Personal View site logo
Make sure to join PV on Telegram or Facebook! Perfect to keep up with community on your smartphone.
AVCHD maximum image quality settings and testing
  • 1002 Replies sorted by
  • @B3Guy That double image you see in your clip of you friend cleaning his teeth is a type of lens flare...You will see that especially if you have a filter over the lens.....What lens were you using?
  • @Ralph_B - I'm more than rejoicing, we shot a couple of pickup shots with the 132 setting today... just looking at the dailies... amazing.
  • @Chrispy Konica Hexanon 40mm 1.8 aka "Old Softy". It doesn't seem like a flare thing at all. To crisp and precise, very "digital artifact". but this is off topic. I have opened a new topic here: http://personal-view.com/talks/discussion/870/low-gop-stutter
  • Looks good Shrigg!
    Man, that guy reminds me of Jerry Garcia. lol
  • One interesting effect from the higher bitrate settings... the Ex teleconverter looks worse and worse to us, the more data thrown at it. I think the stock bitrate helped blur/smooth it before, and now we're seeing the warts and all. Other have a similar experience?
  • @cosimo_bullo
    Unfortunately, I also noticed this ailment. I thought that just seems to me, but in the mode ExTele, banding is more visible even at ISO160.
  • In ExTele mode there is no pixel binning. This certainly illustrates the advantage of pixel binning.
  • @Mihuel It would be weird if the fixed banding some of us are seeing appeared in the same place in ExTele mode. You're not seeing that are you?
  • @cbrandin
    Higher bitrate reveals, unfortunately, sensor failures that are masked by the technique of pixel binning. It makes sense for me.
    @sam_stickland
    Why weird ? Do not kid yourself, we pull out of this machine more than predicted by its designers as we pulled away imperfections assembly line. Nothing is perfect, and even more GH2 :-)
  • @Mihuel - because when you use ExTele mode you read off a different smaller section of the CMOS sensor. If the FPN noise was the same in ExTele mode it would suggest it's being generated by something else (possibly later in the chain or perhaps from some interaction with lighting). I'm going to test this and post the response in the banding thread.
  • @sam_stickland
    (possibly later in the chain or perhaps from some interaction with lighting)
    ok, but I saw FPN with sunlight. Testing its a good idea, I will be happy to read your conclusions from the tests in the banding thread. :)
  • This may be an annoying question, but here goes: Chris, I was away from personal-view for a while, and so lost track of this thread; I don't feel like digging through the whole thing to find the most up to date max-image quality settings you propose, so, I was wondering if you could put this on the first post of the thread? Maybe that way, it will be easy for people to see what is up to date without having to read through everything? Just a thought. Thanks for ALL your hard work!!!!!!!!!!
  • @Vitaliy_Kiselev IDK if this is a good idea or not, but I thought it could be useful if there was a spot on each forum member's page where they could put their personal settings. This way, if I want to see what settings Chris or Kae or myself or anyone else is currently using, I can just go to their page and see (as long as they keep it up to date, of course)
  • @B3Guy

    Yep, I have such interface prototype, but I need good PHP coder to make this work.
  • Yes that would be a great system where it would be like a repository of the latest setting from testers with a description, etc.
  • To come back to the talk of which settings work best for you, I come to the following conclusion:

    cbrandin's setting 66 (in all different versions (i50 and p24)) crash on my gh2. A message states that the cam ran out of bandwidth and I have to remove the battery to shut it down.

    stable, and visually best results (IMHO) I get with cbrandin's 44/32-settings, modified to AQ3 and GOP6. I just have a feeling GOP6 produces less blocks than GOP12, though I don't have software to proof that. Just visual comparing.

    Picture Profile I found the best as:
    Nostalgic
    Contrast +2
    Sharpness -2
    Saturation +2
    NR: 0 (with -2 I think I had much more noise – fine grain – but it is better for me with 0)
  • @neveraholiday can you show us some streamparser graphs of the 44M running at AQ3 and GOP6 of shots around a couple of minutes long. Would really like to see if the stream cadence issue happens. If it doesn't I, and quite a few other people probably, would be very keen to have such a low bitrate GOP 6 setting.
  • I've been casually following along and it seems things are getting a bit convoluted. Who's got a good set of settings covering all the various modes with some decent bitrates and image quality tests to back them up?

    John
  • @neveraholiday,

    Although I have not spent much time with other modes, Chris's 24H (1080 24P) 66M AQ2 setting has never failed for me. You are the first person to mention Chis's 24H 66M AQ2 setting failed. What SDHC card are you using?
    Sounds like a card issue to me.
  • I use Sandisk SDHC Extreme Pro 32 Gig with 45MB/s. Fastest you can get. As for Streamparser, I didn't install that yet cause I don't have much time. And it will take time until I find time to install it. So you'll have to wait, sorry. Anybody else who crashes on 66?
  • @cbrandin Did you ever try IQ=0 and Q=0 and what is its AQ equivalent? I only mention this cus other codecs when QP is set to 0 allows the encoder to go off and get the best possible Q. My tests show very high frame sizes but it needs harnasssing.
  • unless its the default (everything unchecked?) however, IQ =20 so it isnt the default.
  • @neveraholiday I use the same cards (though only a 16Gb and an 8Gb), and Cbrandins 66M setting is what I use everyday and I've never had a single problem. I've also used driftwoods 132M GOP3 with no problems too. Are you sure your card is genuine ? I also used these settings on slower Extreme SDHCs with no issues too.

    edit : you are formatting the card regularly ?
  • @driftwood

    Setting a very low Q value doesn't work with real time codecs. You might be able to do that if you don't care how long encoding takes, but that's not the case with a camera. The possible Q values has to be limited because there simply isn't enough processing bandwidth available to actually try all possible values, and strange fallback things will happen.

    Chris
This topic is closed.
← All Discussions