2

Live playback is almost fully flawless. Software like OBS can do transitions and greenscreen / chroma key effects live, without the need of rendering. So it would make sence if "rendering" would at least be as fast as realtime video playback is.

I do understand that videos produced with rendering are not compressed (are they?), but it's very hard to see the compression on the video, especially if video sharing sites like YouTube also compress the video. It would make sence in this case as well to have an option to export the video without rendering.

  • I did not want to put this in the question itself, but probably this question have been asked already, if not I would be suprised, but I couldn't find any question related to this with my keyword searches. – Joseph Feb 20 '18 at 20:22
  • It isn't really clear what you are asking. Do you want to know why some software has a slower than real-time rendering process while other software is able to manipulate live video with no (noticeable) delay? – Michael Liebman Feb 21 '18 at 3:48
3

When you export a movie from a video editing software, there are, broadly speaking, three steps involved.

  1. Rendering: reading each frame from source media and applying assigned effects to each individual clip. Compositing all tracks to form a final image. This part happens during live playback as well. However, most software will use faster and cruder methods during live playback, depending on hardware capabilities. So, the results may not be mathematically equivalent. But the average viewer may not notice the difference.

  2. Encoding: This is the compression part. One or more rendered frames are fed to the encoder, which then performs motion estimation and other methods to identify exploitable redundancies. Using that information, the encoder generates an encoded bitstream. This part, depending on the encoder chosen, takes significant CPU time.

  3. Writing: The encoded bitstream is then written to disk. If the frame sizes achieved in step 2 are small enough, then this constitutes a minor portion of the export process. But if not, disk I/O can add delay. This will usually be the case if step 2 is skipped and uncompressed frames are written to the file on disk.

During live playback, steps 2 and 3 are skipped. The rendered frames are stored in RAM and played from there. If the buffer overflows, earliest rendered frames are dropped, and re-rendered if requested. So, the difference in duration between live playback and render to disk is due to the better methods used in step 1 and the addition of steps 2 and 3.

  • If should be noted that there is software that is able to stream live video to CDNs (thus adding the encoding step) and also simultaneously record (almost) uncompressed video on disk (step 3) – mcont Feb 27 '18 at 19:04
  • @MatteoContrini I was just about to comment that, for example, OBS can do greenscreen effects when writing video to disk as well. – Joseph Mar 7 '18 at 14:50
  • I think the main difference is that live production software is specifically written and optimized for doing everything live, even at the cost of dropping frames or lagging a bit, if you're asking too much to the software. Video editing/rendering is instead optimized for generating an output of flawless quality no matter what resources are needed – mcont Mar 7 '18 at 15:00
0

My team live webcasts and records church services weekly using a hardware mixing solution (BM ATEM 1 M/E). The live stream goes out immediately, where it is buffered, re-rendered, re-encoded, and written to cloud storage by our live stream provider. You don't skip the steps 2 and 3 in @Mulvya 's answer, you merely delegate them to your live stream vendor.

The live stream is actually reduced quality because it is re-encoded from 1040p to 720p, and is subject to dropped frames and the vagaries of the internet.

After recording, a 40 minute sermon's raw footage is roughly 200 GB in size. We edit the recording to polish the presentation, hide production errors, remove dead space, and ensure copyright/license compliance, then render the edited presentation for archival and delivery on YouTube.

For special occasions, we will add several offline cameras to the mix, and up to 25% of the final rendered video will be from footage that was not part of the live webcast, often footage that was taken during dress rehearsals that would not be possible during the live presentation.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.