Personal View site logo
AVCHD maximum image quality settings and testing
  • 1002 Replies sorted by
  • @neveraholiday

    I'd be suspicious of your cards too. I use the cbrandin 66 AQ2 settings with no problem on a Transcend Class 10 32Gb card.
  • @neveraholiday

    I also think it's your card. I have the same card and those settings work fine for me. There are a lot of counterfeit cards out there.

    Edit: I was wrong about using the same card as neveraholiday. Mine is the 32 gig Sandisk class 10 30 mb/s card and it works flawlessly.
  • don't know how many times this has to be gone over. the 45mb/s card is suspect. The gh2 can't take advantage of it's superior speed. It has inconsistent results on gh2 because some of the write speeds are really slow- on the gh2. They are not optimized to work together.

    Users have gotten away with using it at lower bitrates, not at these huge bitrate settings. There is nothing faster with the consistency required than 30mb/sec sanddisk.
  • @chauncy Yep, saw the thread. However, a lot of us are using these cards on high bitrate settings and we're definitely using the bitrate we've set (streamparser graphs), and for shots of good lengths too. So, suspect or not, the majority of us using them are having absolutely no issues with them at all. So. I see little point going over it again either, especially as there is a thread all about it, and you're the only person I can see questioning their ability whereas most of us have had no problems with them. Apart from neveraholiday who else has had problems with them ?

    Edit: Don't get me wrong, as a result of the great information and research on the other thread I won't buy any of these cards again, I'll get something much cheaper. However experience so far shows that the write speed is definitely good enough for the bitrates we're running at for most of us. In fact I think the extreme III cards that get the nod in that thread are probably over specced for the 66M settings as these 45M Extreme Pros are definitely handling it.
  • Chauncy, I'm with you on this- stick to the Sandisk Extreme 30MB/s SDHC cards.
    The 45mb/s cards may be just "hanging" on to working under those conditions. Maybe it's a random write speed issue.
  • I have a problem with the 66mb setttings. Works great with Sandisk Extreme Card 30MB/s at 1280x720 50p and 24p, but fails when using FullHD 50i every time. I am not shooting bushes of death either. I wonder if people aren't testing the 50i settings as they aren't usually very popular. Are most of the tests done with 24p?

    What's even more weird is that I just tried the 66 with a class 6 card SanDisk Extreme 20MB/s card and it records with 720p and 24p but not 50i. Weird indeed.

    So for me at least 66Mb is not working. I have gone to the 44Mb settings and these work on class 6 cards too
  • @rsquires I don't think the 50i settings work at all, I have had similar problems. I don't use them, and yes I figure you're right they aren't being tested by people.
  • @stray It would be nice to know that really. I want the camera to work in all modes so I have flexibility in camera. Obviously one could have a load of saved firmwares for the different settings that allow for different scenarios. But as far as these threads go it would be great to see some disclaimers here or there.
  • @rsquires
    I assume you are referring to cbrandin's 66M settings ...
    If so, also load cbrandin's 44M settings.
    Copy all the settings relevant for 50i from the 44M set to your 66M set.
    Works trouble-free for me in all modes (I also use 50i a lot).
  • @Vitaliy_Kiselev I'm happy to help with the PHP coding. It's been a few years since I've coded in PHP (I've since moved onto to Ruby/RoR and Python) so I may have to study whatever framework you're using (if any), but it sounds well within my ablilities.
  • 720p Latest: Having done nearly 120 720p/1080i50/i60 combi tests most of which eventually fail either pappas/bush of death tests, buffer tests, we are still some way from finding ideal settings... the search continues.
  • A large proportion of the blame for obtaining decent 720p is that it is using a variation of the codec 'AVCHD Lite' which is of course designed for less powerful cameras. Basically this means its recording 50/60 fields NOT frames per second inside the 1080i50/i60 wrapper. This equates to 25 frames & 30 frames respectively NOT 50p/60p!

    Therefore, I suspect that we will not be able to push 720p too much unless Vitaliy can shed any lite (pardon the pun) or confirmation that there are two versions or a bastardised 24p/ Lite version of the codec running on the GH range.
    Switch your camera over to 720p and notice 'AVCHD Lite' appear in the info screen.
    See http://shedworx.com/avchd-lite-50p
  • @driftwood Sorry asking this, but what is you current best image quality settings? I would like to try it on my next work. Thanks!
  • Driftwood, how high were you able to push 720P?
    We may have to live with modest settings of 42Mb/s or lower e.g 32Mb/s.
    This is still far better than 17Mb/s stock.
  • Its nearly there... I know the correct buffer setting working on the minimum GOP and bitrate. I hope to have a 100% stable setting soon... Don't expect massive results.
  • @driftwood
    re: http://shedworx.com/avchd-lite-50p

    Maybe I'm misunderstanding the article, but it seems to be saying that there is no true 50p/60p on the GH2.

    However, I just did a quick check of some 720p60 footage I have on my laptop and found that if the shutter speed is longer than 1/60 the footage consist of repeated frames, effective frame rate therefore 30 fps.

    But at faster shutter speed, e.g 1/250, the camera definitely delivers 60 fps. Here's a clip shot at 720p60 that shows definite movement at every frame. Go to the Vimeo page and download the MOV - it's just a re-wrapped copy of the original MTS:

  • @bdegazio I guess it comes as a bit of a bombshell - it did to me. Nope its AVCHD Lite implemention. Check your camera - select 720p - see the message scrolling upwards - AVCHD Lite. Poor man's version of 720p50/60. Its fields NOT frames you're seeing - doesnt matter what you think you're seeing - its repeated fields. This means you are seeing 25 frames and 30 frames respectively. However, we're gonna do our damndest to get the best bitrate to buffer ratio sorted. :-)
  • Sorry Driftwood, but you're wrong. The camera says "AVCHD Lite COMPATIBLE", not "AVCHD Lite".
    Go here for a clear definition of AVCHD Lite:

    http://en.wikipedia.org/wiki/AVCHD

    AVCHD Lite is 720 30P, that's all.

    Rest assured the GH2 is giving you a true 720 60P.

    And what "compatible" means is that AVCHD Lite software can playback 30P from the GH2's 60P stream.
  • @towi I am using the 44mbit settings and they seem very stable in all modes. I'll try a hybrid 66/44 see how that goes
  • @driftwood:

    720p50/60 delivers real 50/60 fps non-interlaced. It's encoded without B-frames by default. By setting "Encoder setting 1 720p=3" you can make it record real B-frame 720p.
    Drawback is that you have to record at least a second of 1080p before 720p (probably to initialize something)

    My tests show that stock-GOP B-frame 720p delivers absolutely stunning picture, with spatial resolution akin to Canon 1080p. Even better, you don't need high bitrates to achieve good IQ. It is very effective and 40Mbit/sec is plenty enough.

    It would be very nice if Vitaliy could check what needs to be initialized before recording 720p with B-frames, so we can avoid shooting 1080p first.
  • Grunf >It would be very nice if Vitaliy could check what needs to be initialized before recording 720p with B-frames, so we can avoid shooting 1080p first.

    I will buy the GH2 again when this happens, until then the GH1 720p V2 75Mb peak is very good and doing the job.
  • All my tests for 720p are trying to keep it to P frames - and to work without having to record 1080p first. Still doing loads of tests... spreadsheet getting longer...
  • Can someone post here quickly their streamparser and info chart with best so far found settings under 720p - I would like to analyse it - test its stability - and I want to test my latest setting against it. I think we're getting somewhere. :-)
This topic is closed.
← All Discussions