Any one know how digital video works?

[scv76]

I have been looking into cameras that could be used to video record golf swings, and have a few questions that none of my associates can answer.

I have examined some golf swing video that was supposedly recorded at 120 frames per second. When I play the video back frame at a time, I don't seem to see all the frames that I think should be there. (120 frames should go by in one seconds time) On the down swing, the club head moves something like 45 degrees down the swing arc.

The video has an example at 100% speed and 20% speed. The poster says he used an editing program to slow the video to the 20% of full speed. The 20% speed video has about 5 times more frames viewable on the downswing portion than the 100% sample.

I'm having a hard time understanding why the number of frames differs between the two examples. I would think the number of frames would be the same, but the 20% sample would take 5 times longer to play.

Are the video players limited to about 30 fps?

Is the codec algorithm limiting how many frames I can really see?

If I film a stopwatch at 120 fps shouldn't I be able to go forward a frame at a time and see 120 frames for each second that passed on the watch?

Here's the  link to the swing video I have been looking at:

[newtogolf]

There are a lot of variables when discussing digital video and frame rates.  One key variable is whether the video is interlaced or progressive.  In interlaced video it actually takes two frames to make one, as one frame contains the even lines the next contains the odd.  So an interlaced video recorded at 120 fps has a net fps of 60.

The other variables relate to the compression method and player used to review the video.  Higher frame rate videos take up more storage and are not supported by all online viewers.  I believe YouTube has a max frame rate of 30fps.  Any videos you upload that are higher than 30fps are compressed down to 30.  I hope I answered your question.

[scv76]

Thanks for the response.....it's funny, I was just going to edit the post and ask if by any chance you tube was limiting the videos.

The  "interlaced video recorded at 120 fps has a net fps of 60" information is something I had not thought of.

The camera in the video is a HD Hero2 and here are it's resolutions.  The 120 fps is for WVGA, so I suppose that is interlaced. I'll have to find someone who has some raw un-edited footage so I can see if it's really only at 60fps due to interlacing

• HD RESOLUTIONS:
  • 1080p: 1920×1080, 30FPS
  • 960p: 1280×960, 48FPS + 30FPS
  • 720p: 1280×720, 60FPS + 30FPS
• STANDARD DEFINITION RESOLUTIONS
  • WVGA: 848×480, 120FPS + 60FPS

[Gifted Golfer]

Thank for the information.  I am looking for one too.

[Clambake]

I think the thing that you're observing here isn't related to progressive or interlaced video or the actual frame capture rates, but is actually a side effect of the digital video compression works.     Without compression, the camera will take a snapshot at each frame, digitize, and store it as is.   This takes up a LOT of storage space in the camera, so the manufacturer used video compression to reduce the storage so you can store more than a few seconds of video on a memory card.

In video compression, a key fundamental is to take a full picture every few frames, and then mathematically fill in the frames in between these based on some very complicated algorithms.       The full picture is called an I-Frame, or "Intra-coded Frame".     The frames in between are called P and B frames, for "predictive" and "bi-predictive" frames.    These include some small amount of knowledge about the image in that frame, just enough to let the video decompression engine figure out what the frame should look like.     But after awhile these predictive frames become a lot less accurate, so another I-frame image is taken and stored, and then the subsequent predictions are based off that new image.    The algorithms used also take advantage of the way the eye perceives motion to further improve the efficiency of the compression.    For example, in fast motion the video compression algorithm designers have found that detail is not as important since the eye is occupied more with the motion itself, so some of the detail information can be thrown away without impacting how it looks on full speed playback.    This last bit is one reason why when you pause your DVR the image may look like it has a lot of digital artifacts, blurriness, etc. when it doesn't seem that way when watching the image full speed.

Back to how video compression reduces storage, one simple example is imagine a fixed video shot of a landscape with a person walking across.    The landscape itself doesn't change from frame to frame because the camera isn't moving, so the video compression engine just says "stay the same" to the majority of the picture.   The only information actually needed for these shots is information about the person walking.   But it gets even better - you wouldn't have to send that information about the person walking for every subsequent frame either, as you can predict where the person would be in the next image since they're walking one direction.     In this example, the very first I-frame would have everything, and then the next frames could be P-frames that are just a little bit of information about what the person looked like and what direction they're heading.     This all lets the device store a lot more information in the same amount of storage space, and when played back, this information would be "decompressed" by doing all the algorithmic calculations from frame to frame.

The video on your cameras is similarly compressed, but when you do a frame by frame playback you're really playing back an I-frame, then another I-frame, and another, and so on.    It may also try and put in few of the B-frames as well, but typically in video compression engines these are skipped in "trick mode" playback such as slow motions, fast forward, and frame-by-frame.      So even if you're camera says that it is doing 240fps, it probably isn't storing it that way and then when you play back you're getting what appears to be gaps in the images.    The way around this is to go into the camera and change the settings (if available) to record uncompressed or raw images.   This basically records every shutter shot as an I-frame without any of the predictive compression.      You'll probably only be able to record seconds of video instead of many minutes, but it should give you back the fidelity you're looking for.

Sorry for the long response, but you DID ask "how does digital video work"!

[scv76]

Thank you for the response..it looks very informative; now I need to digest what it all means

[Clambake]

The simple answer is if you want smooth frame-by-frame playback, record the video in full uncompressed format.       Hope that helps better!