I built a camera like that out of wood a couple years ago. It uses gears on the inside to pull a regular roll of film through and expose it through a narrow vertical slit as you rotate the whole camera on its center axis. It's sort of a broad-stroke photo.
http://nnife.com/d/?dimension=33
Here are a few shots that turned out decently: http://nnife.com/p/?dimension=20 These are just film scans. You can by the light bars see on the photos that it was very hard to rotate the camera at a constant rate.
You can also see that as I went with/against the traffic, cars were elongated or shortened. Really interesting phenomenon. I was working on this with a painting teacher at the school I go to, whose fascination with unconventional forms of photography got me interested as well. Unfortunately I was a woodworking noob when I made it so its success rate was too low for me to keep using it and I never took many more photos than that.
If I ever get around to building a better version I want to try an idea he had: mount the camera in a spot and expose it over the same piece of film at three different times of day using red, green, and blue filters. That would add another level to emphasizing time in a photograph.
The teacher I was working with is named Nicholas Evans-Cato. The experiment started off as just one to make a film panoramic camera. You can see why he'd be interested in that if you look at his work. http://www.georgebillis.com/artists/nicholas_evansCato.html
I'm sure it's been done before, but it was a really fun experiment. Changed the way I think a little bit.
Note that this is slit-SCAN, which is a bit different in that it moves the "slit" across the face off the sensor, rather than capturing from a static slit over time (the same technique was used in Star Trek:TNG to show the Enterprise going into warp). Nonetheless, you can play with time similarly.
There's also the (admittedly more expensive) wonderful Widelux camera and its various Russian knockoff models. Jeff Bridges uses it to wonderful effect to document on-scenes at most of his movies, and then present the staff with a book when shooting closes. http://www.jeffbridges.com/camera.html
You can just make a video, and select a column of pixels. Then place the different columns (color-wise) you get from each frame next to each other. The resulting image will have the same width as the number of frames you used and it will probably have a lot of artefacts considering the quality of today's (home) video technology. But it is essentially the same.
And pretty much everyone here would be familiar with a couple of artistic uses of slit-scan photography - the opening credits of Doctor Who ( http://h2g2.com/approved_entry/A907544 ) and the ending of 2001: A Space Odyssey.
I thought that as well. I can't imagine anyone who reads athletics/cycling reports wouldn't have recognized those for what they were. He's not far wrong in the article, to be fair - just in suggesting that the artist invented the technique. But that suggests he did no research beyond a misleading gallery press release, since just googling the term brings up lots of interesting stuff.
I don't mean to denigrate the art though, because the pics are still striking and interesting.
I was also reminded of that weird final scene in the Animatrix short about the runner when he has some kind of warp spasm before transcending and waking from the Matrix. That just gave me a renewed appreciation of animation.
It seems like it would be simple enough to convert a video into an image like these by isolating a single column of pixels from each frame and putting them together chronologically.
#starting at 467s and outputting the images of the next 3 seconds
ffmpeg -ss 467 -t 3 -i video.mkv output.%04d.png
# crop image to 1px slices 280 pixels from the left.
gm mogrify -crop 1x+280+0 *.png
# stitch them
gm montage -border 0 -geometry +0+0 -tile x1 *.png image.jpg
Perfect example how easy and straightforward it is to use command line utilities.
Now someone just has to find movies with static camera scenes that have a lot of horizontal actor/object movement.
Corollory : One could produce a whole range of image0%d.png from vertical slices through the video, and stitch them together into a movie. Essentially rotating the movie abouts its y axis (in the x-t plane).
If we assume 50fps, then every 1000px wide video would become a 3-4 minute short. But movies would require a very wide screen...
Cool demo. The 'gm montage' step isn't working for me and I'm finding its options completely non-intuitive. I produced a series of 1979 PNGs from a 66-second video, and when I stitch them with:
Note that I made a couple changes: to speed things up I used "-vf crop=1:1080:280:0" with ffmpeg to skip the separate cropping step and output 1x1080 PNGs, and my version of gm didn't accept the -border option so I assumed you meant -borderwidth.
What's up with the crazy widths on the result? What is GraphicsMagick doing?
This is how goal line cameras work - there's one at every race course and sporting event around the world. I had a weekend job running one at a race course growing up.
Nowadays they are digital I guess, but the technology is the same - you record only a sliver (the goal line) - the analog version has the film running past the sliver on spools, with the speed being adjustable based on how fast objects you want to record.
It might be interesting to see a video where each frame is one photo of this, and over time it goes through the different X-positions. Effectively swapping the horizontal and time axis in a video.
If you start with a video you can do this. You'd transform an X x Y x T dataset (X x Y dimensional frames shown over a period 0-T) into a T x Y x X dataset (T x Y dimensional frames shown over a period 0-X).
For a long video you'd get a very wide image that varies over a pretty short duration. A 1920x1080 video clip that's 80 seconds long @24 fps would come out right. It'll still be 1920x1080 and 80 seconds long after transformation.
Horse as well. Throw in a well positioned tall but narrow mirror on the opposite side of the track, and you get a full view of how the race finished in a single photograph.
As many others have mentioned, this is a well-explored field. Golan Levin maintains a very extensive list of slit-scan camera experiments, check it out for some awesome examples:
This was made on a large format camera -- probably something like 4 x 5 inches -- with a focal plane shutter. The shutter has a huge distance to cover -- 4 inches -- and it takes a while for the slit to cross the whole film frame.
The camera panning on one direction caused the spectators in the background to be distorted in one direction, while the car was moving faster than the camera and is distorted in the other direction.
Its interesting to note that the distortion was mimicked in cartoons as a way to show speed.
Ward Cunningham and I built something similar with an old Mac mini and an external firewire iSight camera when we both worked at AboutUs a few years back. We pointed it at the train tracks behind the office and got some cool images of trains as they passed by. I don't think we saved any of the images, though -- it was just a quick after-hour hack project.
On a related note, the new Panorama function in iOS 6 seems to turn the iPhone's camera into a slit scanner using the phone's accelerometer for the rotation rate to stitch together a pretty good panorama on the fly.
"...this technique uses spatial + temporal data stored in a 4D Space-Time Continuum, and 3 dimensional temporal gradients (i.e. not just slitscanning on the depth/rgb images, but surface-scanning on the animated 3D point cloud)."
CINECHAMBER CHANNEL
BIOSPHERE - GENKAI-1
Official video by Egbert Mittelstädt From the album N-Plants
It's performed in the cinechamber which is four huge video walls with I think an eight speaker array. The time distortion is amazing and you are totally immersed.
This reminds me of something I saw a while back (it might have been on Boing Boing or something, and I'm sure it made its way across the web) of a guy who had a camera he'd made out of an old scanner. Basically, the scanner would move across the surface very slowly so it captured a slice of time. I wonder if that sort of technique is being used here.
Look up "The Fourth Dimension" movie by Zbigniew Rybczynski - at first sight it is hard to believe it was made in 1988 because it looks like a product of some heavy digital 3D processing. In fact it is each of the 625 horizontal scan lines shifted a bit in time relative to the previous one, an idea very similar to the one described in the article.
I did something similar once with a flatbed scanner. Put my hand/face/whatever on the bed and moved it parallel to the scan head. Made for really long creepy fingers or a smushed up face.
every time I see parts of human body like that, I have a flash backs from the very first time when I saw Aphex Twin's Rubber Johny. Worth watching, IMHO :)
kind of similar work of zbig rybczynski way way back (using analogue gear). on a side-note, discussed this with sean of autechre on how to achieve the same effect in maxmsp/jitter by delaying each horizontal line :)
This technique was used in a slightly different way on NRO satellites. I think KH-7 was the first (GAMBIT http://www.thespacereview.com/article/1279/1 ). They would get the film moving at the same speed as the ground that they were covering. The constant momentum of the film moving was easier to counteract than the constant back and forth motion of a big shutter if they were capturing film "plates".
I was looking for this recently! There is a photo, either form this artist or someone else, that was taken at some sort of festival, a photo of all the people walking by.
It was featured in some website that made the rounds on reddit or slashdot or somewhere.
You can get the same effect with a flatbed scanner, which operates on the same principles but assumes a fixed subject, so the camera moves while scanning. I can't find a link now, but there is a fad of rolling your face on a scanner and sharing the result on a website.
Here are a few shots that turned out decently: http://nnife.com/p/?dimension=20 These are just film scans. You can by the light bars see on the photos that it was very hard to rotate the camera at a constant rate.
You can also see that as I went with/against the traffic, cars were elongated or shortened. Really interesting phenomenon. I was working on this with a painting teacher at the school I go to, whose fascination with unconventional forms of photography got me interested as well. Unfortunately I was a woodworking noob when I made it so its success rate was too low for me to keep using it and I never took many more photos than that.
If I ever get around to building a better version I want to try an idea he had: mount the camera in a spot and expose it over the same piece of film at three different times of day using red, green, and blue filters. That would add another level to emphasizing time in a photograph.
The teacher I was working with is named Nicholas Evans-Cato. The experiment started off as just one to make a film panoramic camera. You can see why he'd be interested in that if you look at his work. http://www.georgebillis.com/artists/nicholas_evansCato.html
I'm sure it's been done before, but it was a really fun experiment. Changed the way I think a little bit.