Tips for choosing right video compression & format for lectures

[Omk]

Hi,

I'm new to video compression and the specifics of video formas. Previously I have experience with mp3 encoding, but not with video, only watching. I have done some preliminary study already and installed some software on my Ubuntu system. Now I have HandBrake installed, with approppriate codecs, including x264 encoder.

The project: I have 16 days of course lectures recorded on HD video, taking 200GB disk space. The lectures were recorded by a few students using different camcorders. I would need to batch-convert them to as small filesize as possible, still being viewable on computer screen.

Being lectures, the video quality is not very impotant as it is mostly the audio that carries the information, but there are some diagrams and tables projected on a screen by the lecturer, and these must be readable. Audio was recorded with the built-in mic of the camera catching all the noises from the lecture hall, so the audio quality is not good.

As I am completely new to this, I dont really know which settings I should change and which resolution to choose to reduce the filesize. Today I tested some, converting to MPEG-4, but it reduced the size of my test file only from 44 MB to 42MB.

Could you give tips which settings would give the appropriate reduction in file size, and which file format I should use? How much reduction can I expect from the current 200GB?

[aedipuss]

what you need to do is pick your output resolution first. what are the files going to be viewed on? how big a display? 1920x1080, 1280x720, 640x480 are common sizes. then pick a bitrate. it's usually easiest to use crf, with bigger the crf the smaller the filesize and lower the quality. experiment until you find a number you can live with. i normally start with 18.

[smrpix]

What is the running time of this 44mb file?

[jagabo]

It doesn't matter what codec you use, or the nature of the video:

file size = bitrate * running time

If you want smaller files use a lower bitrate. It's like "miles per gallon". If a vehicle gets 30 miles per gallon it will go 30 miles on one gallon of gas, no matter what it looks like.

The codec and settings you use, and the nature of the source video (frame size, frame rate, noise, shaky camera work, action, etc.), will determine the quality you get a specified bitrate.

Many encoder use "constant quality" encoding where you select the quality you want and the encoder uses whatever bitrate it deems necessary to deliver that quality. With that type of encoding you don't know exactly how big the resulting file will be. But lower quality settings will give smaller file sizes. And, in general, slower settings (where the encoder looks harder for ways to compress the video) results in smaller files.

[Omk]

Thanks Aedipuss & jagabo, great. This was exactly the help I needed

smrpix: it is just a short clip, I tested with that one. It is 0:31 (31 seconds). Original file format is .MTS

Will try now to figure out what is the best and still acceptable quality.

[Omk]

Wow great,

I reduced the resolution to 1280x768, set the encoder to the slowest setting, RF to 29 and audio bitrate to 128. With these I got the 2.1GB file reduced to 210MB, with quality that serves the purpose well.

Thank you guys for introducing me to the video compression basics This will be very useful in the future for sure ))

[jagabo]

Be sure to keep your originals if you're going to use RF 29 -- that's pretty low quality. There's no reason to use the x264's slowest setting (placebo). The gains from going higher than "slow" are tiny.

[smrpix]

Wow great,

I reduced the resolution to 1280x768, ... I got the 2.1GB file reduced to 210MB, with quality that serves the purpose well.

Unless you have some really compelling reason to customize, I would stick with 1280x720 which is an actual video standard.

Glad you got it sorted.