This is an old revision of the document!
Table of Contents
Here some information about what I use to make the videos for the podcast.
general information
release
This podcast is not the main project. The projects I talk about are. And the video is just the final part. Projects need time, particularly because it is often about something new I have to learn first. Then comes the video. My goal is to release an episode every month.
resources
The opening music is from Voodoo by Opusvertigo.
The font for the opening screen is Voodoo Spirits by Levi Halmos.
recording
camcorder
To record my videos I use a Panasonic HC-V707EG-K. It was one of the first to offer 50 fps at 1080p. It's not bad, but lacks a couple of things:
- no headphone connector
- no manual focus ring
- 60 frame/s only for the US model
- no remote control or connection for a remote
- the external microphone recording is sometime very noisy
For the audio I have Audio-Technica ATR3350 omnidirectional condenser lavalier microphone. But when plugged in the camera it generates a lot of noise in the background. I could not figure the cause, but it prevents me to get better audio recording. And the camera built-in microphones aren't too bad.
screencast
To record the screen I use SimpleScreenRecorder. It's simple and does the job well.
The audio is recorded using a Samson C01U USB microphone.
editing
video
Currently I am using blender to make my videos. Most know it for its 3D modelling, but it can also do video editing. It is slick (I avoid software larger than 100 MB), stable, and does the job quite well.
I tried several non-linear video-editors:
- I started with Kdenlive (v0.9.10). But it crashes too often, it has to be restarted because some actions can't be performed in the timeline for unknown reasons. What made me change was the speed video effect. It does not stack with the rotate effect, and it does not work correctly (it uses video before the beginning of the video part, and does no go until the end)
- Pitivi freezes when I open a folder with H265 videos, and does not provide speed control.
- Lightworks (v12.0) does not work on Arch Linux. It shows a empty screen.
- Shotcut does not provide speed control.
- Cinelerra does not support a lot of input media. But I would like to test it further.
- Blender does not provide Motion-Compensated Frame Interpolation. This means that the input video frame rate has to match the project frame rate, else it gets unsynchronized.
To encode the video before and after editing I use ffmpeg. It does a good job at transcoding videos. And for the final video packaging I use mkvtoolnix.
audio
Audacity does the job well and provides a nice noise removal function. But it does not provide a command line interface and can not be scripted.
SoX is small and fully scriptable. Its noise reduction is not very good, but the rest works well.
subtitles
To make subtitles I use Aegisub. On the first pass I transcribe the video with approximate timing, only using the keyboard. I correct the timing with the mouse on the second pass.
Until episode #007 I transcribed the complete videos in the subtitles. Starting episode #008 I use the subtitles to add comments, details, and correct myself.
work flow
To help me making videos, I have a bunch of scripts which do all sorts of side tasks:
- compress the raw recordings
- clean the audio
- render the video
- extract the chapter marks
- package the final video (with chapters, tags, cover, and comment subtitles)
- create a torrent
- upload to my server
- upload to youtube
But I correct and extend them for each new episode.
I just have to make the recordings, put them together, comment the video in subtitles, run the scripts, and finally click on the publish button.