TimeMachine: backup couldn’t be completed … fix this

Written by  on June 30, 2015

Ok, TimeMachine runs from time to time, backups pile up and the disk is filled. So what now? Format it? No! Drop the oldest backup!


Initial inspiration for the solution was derived from this thread at stackexchange.

Qt: find the libs (Linux)

Written by  on June 26, 2015

I let synaptic install the libs, the qt-creator, the other stuff and their dependencies. Upon creating my first project I noticed NO kit was configured. Compiler was doable, but what about the libs? Tried this:

Check all existing tags for a certain tumblr

Written by  on June 25, 2015

Before you start any implementation: check your field of interest for existing solutions! This may sound awkward, but re-inventing the wheel is not what you should do with you spare time!
I was thinking about a simple and easy way to fetch all existing tags for a certain tumblr. Since they are not presented directly in an obvious way testing the commonly used things like “me”, “myself” or “own” helps, but takes time.

Looks like Benjamnin Horn did something like this. Together with some explanation.
The working example for doing your “researches” can be found here.

Good work!

CodeCombat: learning Python the cute way

Written by  on June 23, 2015

I am currently trying to enhance my skills by learning Python. I made a choice between Ruby and Python after I was quite frustrated sometimes while writing the contactSheet-script. Bash is missing some general debug-abilities (or I don’t know them ;)) and the way to control the code-/data-flow was quite limited from my POV. I came from high level-strong typed-precomiled C++ and then hit bash ..

CodeCombat is a browser-only, completely OpenSource “game” where you lead your protagonist through its adventures by using code-snippets. One after another simple and more complicated statements are introduced, control structures like loops and alternatives follow. You can chose between different languages – besides Python also JavaScript is available.
It is quite amusing to guide the hero(ine). I finished already 22 levels.


For in detailed explainations I stick to: “Learn Python the hard way

Create an usable linux-usb-drive with OSX

Written by  on June 22, 2015

Based on this tutorial (for Linux itself, use Unetbootin).
Open a terminal and then run:

Used it twice to create first a Kubuntu-stick (did not like the desktop-enviroment) and then a newer for Linux Mint (latest version).

Pos, Neg or Koda? What SF8-HDR-type does the file have?

Written by  on June 17, 2015

I needed a quick and dirty solution to automatically check for the type of the given HDR-scan. Since some custom makernotes are used, the question was how to get by using exiftool the necessary information. A quick comparison with hexdump and diff showed the difference in the binary output.

So this tiny script will return the type of the given input as exitcode (checkable in terminal via “echo $?”): 1 for positive, 2 for negative and 3 for kodachrome. 0 for errors … and yes, this negates the POSIX-semantic! I guess someone more capable could turn it into a one-liner. Nevermind: it works and this is sufficient for my usage.


OSX: How to get imageMagick (working with fonts)?

Written by  on June 15, 2015

Hmm, OSX offers out of the box a lot of UI-programs, but is missing most of the *NIX-fun. So if you need imageMagick for your work, you are missing it. The prepared installers are missing the support for the font-manipulation.
Of course, you can download and build all packages by yourself from scratch … or you use one of the nice packet-managers like fink, macports or homebrew.

The fastest way to get a usable OSX (tried with 10.10 and 10.8) for the contactSheet-script is by executing those simple four commands in your terminal. Existing internet-connection assumed:

(Linux: use your favorite package-manager.
Windows: :DDDDD I guess it won’t work until you also get some bash-replacement.)

Check afterwards the current configuration. The list of delegates should show “freetype” or “fontconfig”. If both are missing, then rendering the filename onto the thumbnails won’t work!


createContactSheet 0.4

Written by  on June 14, 2015

Okee, sorry for the continued spam about this small bash-project, but I am a little bit excited. It turned out much more powerful than expected. I learned a lot new and really useful stuff. I am also really confident now about the future of “bash & me” 😉

I improved the latest state with an idea a good friend pointed out: why don’t you add the original file name to each thumbnail? For a better overview? This is done also by imageMagick (convert); also append now the base-directory as footer-note. Because for my workflow the dir-name is the scan-date and the emulsion-type.
Some other improvements are: changed workflow (negative-conversion on each picture instead of the whole combined one; therefore no “filling of the last row” needed); script-parameters for help and version and for suppressing the negfix8-conversion; cleaner code.

I have also some more ideas in mind, but I think for now it is enough. It was fun, but I also want to discover some other areas of IT!

Download this: createContactSheet_0.4


createContactSheet 0.3

Written by  on June 13, 2015

I fixed a small glitch, which was included in the first releases: input-filenames with spaces were not treated correctly therefore no result was created. Fixed by some string-replacement and setting the IFS-variable which dictates that bash should treat spaces not as separators.
For the next release some filename-embedding into the thumbnail is planned. But this has to happen after the negfix8-conversion else it would alter the conversion itself ..

Download this: version_0.3

Small hint: git offers a really nice exporting-function which will just include the current state of the given branch:

You want a contact sheet with that?

Written by  on June 3, 2015

Slogan is like always: “If you are not ashamed of the first release of XYZ, then you waited too long or have no self-reflection.”

It would be nice to have a preview shaped like the original contact sheets of a HDR-film scan. So you could screenshot the thumbnails and print this.

Or you try the following script:

  • it takes all TIF from one directory (assuming they are HDR-scans from SilverFast)
  • creates thumbnails with a longer side of 400px into a separate directory
  • appends six thumbnails for each row and the combines all rows to a single picture. If the number of pictures does not fit, the last picture is repeatedly appended (this is necessary due to the negative conversion)
  • ‘negfixes’ them (negative conversion done by the great original script of negfix8 from JaZ99!)
  • spreads the contrast in two diffent approaches (check yourself which is your most liked version)

As a result you get something like this:


As you can see: the black strips at the left and right side of the pictures still exist. But since I scan always the HDR-files with top and bottom cropped, the black borders don’t show up there. Time to change my workflow?



  • download & extract the two scripts
  • open a bash-terminal an navigate INTO the directory with the HDR-scans
  • run “/PATHTOSCRIPT/createContactSheet.sh” and wait for 2 to 3 minutes

Future plans:

  • improve the structure of the code and add some more comments and make it less error-prone
  • full clean-up after the creation (by default; suppresable by param)
  • suppress the negfixing by param so also E6 (slide)-scans can be ‘contact sheeted’
  • make the script use the general temp-folder and put just the final two contact sheets into the target folder (so the data stays clean)
  • add the filename in the bottom-right as transparent overlay so you know which thumbnail belongs to which picture


edit 20150603: Just uploaded version 0.2 which improves performance a lot and cleans everything up and has a nicer structure!