Homemade Digital Cameras

Michael Golembewski writes "For the past three years, I've been taking apart cheap secondhand flatbed scanners and turning them into homemade large format digital cameras. They are well over 100 mexapixel in resolution, and produce results that are both similar to and significantly different from traditional digital and conventional cameras."

Re:why not just post-process?

[penguin-collective]

The grayscale, banding, and vignetting are easy.

For the scanning effects, you take a video or continuous shooting (most digital cameras support both) and simulate the scanning by taking scanlines sequentially from successive frames.

Re:why not just post-process?

[HaMMeReD3]

If you rtfa it's because the distortion from a camera like this is because it captures the image in a series of scan lines, each scanline is a small exposure, but the entire picture is made up of thousands of slices. This is different from a regular digital camera where the entire sensor is exposed which gives a motion blur effect. A scanner camera does not give you motion blur but actual clear distortion as the world changes during the exposure.

It's a different effect and it's captured through optics and time, it can not be replicated perfectly in a computer and photoshop is not the solution to all problems artistic. This is a very creative idea and I plan on personally converting my 4x6 scanner to take picture like this. It's an original idea, I'd be interested to see if color filters on the lens would allow you to take multiple exposures for red/green/blue and mix them in the computer to create a color image. I think it could make some very interesting pieces of art.

Re:Amazing tech skills with art value!

[dirvish]

I agree. This is very unique and interesting art. There are some very unflattering portraits on the oddities page [scannerphotography.com]

Re:Here's another idea.

[robbak]

Self-reply: they are called "line scan" cameras, for anyone searching .

Time-lapse photo finishes

[Kayamon]

This is kinda similar to the technology they use for doing photo-finishes in track & field races.

See here - http://www.sportingworld.co.uk/newyearsprint/pics/ 2004/photofinish_75pc.jpg [sportingworld.co.uk]

The best ones are when somebody puts their feet on the finishing line, and it gets stretched out to several "metres" long.

Argh!

[HoneyBunchesOfGoats]

It seems he figured out how to script the camera to take continuous photos, and animated them into movies [scannerphotography.com]. Unfortunately, these movies require Flash 8 to play, and the latest Linux Flash Player is v7! This is the second site today that has kept me from viewing content because of this issue (though in this case it simply seems inadvertent; by contrast, you can't access any of animationmentor.com [animationmentor.com] with Linux).

Re:why not just post-process?

[penguin-collective]

If you rtfa it's because the distortion from a camera like this is because it captures the image in a series of scan lines, each scanline is a small

But it is the same way existing digital line cameras work, and it's the same way film-based line cameras work, yielding, not surprisingly, the same effects.

It's an original idea,

No, it's not. Even the consumer-scanner-as-large-format-camera is old.

I'd be interested to see if color filters on the lens would allow you to take multiple exposures for red/green/blue and mix them in the computer to create a color image. I think it could make some very interesting pieces of art.

You mean like Technicolor? Or like Autochrome? Or like three-CCD analog camcorders, digital cameras, and digital camcorders?

Re:Brave guy

[ars]

Just in case here is a coral cache of the site: http://www.scannerphotography.com.nyud.net:8090/ [nyud.net] I clicked on every page to load it up into the cache.

Re:why not just post-process?

[penguin-collective]

This effect is impossible to create to anything near this level of detail or clarity using traditional digital tools.

You can buy a high-resolution scanning digital camera off the shelf, which gives you exactly the same distortions but actually produces excellent still images. You can buy a used Horizon camera and get the same effect on film, minus the banding, stuttering, and poor focus. You can look on the web for "slit-scan photography" (used, among other things, in the film "2001"). You can do this sort of thing with any old large format camera. Or you might look around the web for the same hack done in the mid-1990's, multiple times.

Finally, I really, really, really don't understand why these types of comments are made. It's so infuriating - do you have any sense of exploration and experimentation? Or understand the desire to tell others about your experiences?

No, what's infuriating is when people do the same "hack" over and over again. At some point, it ceases to be a hack and just is a pathetic display of inexperience.

Re:Recycling in a Good Way

[dangitman]

This is a good idea for recycling old equipment. I have several of those laying around,

Several of what? An old scanner is pretty useless without a decent lens with large area coverage, and a housing to mount it in. That's not exactly cheap. If you have old large format cameras or lenses just lying around, then getting a scanner is the least of your problems.

I don't know about you, but I have Horseman 4x5 cameras coming out of my ass.

Re:why not just post-process?

[Anonymous Coward]

Why have a chip on your shoulder about non-elitist equipment?

I don't. I have a chip on my shoulder about people claiming something as artistically and/or technically new when it has been done numerous times before, and often better.

Here [sentex.net] is one link. Here's [rit.edu] another one. There have been a number of other variations, including leaving the scanner in the film plane of a LF camera.

Re:Argh!

[heatdeath]

The movies are actually not very good at all. (The pictures themselves are far more interesting)

Re:115 Megapixels?

[heatdeath]

The scanner he uses on his primary camera is 600x1200 dpi, so he's clearly not talking about that. (a full 8x10 scan would be 65 megapixels, but the 4x5 frame would only be about 15 megapixels) Kinda confusing as to why the images are so large then...that's not much bigger than my DSLR. Even a RAW image with my 8 megapixel camera is only about 8Megs.

Similar tinkering

[Alkind]

Andrew Davidhazy has done similar things at the Imaging and Photographic Technology
Rochester Institute of Technology years ago. His site is interesting

http://www.rit.edu/~andpph/ [rit.edu]

Many have done the same later on. I got through a Christmas period converting a Umax page scanner to a panorama scanner. It was fun.

http://www.pigment-print.com/Panorama%20Camera%201 /index.html [pigment-print.com]

Re:Amazing tech skills with art value!

[gormanly]

ADD ? Attention Defecit Disorder ? Advanced Dungeons & Dragons ? WTF??

Re:why not just post-process?

[hummer]

The similarities with slit-scan photography immediately stood out to me as well.

For anyone that's interested, there's a reasonably good page describing the technique here [rit.edu] and pages about it's application in the stargate sequence of 2001 here [seriss.com] and here [underview.com].

It's possible to fake the technique in Adobe aftereffects with the time displacement filter too.

Coral Cache (just in case)

[HazE_nMe]

http://www.scannerphotography.com.nyud.net:8090/ [nyud.net]
With that many pictures I would have used the coral cache link in the summary.

Re:115 Megapixels?

[bogado]

A raw image should have more then 8Megs, since each pixel has at least 3 bytes, but since many cameras that provide raw uses 16 bits per channel this would acount for 6 bytes per pixel. On the other hand, if your raw has the information direct from the ccd that usually is a black and white sensor that has a colored mask in front of it this would make each pixel to have a single channel (usually in an array of 4x4 squares that hold red and green in the first row and green and blue in the second).
So in this configuration the raw file would hold 16Mb more or less. If this file is compressed with a non-lossy (gzip, zip, bz) compression it can be expected at least a 2x compression rate, so it would re-shrink it to 8Mb.

So I guess that it is not that obvious that a 8Megapixel camera will have a 8Mbyte raw file, even if it seem obvious.

Re:Dupe, happended 7 years ago

[scharkalvin]

The guy in the earlier link gutted the scanner and put the guts in a box with
a lens. In the newer article, the scanner was used as is, but with the lamp
removed. I like the latter approach better. So this isn't a dupe,
but a new improved method.

Re:115 Megapixels?

[maxwell demon]

On digital cameras each pixel delivers only one color channel by using a Bayer filter [wikipedia.org] (unless you have a 3-chip camera, of course). The colors are then interpolated into those pixels which measure different color channels. Therefore you have only 1 to 2 bytes per pixel in the raw format.

Mirroring Help!!! (SITE'S BACK UP)

[majk_g]

Well, I didn't expect that to happen... My bandwidth limit was drastically exceed this morning. As you know. Luckily, some great people have offered me a bit of help... so you can see the project again. I've put up a link on ww2w.scannerphotography.com Thanks so much for looking at my work... I hope you enjoy it! Mike Golembewski

Re:Time-lapse photo finishes

[Anonymous Coward]

The photo linked to was done by a FinishLynx timing system. I setup and run the one for the local high school. The results look odd, but once you understand what it is doing it is obvious. First notice in the image the track has white lanes with black separators. Ever see a track like that?

The image you are seeing is best thought of as a graph, not a photograph. The camera/system is recording the image/pixels from a very thin line (maybe 2 pixels) across the track at high speed. Think of the output as a graph where time increases to the right. The red line you see in the image is used in determining the times. It is used like a cursor. Start (time 0) is at the left edge and as you move the cursor right you can read the x value (the time of that particular set of pixels. The Y axis is the pixels across the track, with the bottom being at the camera, the top across the track.

The camera is set up on the center of the finish line. That is why the lanes appear white, because at that point on the track, the lanes are white and the dividers are black.

Gaffer's Tape = Duct Tape

[SoupIsGood Food]

Next time try some Gaffer's Tape... reknowned in the studio photography world to stick like a sumbitch, leave no residue when peeled away, and block light completely. It's a high-threadcount black cotton tape with masking tape adhesive, and it realy is dark-safe.

SoupIsGood Food

You can distort with a 35mmSLR too

[cellocgw]

35mm SLRs have mechanical shutters. Most (not all) work by releasing two curtains in sequence. The first retracts from one side to the other, letting light reach the film. The second extends in the same direction shortly thereafter, covering the film again. What you get, then, especially for short exposures, is a negative which was exposed at different times from one edge to the other. Normally you can't see this effect, but it is possible to capture (or miss, depending on the direction of motion) highspeed events in the field of view.
Essentially the shutter mechanism is acting like a mechanical line-scan.