Personal View site logo
Official Low GOP topic, series 3
  • 1008 Replies sorted by
  • @toxotis I guess it's a matter of preference, however I should point out that those are from still photography guides, I work with DoPs with 20+ years experience and they all underexpose by at least one stop, and this advice is seconded in a lot of books on digital video cinematography. In good part that is because life-like skin tones are best generated when you keep that in 18% or lower (eg where a grey card would register), aka Zone 5 (maybe Zone 6). You go above that, which you would if you are exposing to the right, and you start to push skin tones into a zone where much of the colour detail is lost and they get washed out, "chalky" and unrealistic. Once they are in this zone it is difficult to pull them back in colour correct because you can't get back colour detail you never had in the first place (whereas if you are underexposed, you can always loose a bit). Anyway, this is my last comment on this topic because I realize it is off topic somewhat -- the connection I would make is that with GOP1 and high bit rates, you can rescue under exposed skin tones.

    You may have seen this already, but check out my video at
    for an example of how under exposing leads to proper skin tones -- it got crunched a bit more by vimeo, and yes, a lot of it pushed off to the left and goes to black, which might be a problem with landscape photography, but for video work, the golden rule, which I think most people do not realize, is expose for skin tones, which generally means under exposing one stop and, if necessary, fiddling a tiny bit with colour correct in post if it is too dark, which thankfully, we finally can do with driftwood's high bit rate low GOP patches.
  • @lolo High ISO = more noise = more work for the encoder => higher bitrate => more chance of a card speed error?
  • @sam_stickland ups, sorry, you are right i should analyse first the bitrate of the 2 vids, before jumping to conclussions
  • @Tobsen Np! I guess the bottomline with high-iso + high bitrate patch is that it can be used for a certain aesthetic.. but it might not be useful for everyone.
  • seAQuake test: Here's one for the great mathmaticians. I just used Panasonic's utility 'Restore AVCCAM' which recovered a first recording on the card 00000.mts which had failed to span to... wait for it... 8.59 Gb!!!! A strange anomolity or is there more to what is going on than we know on the Panny seafloor!? I mean 639mbps is like a UFO flying out of the water...

    btw The recording looks fucking amazing.
    639Mbps 8.59 mts file after restore.png
    984 x 828 - 113K
  • @driftwood Very strange and very interesting. I seriously want to see that 00000.mts file!
  • its a little personal - my girlfriend - nothing rude - but f2 on a 50mm macro - up very close - exposes all her age lines etc... so I dont think Ill be uploading it to Dropbox. But I would like others to try out Panny's software and report - record a 1080p24 using seAQuake and leave it till it hits the 4gig limit and stops by itself. Bring in card to computer and use Pannys AVCcam restore utility. Tell me what happens once it's restored to your hard disk. It could be an anamolity of course.
  • I was able to get 720p seaquake working stable and playing back in camera through my zoom autofocus tests with the 14-140 lens. I only use sandisk 30mb to eliminate that variable.

    Took aq down to 1 and lowered the frame limits for 720/60 by two thousand.

    This is the first high quality 720p patch that looks stable so far through this test.
    24p has always been a rock.

    I don't discuss interlaced or use it because it is evil.

    UPDATE: if in camera playback doesn't work, turning the camera on and off may fix it.
  • ah this thread is always full of goodies. yummy yummy!!!

    @driftwood that's like 2.5GB data for 30s duration. sounds great for a short commercial. if your GH2 passes out, see if it spits out a number. it might be "the" number from π (1998). haha. many thanks for your work.
  • @BenjaminDUPLAIX

    Remember the UHS-I is problematic, and the 30MB/S (video hd or not) seemed more reliable.
  • @Driftwood: I tried 50i with your newest patch - I don't have streamparser but - it seems like proper Interlaced footage rather than PsF. Am I missing something or how have you managed your PsF?

    Very interesting with the file restore.. I'll try it out during the day!
  • Oh man @randolfo never mind the hack, that is awesome music. BTW the video is great! Al
  • @FGCU A failed recording recovered to 8.59gb using Pannys restore utility.
  • Maybe the tool recreate P-frames between all the I-frames ...
  • @lenuisible That could explain why the file is roughly twice the size it should be.
  • @driftwood seAQuake is amazing.. It even looks good at 2x! :) Seriously though, it is insanely clean - interestingly it also shows the weaknesses of the camera (and 8 bit colour) - banding becomes more apparent in some conditions. I haven't tried to work in post with such footage yet - I guess it's easier to try and avoid banding in the shoot than to fix it in post.

    Leaving for Rome tomorrow morning, will shoot as much as I can with it!
  • I have done some careful pixel peeping comparing Seaquake to bkmcwd 176 gop3 q18. The image quality is nearly identical with the ever so slightest edge to bkmcwd. I am comaparing for noise quality and detail in the shadows as well as picture detail. All kinds of motion also look fine to me. What am I missing and what should I be looking for to make a more complete comparison?
  • If it looks good to you, that's all that matters. ;) Sometimes ignorance is bliss. Seriously, when you know how to look for a problem, you start seeing it everywhere. All of these high bit rate patch settings deliver excellent video quality. Practical considerations become more important than squeezing out the last ounce of video quality.
  • @Jspatz It wont be in the macroblocking of the dark areas. seaquake goes much lower unless bk has copied my encoder settings :-) For me INTRA is more preferable to GOP3 on the movement - its slightly smoother - just got the edge - sorry to say, but thats my opinion.
  • @driftwood @Jspatz @balazer bkmcwd did copy from your GOP1 176mb/s patch driftwood, I was reading from his post. So that clarifies the pixel peeping part.
  • *** NEW *** 'SPANmyBITCHup' INTRA Patch. This is the baby version of 'seAQuake' and lill' brother of 'AQuamotion v2', which will span and record at the lowest minimal possible settings for all modes with gopod quality video. Everything is GOP1 I frame bar 720p60 which is still GOP3.

    This patch should work on just about any card, still record decent pictures, whilst giving you very long record times. Tested on all modes plus EX TELE and 1080p 80% modes. Now PLEASE stop asking for spanning INTRA! ;-) Test away.
    Driftwood - 'SPANmyBITCHup' INTRA Spanning and Any SD card in the Type 6-10 range - setf.zip
    633B
  • @driftwood THANK YOU!!!! (for both). Now if I only hadn't rented out my entire kit for this week, I could try some of these insane patches. Can't wait!

    PS: Any idea if the spanning settings would work at 24L and SeAQuake at 24h with the new Ptools?
  • *** NEW **** AQuamotion' version 2. Better quality video with spanning - shorter recording times (as the better quality increases bitrate/video quality) than the newly released minimalist patch SPANmyBITCHup.
    @JDN ver 2 dedicated to you ;-) Includes better i frames than SPANmyBITCHup and indeed spans on all settings. 720p60 is GOP3 btw ;-)
    Driftwood 'Aquamotion' ver2 AQ2 all round and spanning - great for EX TELE and 80percent recording - better 1080p24 quality with spanning than SPANmyBITCHup - setb.zip
    756B
  • @JDN no chance of sharing between H and L settings because of the standardised frame limits for only one overall setting. sorry.
This topic is closed.
← All Discussions