and 6400 just now in 100Mbps 24p and none of them was able to get past 4minutes, and whilst I can 'preview' them in playback, it just plays 1 second before giving an error message stating something's wrong with the card. Now I will try ISO5000 to see if it will make any difference.

    But overall, I hope you are not upset or frustrated with me, that is not my intention whatsoever. Hope you understand :)

  • @peternap Thank you for posting your frame grabs on the previous page! That cannon firing is a stunning one-frame transition, truly an acid test of the camera's ability to track instantaneous motion. What is undoubtedly happening under the hood is that the encoder is using Intra-frame-type macroblocks to encode the completely changed areas of the canon firing, while using Inter-frame encoding on the rest of the B-frame. This is an example where the coarse B-frame Quantizing Tables used in the unhacked firmware would probably not be able to maintain I-frame image quality in its P and B-frames.

  • @Lpowell @driftwood

    I write here because I already posted here my video showing the codec suffering shooting a lake bottom through transparent water with small ripples. The previous clip was shot in VMM 80% mode. Yesterday afternoon I came back shooting the same subject. Unfortunately in the afternoon I found a completely different light and I could not replicate the same scene. Anyway I managed to get a couple of clips. I preloaded my GH2 with SEDNA A20 then I flashed the camera with FM 2.02. Both clips are in 24p cinema mode shot withthe lumix 20mm with a marumi polarizer. Unfortunately while I changed setting the light changed rapidly so the two clips are not very useful. @cbrandin streamparser shows that clips were recorder nearly a the top bitrate offred by these settings.

    This time it seems that things went better than before also if the light is more forgiving than before.

    Lesson learned: I throwed away HBR and I will use 24p cinema mode for this kind of scene. It's a shame but it's nearly impossible to spot flaws on the camera display and you'll discover imperfections once at home on a proper computer screen.

    Again, if you are curious on the codec behavior don't look at vimeo clips but you have to download the original files.

    FM 2.02 24p cinema mode

    SEDNA AQ20

  • I shot this last night to see how blacks would hold up with a single spot light. Did a little color grading in colorista II, which was my first time using it. The blacks looks great and I didn't really see much noise at all. take a peek and lemme know what you guys think.

  • @LongJohnSilver I downloaded your test footage from Vimeo and had a look. You're right about Vimeo's web compression - it does a very poor job of rendering the details that can be seen in the original footage.

    Could you fill in some more details on your shooting setup? Although the original Flow Motion v2 footage looks virtually flawless, it's not clear to me how the situation differed in the case of the Sedna AQ20 footage.

  • Hi Lee,

    I was shooting with a tripod and the 14-45 with a marumi polarizer on. I arrived ther with the sedna preloaded. I took a shoot and then I changed settings loading FM 2.02. In the meantime the sunlight changed a lot so I had to move the camera to find a similar spot.

    Hoping to get exactly the same shot as the previous time I have to go there in the morning. That part of the resurgence is well lit only in the morning.

    Anyway It seems to me that shooting in 24p cinema mode things get better.

  • @Lpowell "You're right about Vimeo's web compression - it does a very poor job of rendering the details that can be seen in the original footage."

    Of course the compressed video could never be as the original but expecially with the original MTS files, Vimeo struggle to compress them properly. Actually during the import Vimeo alerts you that the clip has a bitrate too high and that conversion could be poor. Anyway is always better than dropbox to share mts files ;)

  • @EH_Films Thanks for uploading your sample footage, I like the dramatic results you got with a single key light. The subtle noise levels in the near-black shadows look natural - I take it you didn't add or subtract any noise in post?

  • a total newbie here as I picked up a gh2 about a month ago and just added my first hack yesterday. I put on "quantum X Span my bitch up v2b" and i am using a sandisk Extreme SDHC card 32 gb 45mb/s. I tried it with both my panny 14-45 lens and my canon fd 50 mm lens. It froze up on each one when i tried to shoot windy trees in HBR mode resulting in taking out the battery. Also handled it fine in HBR mode while shooting my dog who was being fairly still while taping. so i then switched to 24 p cinema and it handled it fine on both lenses.I guess my question is: is this just a card issue. if so i will pick up sandisk extreme pro 95mb/s. Thanks for any help.

  • @kdurk Though your question about Quantum X patch reliability is off-topic in this thread, while you're here you might want to check out Flow Motion v2.02:

    http://www.personal-view.com./talks/discussion/3337/gh2-flow-motion-v2.02-spans-in-hbr-25p-24h-720p/p1

  • @Lpowell ......I kind of feel like an ass now. I did add noise in post. Its cinegrain 35mm 500T exposed. I've been trying to tack it on to everything I do seeing as it cost me 700 bucks to buy. I figure I try to get the most out of it I can. But I can tell you that its so subtle you can't really tell a difference with it off. I should have tagged that on. My bad. It was shot with the 20mm pancake for the hallway shot, and the nikkor 50mm 1.4. I tried all the modes I could including 1080p, HBR, 720p overcranked, and ETC mode for the super close ups at the end.

  • I'm very pleased with the results I get with Flow Motion 2.02. I tried several of the other hacks but like the look I get with FM. Here is a video from a concert a few days ago of a local acapella group called The Coats. Yes I know I was too far back to get good audio but they put the students in the front section and the rest of us in back.

  • @RockHunter Nice camera work, and I thought the audio sounded fine considering the acoustics. I downloaded the edited 1080p MP4 file from Vimeo to check out its 13.3 Mbps image quality in more detail. The one issue I'm concerned about is the banding on the blank gray wall behind the performers, which is noticeable in close-ups. Would you be able to upload any of the original MTS clips recorded by the GH2? That would clear up the question of whether the banding is due to the MP4 compression rather than the camera itself.

  • @LPowell Hi Lee, how you doing? I hope you are completely recovered from the surgery already. First of all, let me thank you for sharing your settings and your precious knowledge. And now the inevitable stupid question: You know, the peak bitrate for 24H and 24L in this setting is supposed to be 100Mb and 60Mb but when I load the ini file in Ptools it shows 95Mb and 32,5Mb. I supposed there's something working underneath that my stupid mind is missing...

  • @LPowell. Thanks for looking and for the comments. Actually I just redid the file and reloaded it. It seems that I forgot that I recorded in 720/60p but rendered in 1080. So I redid it in 1280 x 720. Perhaps that helped with the banding.

  • A question. Many of us post our videos on Vimeo. Vimeo has preferred settings that they want us to use in creating the files for posting there. https://vimeo.com/help/compression

    Those settings are:

    Video settings: H.264 (.mp4) 24, 25 or 30FPS 5000 kbps 1280 x 720

    Audio settings: AAC 320 kps 44.1 kHz

    We also know that Vimeo converts the files after we upload them. So to get the best quality video and audio, should we follow Vimeo's recommendations or are there some adjustments to these settings that will not only produce better quality than the recommended settings but will work sucessfully on Vimeo?

  • @RockHunter I routinely upload my original GH2 MTS files without concern for Vimeo's settings guidelines. As you know, Vimeo will convert each video to their own web streaming format, regardless of how the original was encoded. One of the reasons I use Vimeo is to distribute original GH2 sample files to those who wish to download them, and re-encoding the file would defeat that purpose.

    I downloaded the 720p version of your video and while it does look sharper, I still see some banding on the gray wall in the back. Would it be possible to make one of the original GH2 clips used to make that video available for download via Vimeo as well?

  • Sure thing. Here it is.

  • @RockHunter Thanks very much for the download. Your original Flow Motion v2 footage looks very clean, no perceptible banding on the gray wall at all :)

  • @LPowell Good to hear that the original .mts file was ok.

    I'm really happy with the results I'm getting with Flow Motion. It is way better than any camera or camcorder that I've used in the past.

    Since the raw file is clean, what settings should I use in processing to make the final product on Vimeo look as good as possible? I've uploaded files to Vimeo in many formats including .mts, .wmv, .mpg, mp4, .avi and Vimeo can manage to convert them (but not always) to their standards for posting. I would be interested in hearing what others have to say about this.

  • @RockHunter As long as your original video exceeds the quality of Vimeo's version, I doubt it makes much difference - other than the length of time it takes to upload the file to Vimeo.

This topic is closed.
← All Discussions