Merge MP4 with PowerPoint slides without re-encoding?

[yetanotherlogin]

Hello,

Considering it takes hours on my slow workstation to render a 1-2 hour video simply to overlay a few slides in a second video track… I was wondering: Provided the slides have the same width+height as the video, wouldn't it be possible to insert the slides at keyframes, and simply merge the two without needlessly re-encoding the whole thing?

Thank you.

[Cornucopia]

Yes, if you encode the ppt to the exact specs of the video (don't forget to include audio -even blank- if there is audio with the video). But you may need to use a different app than vsdc.

Scott

[yetanotherlogin]

Has someone done this? Any HOWTO I could use?

[abolibibelot]

With Avidemux or VirtualDub2 you can view the video and jump from key-frame to key-frame, so as to visually select the best possible location for the slides insertion.
With ffmpeg you can convert a still picture to a video of pretty much any existing format, including MP4, and of any duration. I don't remember the specifics, someone may provide you with a sample command to do that.
So the tricky part to keep everything in sync would be to create a short clip from each slide with a duration corresponding exactly to the number of frames that you delete from the original video, between two key-frames. Otherwise, adding those clips with blank audio would probably be awkward.

Or perhaps use an editor that is able of “smart rendering” ? Meaning : only re-encode the parts that have changed, and copy the sections which haven't in “pass through” mode. The Magix editor which I use has that ability, but apparently only for AVCHD format. If yours has the same behaviour you could, I suppose, losslessly convert the MP4 to M2TS (AVCHD) with TSMuxer, then edit to add those slides, then “smart” render (which should take a fraction of the time needed for a full rendering), then remux to MP4 as required.

[smrpix]

Don't mean to sound too cynical or skeptical -- but I'll bet by the time you're finished measuring, calculating, debugging and encoding, you could have just done it in your NLE and let it render overnight. Keep an eye on your time.

[abolibibelot]

Sometimes losing some time in the short term can end up saving a lot more of it in the long term.
I couldn't say if that's the case with the issue being exposed here, but generally speaking I think that it's a good thing, especially on such a help forum, to admit that someone has a good reason for wanting to do what they ask how to do (unless there is a strong indication that they don't know what they're doing whatsoever, or have such severe misconceptions that the question doesn't make any sense whatsoever, or if they are adamant that they belong to the feline species), if only to try and realize later on that it wasn't worth it after all (it can always be worthwhile as a learning experience, which could prove useful at some point in unexpected ways), rather than questioning right away the rationale of why wanting to perform the task in question in the first place. Also, someone could have different priorities than what you would consider reasonable or efficient, and it's alright ’cuz they are (for a few decades at least) human beings who have the right to make errors and run fools' errands if they wish to do so with their time. Some great discoveries have been made by someone who “obsessed over low-priority things”, and efficiency shouldn't be the only criterion of decision for our actions in this damn world. (If anything, the quest for efficiency has done a lot of harm in history, and continues to do so at an ever increasing rate.)
Sorry for the rant, I'm a bit upset these days, for good reason shall I say – well, those are my reasons anyway, even though someone very reasonable might say that they're meaningless and that I should just “move on” right away.

[Cornucopia]

My long history here, and troubleshooting & helpdesk best practices, tells me its almost ALWAYS best to start with the why, as that further clarifies the target, and also helps to rule out things.

But this should be quite simple.
Use ffprobe, mediainfo, etc to find out what the current encoded video is in detail.
Encode the PPT with those settings. If that isn't easy direct from ppt, use a lossless intermediate and then re-encode to the specs. Call it B.
Using avidemux, etc., cut on the gop boundary where you want the insert point. Save the earlier clip and the later clip. Call the A and C.
Append A + B + C. Save as D.
Done.

Scott

[smrpix]

My long history here, and troubleshooting & helpdesk best practices, tells me its almost ALWAYS best to start with the why, as that further clarifies the target, and also helps to rule out things.

But this should be quite simple.
Use ffprobe, mediainfo, etc to find out what the current encoded video is in detail.
Encode the PPT with those settings. If that isn't easy direct from ppt, use a lossless intermediate and then re-encode to the specs. Call it B.
Using avidemux, etc., cut on the gop boundary where you want the insert point. Save the earlier clip and the later clip. Call the A and C.
Append A + B + C. Save as D.
Done.

Scott

Scott, If we're talking about a single slide, or a handful, I absolutely agree that's faster. When I cut PPT slides into a show I generally deal with dozens or more. That's where my caution is coming from.

[smrpix]

Some great discoveries have been made by someone who “obsessed over low-priority things”, and efficiency shouldn't be the only criterion of decision for our actions in this damn world.

What great discovery do you think would be made by reproducing Panasonic's metadata?

Some of us try to discover ways to do things efficiently, which is also creative and rewarding.

[johnmeyer]

Not entirely sure what you're trying to do, but if your goal is to insert Powerpoint slides into an existing video (insert, not overlay), then you can easily do this in VideoRedo. What's more, you don't need to waste time worrying about whether you are inserting them at the beginning of a GOP because VideoRedo will re-encode just that one GOP. So, you get a lossless copy of all of your video, except a few frames around where you inserted the slides.

VideoRedo

[Cornucopia]

@johnmeyer, assuming you set the encode settings to match incoming (and the codec is one of the supported group of smart-render capable), any smart-rendering nle should be able to do that, but good call. For lots of inserts, that's how I would do it.

Scott