VideoHelp Forum
+ Reply to Thread
Results 1 to 2 of 2
Thread
  1. Member
    Join Date
    Apr 2013
    Location
    Brussels
    Search PM
    Hello folks,

    The question is in the subject: given some camcorder, in my case Panasonic or GoPro, what is the conversion strategy of the built-in h264 encoder?

    I mean, given a video stream, I can use my computer with ffmpeg and x264 encoder to make it follow crf (constant rate) strategy, qp (constant quantization), cbr (constant bitrate) or other.
    Results are different.

    So, what a camcorder is doing?
    I tried to analyse some few videos from gopro and various graphs cannot provide any obvious answer. Frame sizes tend to be roughly close but not exactly equal, there is no also a direct correspondence between size and the scene complexity, or the motion activity.

    Why I'm asking? I need to recode the video, using, preferably x264 in order to squeeze its size (to be able to send it to my remote friends) but also to make it possible to extract from this SQUEEZED video photos (by my remote friend).
    However, the original video taken from the micro-sd card was already encoded upon being written. So, technically, the very original raw data from the CCD was compressed. Obviously, some scenes were compressed preserving better quality, some not. Depends on scenes and motion activity. Thus, we have a transform
    RAW->SDCARD
    Suppose, there is one frame, FRAME1, which lost in this process 5% of its perceptual quality.
    Suppose, there is another frame, FRAME2, which lost in this process only 1% of its perceptual quality.
    My wish is to understand which frames lost what and treat them oppositely in my further computer recoding. Hence, FRAME1 is going to be compressed(loosing) less, FRAME2 is going to be compressed(losing) more.

    At the end I contemplate to be able to extract individual photoshots which are of evenly equal perceptual quality loss compared to the very original raw data on CCD.
    It is also NOT AN OPTION to extract photos from what I have on my SDCARD without recoding as this task MUST BE DONE by my friends who are remote and whom I can technically send only squeezed video.

    As I learned, my task is OPPOSITE to the logic of fixed crf intent. crf determines active motion and compresses those frames more, as "human being does not concentrate on individual frames when there is active motion". I need OPPOSITE, I need to stop in the middle of some active scene and extract a photoshot. The higher quality the better. I therefore do not speak about perception upon viewing video, not.
    Fixed qp is also not great. It will try to compress all frames equally but I'm fine if some part of video has been compressed no too much in the camera, can be jammed more than other parts. SO, some waste of space is expected.

    Any thoughts?
    Quote Quote  
  2. Explorer Case's Avatar
    Join Date
    Feb 2004
    Location
    Middle Earth
    Search Comp PM
    Your workflow would be quite uncomfortable to me. Here’s why: you are jumping through hoops needlessly, trying to control things you shouldn’t have to concern yourself with. Is your video too large for your friend? Mega’s free 50 GB seems royal for most uses. Splitting the video file in segments allows for unlimited temporary cloud storage. If you need to squeeze your videos, how much video are you talking about? From what size to what size? E.g. 5% less seems pointless, but 50%+ less may ‘ruin’ your footage enough for frame grabbing.
    If your friend needs a very limited number of frames, then s/he could select time codes (“the GOP around 0:15:19.4”) from a low res version, and ask you only for those segments, instead of everything. Those segments could be extracted without recompression. Small transfers and everybody is happy.
    To any re-encode, your h.264 video directly from the camera is the “original”, so any further conversion process to preserve as much as possible, will be compared to that first video. There is no way to compare to what the camera originally saw, only to compare to what was saved to the memory card. And you can’t shift compression strategy on a frame-by-frame basis anyway.
    Quote Quote  



Similar Threads

Visit our sponsor! Try DVDFab and backup Blu-rays!