January 27th, 2009 by Rolf
This week it’s geek time. I found in the German c’t magazine an image which can measure the amount of change to contrast caused by an image manipulation. Yes, an image can measure. I contacted the author, Ralph Altmann, and we cooperated in translating this file from Photoshop to GIMP. (Cooperation = he did most of the work. 
)
The sound of this episode is a bit off. I have a new headset for recording – good quality – but I have messed up the mixer settings. Some experiments to do.
Sorry, no TOC yet.
There was a problem with the upload. If you have downloaded before 11:00 GMT, please check your file and reload if necessary.
This work is licensed under a
Creative Commons Attribution-Share Alike 2.0 Germany License.
Loading ...
Truly fascinating!
As you say, it deserves more thought and attention.
~jim
Very interesting. I read the ct special magazine, too but I did not find the time to try this out. Great to see this action in Gimp without adjustment layers. Thank you!
Just two points:
1) You are using layers in Dodge mode and you are presenting us the documentation of the Burn mode. Why?
2) “0 inverted is 0″ Is this really true? I thought that 0 inverted is 255 in 8 bit RGB?
I have another question. What exactly are the different colored lines mean regarding on a real photo we are about to sharpen? What is the difference between the red “line” and the green and yellow line?
I think the aim of this is to be aware of that the sharpening algorithms are clips some tones. And to encourage to use some kind of sharpening which does not clip so much. Right?
I tried a bit around with the gimp file you linked in the show notes and got a result of sharpening the test image without clipping. Using a normal USM sharpening with a edge detected layer mask, inverted and applied a certain curves setting on it. I’ll post this in the forum.
Ported an example for AviSynth
http://forum.doom9.org/showthread.php?t=144638
Don’t know if Gimp has a levels layer, if it does then you can combine the dodge’s & invert into one levels layer.
I have just finished watching the video for the second time and I really enjoy Rolf when he is in what I think of as his Lecture Mode. No, I am not a geek but have a leaning towards science and maths and really enjoy learning how things may be shown to take place. The fact that colours and virtually everything else has a mathematical base in digital operations is exemplified once again.
On a different matter, Rolf, could you please tell me why you and some others insist in pronouncing Gaussian in some peculiar way and not in the pure German which herr Gauss deserves?
@Paul: Dodge and Burn – well, lack of concentration I suppose. I’ll think through theses operations again – seems that I have not understood them really.
The three colours reperesent the effect in relation to the distance of the contrast edge. How that relates to an actual image is a thing to be explored.
@Norman: I have never heard an other pronunciation in English, but I will try it next time. I live not far from his first triangulation point away and we have one of the first dozen or so in Bremen, because the Senat paid a share.
Perhaps I should look into the different blurs sometime.
Some thoughts about this video in the forum: http://forum.meetthegimp.org/index.php/topic,260.msg2056.html#new
You will find e.g. my ideas about the questions raised by Rolf and Paul.
In addition to Rolf’s explanation in the last comment: The wider those lines are separated, the sharper this filter works. You can compare it with the brush. Painting with a hard edge brush will result in clearly visible borders that the effect is working in. painting with a soft edge brush will result in blurred and not defined borders for the effect.
Maybe you can say: The further those lines are separated the more artificial an image will look. (???)
littletank: You would have to say “Gaußsche …” instead of Gaussian. Nobody would understand you if you mix a german pronounced name with an english suffix, I think. And this would be even worse…
I never heard a different pronounciation in English neither.
But the English suffix ‘ian’ is pronounced ‘ee-an’ as in Christian and not ‘shun’, ‘-tion’ is pronounced ‘shun’ as in station. In physics and maths gaussian is pronounced as gowssian.
According to Merriam-Webster littletank is right, look (+ listen) here: http://www.merriam-webster.com/dictionary/gauss and there: http://www.merriam-webster.com/dictionary/Gaussian
Rolf, excellent show (as usual)!
Speaking for myself, i have been pronouncing gaussean incorrectly,
out of absolute ignorance about Karl Gauss.
(http://en.wikipedia.org/wiki/Karl_Gauss)
Now that i know better, thanks to littletank, i will pronounce it
gowssian, as Karl Gauss rightfully deserves.
Doesn’t it all depend on the value of the standard deviation?
~jim
I apologise if I appear somewhat pedantic but Rolf’s English is so very good that sometimes a not quite right pronunciation stands out like a sore thumb.
I am thankful for hints – really! Especially if you hear something wrong repeatedly. Except the “th”
At last someone has commented on the pronunciation of “Gaussian”. In my line of work (electronics research), and from school & uni, I had never come across anything other than “gowssian” until I saw a video from an English photography magazine. Every time I hear “gawshun” it grates on me, and I assume that the speaker is not that educated! Obviously not the case with Rolf!
Good to hear “gowssian” back again!
Nick
(UK)
[...] After watching this video here (Skip to about 19 Min in, the rest of it is kinda boring…) http://meetthegimp.org/episode-085-geeks-only-really/ I decided to replicate this using the files provided via this link [...]
Here’s a TOC:
http://meetthegimp.org/episode-085-geeks-only-really/
Episode 085: Geeks Only! (Really?)
00:18 Welcome
01:00 SimpelFilter
04:00 Open the original test image in Gimp and examine it
12:23 The Gimp version using a layer stack
14:50 Explanation of burn layer mode
19:10 Use this tool to analyse Gaussian blur
20:11 Analyse unsharp mask
20:55 Analyse built-in sharpen
21:50 Analyse selective Gaussian blur
24:20 Have a look at the images we have blurred and sharpened
26:43 Where to find the file used
28:55 The End