Shaving avi files!

[jlrm365]

I made a little mistake with some pre videohelp editing.

The result is I now have one avi with a few seconds of the beginning of the second avi on the end of the first, with it also obviously being at the beginning of the second.

This poses a problem, because I no longer want them seperate and would instead prefer them joined. How can I shave fractions of seconds off the end of an avi file?

Which software should I use and how should it be used?

Also, for future reference, is there a piece of software which can cut out bad little bits in the middle of an avi and then pull the two halves on either side together? I realise this would mean a skip-like effect but to leave the section in would be worse!

I have seen lots of software suggested but I would prefer a simple filter for VirtualDub if one is known or a very simple other piece of software, as I prefer to streamline.

[Steen4]

The answer is simple, depending on which codec you used for your AVI file. Am I correct in assuming (the wording of your explanation was a little difficult to understand) that video file A has a little of video file B at the end, and video file B has a little bit of video file A at the beginning? If this is the case, load the two video files into VirtualDub by loading file A, then APPENDING file B. Move the tab along the timeline with the mouse until you come to the first frame you intend to lose, then hit the HOME key. Continue moving the tab along the timeline until you come to the last frame you intend to lose. Hit the right-arrow key to advance one frame more, then hit the END key. Notice that there is now a colored segment in the middle of the timeline. That will all be gone from your video when you re-save it (under a different filename, of course). Hit the DELETE key, and then move the tab left on the timeline a couple dozen frames or so. Hit the spacebar to review the results of your editing work. If you think you could have done better, just open the file menu, click "close video file" and try again. If you are satisfied with your edit, then go to the audio menu and click "direct stream copy." Afterwards, go to the video menu and click on "direct stream copy." Go to the file menu, click "save as AVI" or "save segmented AVI" (there are file size limitations in certain situations. You can find more on that elsewhere in the forums), name your file, and save it. Post back with your results.

[jlrm365]

You have understood me exactly, so I cannot have been unclear.

Your advice seems like just what I was looking for.

I suppose that would be the first thing to do, before any other edits or filter applications, yes?

I'll indeed get back to you with results.

Thanks.

[jlrm365]

I found it easiest to spot something going past a second time in the overlap, then set marks at just before the first time and just before the second, cutting that piece in the middle. I also tried cutting a little less and then pressing the delete key to take the last few frames. That felt better, as it seemed as though I had more control.

[Steen4]

Good to hear that you got the results you wanted. Just remember that saving as direct streams will edit the frames without reprocessing, so the editing should be lossless. Any other editing mode will degrade the quality of the file.

[jlrm365]

Can you explain that a little more? I still get a small noise when it makes the bump from one to the other, but not much.

Can you describe the results as you have suggested and the alternative?

Thanks.

[Steen4]

OK, I'll relay my own, admittedly limited, understanding of it. Compression for binary files (text, programs, etc.) must be lossless. This means that the precompression file will always be identical to the compressed and re-expanded file. Otherwise, binaries can behave unpredictably...if they work at all. Video and audio files, though, may use differing degrees of lossy compression without hazard. This is because, while programs must be kept intact at the bit level, audio and video files have information that could be interpreted by an encoder as insignificant to the final image, or below the limits of perception (or at least not terribly missed if it were to be chopped). The result is image or sound quality similar to the source, but with a much smaller file size. This kind of 'culling' is called perceptual coding, and is the basis for the kinds of huge compression ratios you may obtain with some compression schemes. The downside of perceptual coding is that this information is gone forever. Each time the file is compressed and re-expanded, more information is tossed, so re-encoding should be done only if no other option is open. If one were to encode a first generation AVI file and encode it as a DivX video, there may not be a noticeable difference in quality, depending on the bit rate and amount of motion. Re-expand that DivX file back to uncompressed RGB, recompress as a DivX file, re-expand again, then recompress, and I'm sure you'll wonder what happened. Generational loss is what happened, just the same as copying VHS tapes from generation to generation. The best way around this degradation is simply to make sure your sources are just the way you want them before the first encode, so you won't have to perform after-compression surgery on them. Failing that, though, you should try to edit without the recompression step. VirtualDub has a direct streams option for its audio and video, but it can't do everything. You can only perform time line editing, such as pasting, cutting, appending, and deleting frames. Once you try to process the video image in any way (say, using filters or changing codecs) direct streams cannot be used, so try to stick with direct streams for already-compressed files unless there is a significant need for reprocessing. By the way, you might not be able to get rid of the bump without demuxing (separating the audio and video streams), expanding, editing, then reprocessing the audio (the video would not need processing, so will be unaffected)...unless there is some kind soul out there who knows of another option.

[jlrm365]

That's a little more indepth than I expected, but thanks.