Dorothea Lange, Russell Kirsch, The Scream, Sun & Moon, Drive OST, Stearns - Glitch'd

Hello all and welcome to my Illuminated Errors project. This is a step by step, day by day account of my progress with glitch art under supervision of Phillip Stearns. Below you will find brief descriptions of the methods I used and also several 'deliverables' from each lesson we're taught.

Day 1:

Dorothea Lange's Migrant Mother (1936), Glitch'd.

Hi guys, glad to meet you all.  I've always been fond of imperfect imagery within digital, however this is my first attempt at glitching intentionally. I looked through all the other projects on here (WOW) and decided I wanted to try out something a little different.

Instead of using a screen shot, I downloaded the Wikipedia commons image of Migrant Mother as taken by Dorothea Lange. I chose this image because it was the first photograph to pop into my head when thinking of recogniseable photos, and secondly because I was interested to see what would happen to a black and white jpeg. Before using the image, I resized it to 500px across to simplify the code and so small adjustments wouldn't appear as tiny lines.

I made a few dozen glitches using both Textedit and Hex Fiend. Using trial and error I was able to work out that (to my surprise) the code is in order with the image. Ie. The top part of the code is the top part of the picture, the middle, the middle part and so on. With this in mind I wanted to see if I could 'find' and 'glitch' the Migrant Mother's eyes...

These are of course not as abstract as Philip Stearns' creations, and I wasn't particularly pleased either. I started experimenting with adding text from the Migrant Mother's Wikpedia page and placing it at random intervals in the jpeg code. I'd choose large portions of text and apply it about a third of the way down. Here are some of the nicer results:

1. This first one has plenty of those pretty macro blocks.

2. I'm not sure how this has happened, but it looks like the Migrant Mother has been inversed and shifted along. It's like a pixel ghost.

3. Rule of thirds, anyone? I'm keen on this aesthetic.

I then switched to the Hex editor to see what sort of results I could get. This is probably the best from a dozen. I found it problematic to get these to work as first because I was editing too high up in the hex code. The macro blocks are sparse, but definitely there. To achieve the look I pasted "Migrant Mother" in intervals throughout the text:

Now I'd experimented and had a lot of fun, I wanted to apply something more than just the technique. I noticed that "Dorothea Lange" and "Migrant Mother" both had the same amount of letters. I switched back to the full resolution file first. Then I used the Text Edit software to 'find' M and replace with D (the first letters). The result was interesting:

These wonderful lines appeared across the image, seeming to emerge from the right and dissappear by the time they reached the left.

I continued using the same picture, adding (or replacing) each letter in the names.

M > D, i > o, g > r, r > o... etc for all 13 letters. Two didn't work because they'd already been replaced. The final outcome looked like this:

Phew.. I'm exhausted now! That's a lot of glitching! But I still hadn't finished. There was one last image I wanted to attempt. Instead of adding to the image like I did above, I individually replaced the corresponding letters to the picture on 13 seperate files (13 letters). With the output, I went into Gimp and created an animated GIF with the 13, cycling continuously at 100ms.

Here it is:

All corresponding letters of Migrant Mother replaced with Dorothea Lange.

Apologies for the large file size (it's 1.6mb). 

That was a lot of fun. Look forward to the next lesson!

DAY 2:

Russell Kirsch, Inventor of the Pixel

- Creating animated glitching GIFs with GIMP.

After yesterday's fun with my very first (deliberate) glitch creations I was looking forward to experimenting more with animations. I wanted a different subject to use and to add a bit of colour to my project. I decided to have a base starting point of the inventor of the pixel, Russell A Kirsch. He famously scanned the first ever photograph into a computer in 1957. Here's a picture of Russell and the 'first ever digital photo'.

I quite liked the portrait of Kirsch so decided I would use that, and also experiment with the first photo. Incidentally the scan is a picture of his newborn son.

For both pictures I went through the same procedure to create a GIF. It took several steps. What I wanted to see at the end was how different or how alike the two resulting GIFs would be from very different sources.

1. Open image in Textedit. Replace 1 with 0.

2. Screen shot image preview of jpg.

3. Open the .png in GIMP, remove 1px grey border, save as .jpg w/ 85 quality

4. Open .jpg in Textedit & replace 1 with 0.

5. repeat stages 2-4 a number of times.

These are the two resulting GIFs from this simple method:

Above, Russell Kirsch 5 frames.

Below, First Digital photo 5 frames.

Interestingly the black and white scanned image turned colour. It was also the less detailed of the two pictures, yet it appears to have more detail than Kirsch's portrait now it has been glitched.

I wanted to apply the same animation process to a colour image as I did yesterday to Dorothea Lange's portrait. I of course chose Kirch's portrait again, but this time used a much smaller res file. I did this because I wanted to make the pixels very prominent in his animation, as after all he did invent the pixel.

Here is the same process I used above (DAY 1, Migrant Mother GIF) applied to Kirsch's smaller, colour portrait:

YEAH! Now that, I like!

So what else could I try doing? I really liked this portrait, and the idea of glitching the face of the man who made this all possible in the first place (no disrespect to you, Phil). I took the original downloaded portrait again and enlarged to 3000px to work with that. 

I used the simple Textedit method again and created 9 glitched files. All using find and replace method. 

File 1: find 1, replace with 0.

File 2: find 2, replace with 1.

File 3: find 3, replace with 2 ... and so on.

Once I had finished doing that, I imported all the images as layers in GIF, shrunk the image down to a more manageable size and animated just as I had done above.

If you suffer from any form of epilepsy, please look away now.

(Link to GIF - http://i.imgur.com/7E2xI9P.gif )

Hey Russell, you're looking colourful today! What an awesome mash from a single image! It's hard to believe that was once Russell! 

Maybe I should put him back in...


(Link to GIF - http://i.imgur.com/VBV8oDV.gif )

That's better.

But still, something's not right.

Edgar Allan Poe said "The eyes are the windows to the soul" right?


(Link to GIF - http://i.imgur.com/Wzg22ye.gif )

WOOOOOAH MAAAAAAN, John Lennon would be jealous! 

OK, I think I'm down there, but I had one last challenge for myself before I called it a day. Glitching a GIF file.

Would it be possible to take the above psychadelic portrait of the great Russell Kirsch and glitch it even further? Is there a point of mashing up already cooked mashed potato? Well, I gave it a try.

I opened the GIF in text edit and once again replaced all 1s with 0s. (I love the link to binary there.) I saved that file and got a strange square looking image. I triend it again with 2s replacing 1s and 3s replacing 2s and so on.

Out of all 9, the best results came in the 5s replacing 4s. It manipulated the image just enough to keep the essence of Roger there.

The GIFs wouldn't open in GIMP so I had to use preview to view them:

It was pretty easy to use "CMD+Shift+4 space" (thanks for the tip, Phil) to get these all into Gimp. I then simply cropped to the image and created a very creepy Kirsch looking animation.

Here it is:

(Link to GIF - http://i.imgur.com/TdKWYeI.gif )

I love how you keep getting a glimpse of his cheesy grin! 

That's me done for today. I really enjoyed animating and I'm looking forward to my next lesson.

Day 3:

The Scream by Edvard Munch, Glitch'd using Audacity (an audio editor.)

For our next lesson we were shown how to edit TIFF (uncompressed images) files as raw data using audio software called Audacity. I found this immediately interesting because we were given another visual representation of the image, in the form of a sound wave. What's better was that you could actually play the image as a sound, and although it sounded indiscernable from white noise, it was interesting to hear what a digital image sounded like, well, Audacity's interpretation of it.

I chose Edvard Munch's 'The Scream' painting to work with and downloaded the image from Wikipedia. The idea of using an artwork which was all about a sound and that had never been given a chance to make one is an interesting premise. But I wasn't about to finally giver the screamer her(?) voice, I was merely using the hidden sound to desecrate the painting.

First I had to convert the .jpg to a TIFF file in Gimp so I could work on it.

The Scream, or 'Skrik'.

The visual interpretation of Edvard Munch's scream as a sound wave filled out the mono channel I opened it in. The sound it made was loud, constant and dirty:

At first I experimented with Cmd+X, Cmd+V (cut and paste) on the sound file, making sure not to touch either end where the header information is. I also discovered that it was imperative to keep the file the same time length, otherwise it would corrupt the TIFF beyond opening when exporting.

Here's an excerpt of images from the dozens I created. Each one a manipulation of the file before it.

After several cut and paste style glitching I took the TIFF back into Gimp to rotate it so I could make vertical glitch lines across the image:

I continued this process until the image was completely indistinguishable. What I was left with was a complete mess:

With these files I had the opportunity to make an animation of the 'Scream Soup' showing the four last manipulations:

Cutting and pasting was fun, but it can only go so far as a tool, so next I experimented with some of the available effects Audacity had on offer.

Using the same process as before, I selected small portions of the original Scream image and used different effects on it. These included 'Reverse', 'Invert', 'Phaser', 'Echo', 'Bass Boost', 'Equalisation', and 'Wahwah'. I then rotated the resulting image in Gimp and applied the effects again to give a crosshatching(?) of Glitch effects.

I really liked this image on it's own. It reminded me of some of Phillip Stearns' textile pieces. I wanted to show it off in more detail. I took 200px square crops from around the frame and enlarged them to 500px without interpolation to create a glitch animation:

I've slowed the frame rate down to 150ms so you can see the variety of different glitches in the one image.

During the exercise I found it interesting that some of the Audacity effects mimicked corresponding filters on Gimp. Invert for example would produce a similar (but certainly not the same) result. Reversing the sound file would flip and shift the image. The 'Echo' effect is one feature I applied several times to the raw TIFF file and ended up with a nice, dark aesthetic. 

On some of the earlier versions you could 'see' the echo getting louder and more intense, thus adding to and darkening the image.

The 'Phaser' effect also rang true with it's nature of adding a type of phase to the image file:

You can see from the sound wave how the same effect is mimicked in the imagery.

This lesson was extremely exciting for me. I wish I played with the sound element of the files more, but perhaps that's for another day. I learnt how all uncompressed code is very similar in nature, if anything is to go by audio and image files, anyway.

Looking forward to the next set of tasks and tools!

Day 4:

Data Mashups using Audacity. The Sun, the Moon & Neil Armstrong.

With Audacity we have the ability to blend two completely different uncompressed TIFF images into one. For this lesson I wanted to use two of my own photographs I'd taken rather than nipping one from Wikipedia as I'd done previously. Because of the phenomenon of the Sun and the Moon being the same size to our eyes in the sky, and my interest in astronomy I thought these would be two good images to use:

Both photographs are crops from a larger frame, shot with a Canon 5Dm2, 400mm with a 1.4x. Neither have had any post processing apart from the crop to fill the image.

A straight 'mash-up' of both images using Phillip Stearns' technique produced quite a beautiful result:

NB. It didn't matter if the Sun track was above or below the Moon track, the image produced was still the same.

For the reason I don't think it's fair to say that the Sun is 'behind' the Moon here. The case is that they are mashed together. Immediately I thought it looked like one of the early faux-colour NASA images of distant Moons and planets. Isn't it also strange that the black space around bothe cellestial bodies changed grey when grouped together?

I wanted to add a third element to the image because it didn't seem finished to me yet. As I was in Audacity still I thought it may work if I added a sound file. Of course there was only one I could use...

"One small step for man, one giant leap for man kind". 

I was able to download a WAV file direct from the NASA website to use with the image. It included a sentence before the famous one which said "OK, I'm going to step off the LEM now". When importing the audio into the image I deleted the first sentence as I thought it wasn't necessary. I was also delighted to see that the WAV file fit snuggly into the Sun and Moon files.

Here's a screenshot from Audacity showing all three tracks:

You can see I left the header on the Moon track, at the right of the audio.

The result looks more like interference to a television picture, mirroring the Moon landing itself back in '69:

Cool stuff!

The audio covers the Sun and Moon nicely but seems a lot more sparse than I thought it would. You can actually see where Armstong said "ONE" for the second time about two thirds down the orb, as a lonely word, seperated from the two phrases.

I had an idea to turn this into an eclipse GIF with the Sun slowly moving across the frame into the Moon's position, however when I tried it I didn't get quite what I expected...

This made me realise how lucky I was on my first exported TIFF of the two. If the Sun's file was just a fraction to the left or right in Audacity it wouldn't have covered Moon. As it happens, it did. This GIF is made up of 21 individual frames and I have left Neil Armstrong's famous speech in which you can se shifting nicely down the image.

I may try again to keep the Sun along the same path, but as it happens, I think I'd prefer this anyway.

Day 5:

All my Love, track 4 from the Drive OST.

I found myself a bit lost with this lesson on what to do. There's generally been a concept or kind of back story to each of my projects so far. It probably doesn't matter, but this project doesn't have any deeper meaning than the piece of music I used is one I like. 

I chose a track called Oh My Love (feat. Katyna Ranieri) which I discovered in the movie DRIVE (starring Ryan Gosling.) If you've not seen it or can't remember which one it is, let me refresh your memory with this absolutely stunning clip from the film - http://www.youtube.com/watch?v=Q5gyRtg262c

I opened the file, and just like in Philip's tutorial it was a plethora of different pixels with no real design that could be seen with the naked eye. In fact, it was pretty grey and bland:

From this it was pretty easy to glitch up the sound and create audio that resembled the original with clicks and whirrs, but to me this was still pretty boring. I also discovered it was pretty easy to mess up the sound file so it wouldn't play, which was pretty frustrating.

I far more enjoy creating a visual piece that I can look at as one. So this is what I continued to do. First off I cropped the image above well below 100x100 and then enlarged it without interpolation. 

You may have spotted that I added the 'Drive' logo to the bottom corner. 

Of course this would look far better as a GIF. I started again with the original mp3 file and cropped it to 100x100. This looked great:

However the GIF was far to big in file size and reducing the colours didn't help much. So here's a 200px version (under 1mb) enlarged. I think you could argue it really suits the '80s vibe the film has.

So yeah, that was cool, but I could certainly do something more with it. I tried adjusting the levels on the GIF, increasing the bottom third up 20 points a time for about 20 frames. I layered them all for a GIF and zoomed in as it cycled through. The result looks like an explosion of sorts:

Playing with audio files is a lot of fun. It was by far the most temprimental method I used. Perhaps it was the file formats I attempted it with, I'm not sure.

Before I finish up I wanted to try one last project. An homage to our tuor, mr Stearns. It's not very good, but I had fun doing it... Here we go...

Final project:

Phillip Stearns. Glitch'd.

Here is a picture of Phillip Stearns. I found it somewhere online. It's a 200px crappy jpeg. But that doesn't matter really, does it?

[Phillip Stearns 1]

To make it useable, I opened it in Preview, zoomed in and took a screenshot of it. and took a screenshot of it. I then opened that up and saved it as a TIFF so I could play with it.

[Phillip Stearns 2]

Ooh, pretty. As it was a screenshot of a zoomed in image there is some automatic interpolation.

I then used Textedit to use the 'find & replace' technique. Replacing every letter from a to m with the next letter that proceeded in the alphabet. Here's a GIF of the 11 frames and the progress. It's interesting, but pretty boring. I also found it interesting that I could recreate some of the glitches I experience in some earlier projects.

[Phillip Stearns 3]

As you jan see, Mr Stearns goes slightly jaundice and does a little chin wiggle. Nice one Phil.

Next I moved on to HEX Fiend to shift about some large bits of code. Because the TIFF file was so large, this wasn't very easy. I spent quite a bit of time on it and this is what I came up with:

[Phillip Stearns 4]

As this was a slow and tedious process, I switched to Audacity and things speeded up quite a bit. Before doing that, I rotated the latest image to shake things up a bit:

[Phillip Stearns 5]

Then, using Audacity to create the same method as cutting and pasting in Textedit and HexFiend made things a lot faster. The image was now starting to look like a glitch pic indiscernable from its original:

[Phillip Stearns 6]


Time for some CSI! I'm going to zoom in and have closer look at this image.

I took several 100px crops and strung them together to create a glitch GIF. 

[Phillip Stearns 7]

I learnt a few days ago that I am able to glitch a GIF! I used Textedit to Cmd-X, Cmd-V a few portions of code around on the above GIF. It didn't look that great, but I was able to pull out one of the resulting frames which looked interesting.

Not only did I use the frame for the next step in the process, I zoomed in once again taking just a 68px square crop and resizing it to 1000px.

[Phillip Stearns 8]

Is it getting boring now? No, not for me. It's just starting to get interesting. We've now zoomed in further than the size of a single pixel. I wanted to bring Phillip Stearns original portrait back into this somehow, so I did this by doing a datamash using Audacity.

[Phillip Stearns 9]

Hmmm... Quite dissappointed Phil's face didn't appear again. Perhaps I'll manage to get it back a bit later.

I put the file as it was back into Audacity and used the equalize effect. Unfortunately this turned it even more bland. Pah.

[Phillip Stearns 10]

Only one thing for it.

Zoom in, crop, enlarge without interpolation.

Then, I will bring the file back into Audacity and use the invert effect. Here's a GIF showing the cropped in version before and after the Audacity invert:

[Phillip Stearns 11]

All this time I've been working in TIFF. I'm in desperate need of bringing some colour back in so I'm re-encoding this image as a jpeg. It should produce some more dramitic results.

Using textedit I replaced the following letters:

S -> G

T -> L

E -> I

A -> T

R -> C

N -> H

S -> (already been done.)

and was delighted with this lovely set of jpegs. Much more colourful!

[Phillip Stearns 12]

This is a set of 6 glitches as a collage. Original files sizes were 500px x 500px.

Time to bring Phil back into the fray with a bit of cheating. Some layer effects using Gimp.

[Phillip Stearns 13]

Hey Phil, looks like youve got yourself a new Facebook / Twitter avatar? Huh? yeah? :-D

Keeping in line with my other projects on here, and as I've become very fond of animations, there's one last treat. A nice, frenetic animated version of the above avatar. 

And now, I think I'm done!

[Phillip Stearns - FINAL]

Now, that's quite pretty if I do say so myself.

If you're interested. The final filename of this mammoth final project was: 


What a beast.

Thanks for looking through my projects. I've really enjoyed myself here and am super grateful for Phil for dispensing his knowledge on us. I'll certainly continue to experiment and practice more.

If you were to ask me my favourite, I don't think I could answer. I'd need to spend many more hours playing before I decided which method I'd carry on using most.

BTW, I've been keeping up to date with everybody elses projects. I've really enjoyed watching everyone get on and have been inspired by so many - I'm sure that's obvious.

Anyway, over and out, I'm done!


Sam Cornwell.




Please sign in or sign up to comment.