New render from extracted keyframes

[morganwars]

OK, after successfully using mplayer to extract the keyframes from a file...

I want to line them up and render them back into a new video, using the same frame rate and gop.

After much searching I'm throwing in towel and need help.

What tool can use to import these PNGs and render back in P B frames?

Thanks in advance!

[poisondeathray]

OK, after successfully using mplayer to extract the keyframes from a file...

I want to line them up and render them back into a new video, using the same frame rate and gop.

After much searching I'm throwing in towel and need help.

What tool can use to import these PNGs and render back in P B frames?

Thanks in advance!

I don't understand what you're trying to do :confused:

If you're making a "new video" , you still need the old P, B frames. They contain information that the keyframes don't have

[morganwars]

But P and B frames are derived for IFrames, why can't hey be re-derived?

[poisondeathray]

But P and B frames are derived for IFrames, why can't hey be re-derived?


Because they contain the predicted information, but store the differences between the original and predicted


Think of of it this way: If you have a panning shot represented by 24frames every second (24fps), and for this example your GOP length is 1 second. There is no way a single keyframe can show all the temporal data in a single frame. How would you be able to derive 23 other moments in time from a single keyframe?

[morganwars]

Sorry, let me restate: I have two files, they are recordings of same show. Both were record at the same bit rate.

I've pulled I Frames from one the files, and and I'm looking to merge those files into the second file, on the premise that more data could allow for some upscaling. Generally the same kind of function as remaking P files.

[poisondeathray]

Sorry, let me restate: I have two files, they are recordings of same show. Both were record at the same bit rate.

I've pulled I Frames from one the files, and and I'm looking to merge those files into the second file, on the premise that more data could allow for some upscaling. Generally the same kind of function as remaking P files.

You might be able to use some sort of averaging function, but the assumption that there is "more data" is questionable

If both recordings were of the same show, and used the same bitrate, and same compression method, I don't see how any of this will benefit you. Presumably they have both incurred the same (or very similar) losses