Notes about learning Cinelerra.

1. Abstract

This file is about learning Cinelerra. I decided to learn Cinelerra because it seems a capable video editor for many operating systems. The context is the following: international travelling is at a minimum now, because people are afraid of the “Global Coronavirus Pandemic” and the conference have been cancelled.

2. Manuals

3. High level analysis

So we need to have a “video presentation”. What does it even mean?

Let’s say, it is going to be a “Video Review” on the subject described in a paper.

A video review should serve the following purposes:

  1. Tell the audience what the paper is about, so that they can judge whether to read it.
  2. Tell the audience who and what is the author, in order to discount his biases, potentially subconsciously embedded into the article.
  3. Include additional visuals that did not fit into the original document.
  4. Promote the author as a useful person.
  5. Add a “human touch” to the paper.
  6. Explain in human words things that are written rigorously into the text of the paper.
  7. Anything else?

What do I need then?

4. Task?

What can I start with? Record a video of myself “just talking” about SICP?

5. Exercises.

In the Cinelerra manual, there is no exercises. It’s a shame! Can I write them?

5.10. TODO

6. Terms

There are common terms used often. Some of them are listed in the manual, but I will repeat them any way for completeness.

Brightness - Light intensity. Sometimes marked with Y. Hue - colour as in “ratio of pure colours”. Saturation - distance from grey.

7. Reading the Cinelerra manual.

7.1. Preferences -> Appearance

What is BT601 and how is it different from BT709 and BT2020.

https://blog.maxofs2d.net/post/148346073513/bt601-vs-bt709

It is the set of coefficients used for transcoding RGB->YUV.

7.3. Cinelerra: Compositor window

The blue dot means that the video is paused.

7.4. What is yuv420p?

This is a format for encoding/compressing raw pixels. Encoding/compressing, because the data is not compressed mathematically, but some data is simply dropped. Presumably, this is enough to keep the image “visually identical”. The block is always two pixels high (I guess it is somehow connected to interlacing), but several pixels wide. The first number is the width. The second is the number of pixels used in the first row. It probably makes sense to make it a divisor or the first number. The second is the number of pixels in the bottom row. It probably should be a divisor of the first two numbers.

However, in principle, having a good algorithm may allow for different combinations.

7.5. What is FOURCC?

FourCC is a stupid extension to AVI container format, a number designed to indicate which encoder exactly was used to encode the image if/when it is mpeg4-family.

8. Notes

8.2. Webcameras: YUV vs MJPEG

You can check what a camera can stream with:

qv4l2
it is a GUI tool for querying cameras
ffmpeg -list_formats all /dev/video0
in the command line

8.4. Turns out, OBS-Studio may be better for my purposes

At least for recording a screencast.

Hm… can OBS actually record several tracks into a file, rather than an already combined track?