Development log

Code, tricks, quick technical ideas, recipes, progress reports, bash, ffmpeg and more…

ffmpeg: audio visualization tricks

Often one wants to share audio online, but it seems like video as a format has many more options: mastodon, twitter, facebook, youtube, all allow to upload video, but not only audio. Here are some ffmpeg tricks, how to add interesting video to your audio file, often autogenerated visuals. All the code is supposed to be used as one line without line-breaks.

 

Audio Vector Scope

The code above creates a mp4 video file with a vectorscope nicely centered inside a 854×480 (480p) video. If you need a 1:1 video, just exclude the pad part:

Documentation on ‘avectorscope’ filter is here: https://ffmpeg.org/ffmpeg-filters.html#avectorscope. One can play with zoom and other options to produce desired form.

 

Show waves

more options: http://www.ffmpeg.org/ffmpeg-filters.html#showwaves

Showspectrum

Above code will create almost completely desaturated spectrum of the audio sliding from right to left. Again, there are various options to tweak, see here: https://ffmpeg.org/ffmpeg-filters.html#showspectrum-1

 

Histogram

Documentation: https://ffmpeg.org/ffmpeg-filters.html#ahistogram

 

Static spectrogram

Sometimes you want to just create a static image.

Above one is in two steps. More info here: http://www.ffmpeg.org/ffmpeg-filters.html#showspectrumpic

 

Drawtext

Add text to any of the above with a “drawtext” filter. More options here: http://www.ffmpeg.org/ffmpeg-filters.html#drawtext-1

MikroTik RouterOS configuration for simple switch

My home network consist of:

  • little linux box acting as Media Server (to play videos off it and projector connected to it) and main routing and firewalling machine. it gets/sets an IP with our ISP and runs a DHCP server and DNS and IP masquerading and NAT-ing for the whole home LAN. The box itself has only one network port but I extended it with another USB network adapter – so, this box is connected to ADSL modem on one interface, and to another router (acting as a switch) on the other interface:
  • two SOHO consumer-level routers configured as simple switches/wireless bridges/AP with disabled DHCP and any routing – they just forward any connecting devices to DHCP on the main linux box.

I ordered a very cheap but seemingly powerful router by MikroTik “hAP lite RB941-2ND-TC” and configured it in the same way as other two routers – one of them of not both will need to be replaced by this new one. So, the configuration is as follows in the web interface – Quick Set:

  • set wireless Network Name if you want the way you want
  • under Internet set:
    • Address Acquisition to “Static”
    • IP address to 192.168.88.1
    • Netmask (leave at) 255.0.0.0 (/8)
    • Gateway to 192.168.88.1
    • untick Firewall Router
  • under Local Network:
    • IP Address: 192.168.88.1
    • Netmask: 255.255.255.0 (/24)
    • untick DHCP Server
    • untick NAT
    • leave UPnP unticked
  • VPN address should stay unticked
  • setup your new admin password, if you haven’t already, and Apply Configuration/Changes.

Now connect the (real) router – linux box to this MikroTik on non-internet port, either ports 1,2,3. You should now be able to connect via MikroTik either with a utp cable (again, not via port marked as internet) or wirelessly.

To go back to configuring your MikroTik, you need to connect to it with a cable and manually set your IP to something like 192.168.88.246 (disable your wifi etc..). Then you should be able to ping MikroTik or open the web interface at 192.168.88.1. Otherwise the MikroTik is somewhat invisible (apart from providing the AP – wifi access point name).

Tearing of fast action on NVIDIA

So with a brand new setup that includes a GLX/Geforce 1070 graphic card (will I ever be able to abuse all that GPU power?) I was somehow dissapointed to find non-uniform flickering, in other words the fast movement of things on screen was tearing. I didn’t prioritize this problem yet, but today I took time to look into it, and the following recipe solved it:

To avoid tearing, there’s a “workaround” – there are two options that seemingly need to be enabled in Xorg/nvidia driver: TripleBuffering and ForceCompositionPipeline (or ForceFullCompositionPipeline). My /etc/X11/xorg.com now contains the following in the “Screen” section:

To find out what is your screens layout, you can do

Also, if you want to fix tearing temporarily without changing any configuration files, use this command with something like that:

Supposedly forcing composition pipeline is causing some degradation in quality/speed in games.

Sources:
https://devtalk.nvidia.com/default/topic/543305/linux/screen-video-tearing-gtx6xx-7xx-kepler-9xx-maxwell-in-almost-all-applications-including-desktop/1
http://http.download.nvidia.com/XFree86/Linux-x86/325.15/README/configtwinview.html
https://forums.freebsd.org/threads/52311/#post-312833
https://forums.freebsd.org/threads/52311/#post-325927
http://us.download.nvidia.com/XFree86/Linux-x86_64/319.32/README/xconfigoptions.html

 

Learning SuperCollider patterns: legato, sustain and using pairs

Yesterday I had problems understanding why does a SynthDef and it’s Synth instance complain when it’s envelope duration is shorter than duration \dur. The following code will produce FAILURE IN SERVER /n_set Node 1067 not found complaints in the post window.

The sound will be generated fine, but the problem is that the doneAction and sustain value in Env.linen result in the Synth instance being freed before Pbind does it. So when Pbind does tries to free it, it’s already gone, it doesn’t exist anymore.

I asked nice SC people on SC FB group and the answer led me to understand that there’s a good use of\sustain parameter in Pbind, which deals with exactly that. So to properly control sustain time, an example shows how it can be done:

It seems like the most important part is that the SynthDef has a sustain argument. Pbind will calculate sustain from \dur so that it doesn’t even have to be defined in the Pbind.

There’s a nice thing used above, which I was looking for a while – how to defined note and it’s duration in pairs instead of on separate lines for each parameter: [\degree, \dur] and what follows is a list of pairs.

linux: auto-login

There’s couple of ways how to setup auto-login on linux. This one is geeky and not very user-friendly, but might come handy to somebody. This is on Ubuntu Linux 14.04.1/trusty with graphical login/desktop manager (lightdm) disabled.

install mingetty

edit /etc/init/tty1.conf so that it looks like this:

The most important part are the last three lines. Change username to what you have.

Edit your ~/.profile and add at the bottom:

Reboot and you should end up in your X session.

Sources:
askubuntu.com/quest…unning-upstart
https://wiki.archlinux.org/index.php/Xinit

supercollider: easy beats with patterns

minor random bliss in two

Learning Patterns in SuperCollider, here’s a little lullaby:

 

SuperCollider Help in dark grey

In order to browse and use SuperCollider help system in inverted gray colours, there are two quick hacks:

  • modify internal Help system
  • style the help on doc.sccode.org using Stylish

To modify internal Help system find the file /usr/local/share/SuperCollider/HelpSource/scdoc.css and add at the bottom/end of the file:

If you use doc.sccode.org and a Stylish add-on in (for example) Firefox to custom-style pages, you can add a new style which is similar to the above:

Processing and Awesome WM

Processing is a Java application and has had troubles in my Awesome setup for a while now in various ways. With some early versions (2.x) the main IDE window didn’t want to redraw following a resize. Now with new IDE in 3.x series this problem is gone, but now the main output window thinks it has a window decoration and is offset in a very ugly way, showing a wide grey bar at the bottom and slighly less wide one on the right (lack of top window decoration and a scrollbar probably.

These problems can be solved with a program called wmname before starting a java application.

(source: https://awesomewm.org/wiki/Problems_with_Java)
You can find it in the ‘suckless-tools’ package on Ubuntu/Debian.

It seems this needs to be run only once in one of the terminals and it then works accross any subsequent commands in the current session. Probably not stupid to put it in ~/.config/awesome/rc.lua?

making of Requiem for The Future: A.I. video

The biggest part of creating this video was to generate satisfactory glitched material. I researched various ways how to glitch images using SoX tool and a glitch in pre-3.0 ffmpeg’s jpeg-le codec. Below is a BASH script which worked through all .mov videos in the current folder, extracted frames, glitched it and stitched back the videos from frames. Here’s the gist of all glitches in a summary:

 

Here’s a full bash script:

Continue reading

jpeg-LS glitching with FFmpeg

In an older version of FFmpeg it was possible to glitch the image with JPEG-LS codec. Newer versions of FFmpeg don’t work anymore in this way, so one must download an old version and compile it (keep it local) – here 2.0.7 is used.

This script takes a video file as an argument, extracts frames, glitches them, and gathers frames back into a video file.

 

 

glitching images & movies with audio effect using sox

Some tutorial will be here soon, but for now, this script that glitches your video file.

Requires bash, ffmpeg & sox.