h a l f b a k e r y
Warm and Fussy
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
or get an account
Notwithstanding the proliferation of digital photography,
film is a
great medium overall. One of its chief advantages is its
whereas I'm not sure if I could find a way to read, say, an
era backup tape, I can be certain that, properly
preserved, film will
for 100 years or more. It also has surprisingly
resolution, and can be read optically by both humans and
without much trouble.
These factors combined make it potentially an excellent
long term storage of data, but this raises the question of
how we get
the data on there. We need a device that's small enough
connect to a desktop computer and cheap enough for the
user to be able to afford. We need a YTK Industries
The color film cartridge is loaded into the device much the
way it would be loaded into a camera. The print head
three LEDS (red, green, and blue), and a hole the size of a
recorded dot on the film. The print head moves across
from one edge to the other, recording a single dot after
movement. When the line is completed, the print head
the beginning, and the film is advanced by a single dot's
the process starts again. To speed up the process, the
actually consists of an array of such individual printer
elements, so it
can print hundreds or thousands of dots with each
on multiple lines at the same time) with the computer
of the order that the data is sent to the printer. When
the roll is
printed, it's removed and sent off for developing. Later,
can be read from the film by any scanner capable of the
resolution, or even (in theory) by a human being with a
loupe and a
lot of patience.
But how much data could we actually get onto a roll of
film? Well, a
high-quality, low-noise film such as Fuji Velvia 50 claims a
power of 160 lines/mm, which translates into 320
102400 pixels/mm^2. A frame of 35mm film is 24x36mm,
864mm^2. Multiply that by 3 because we're recording in
RGB. So, a
single frame of film could store 102400*24*36*3 bits, or
megabytes. A 36 exposure roll could thus hold almost
out some of that storage capacity for error correction, and
still looking at more data than can fit on a CD-Rand you
peace of mind that your data will be available no matter
technology progresses in the future.
..it must be true, it's on wikipedia [not_morrison_rm, Jun 14 2012]
Please log in.
If you're not logged in,
you can see what this page
looks like, but you will
not be able to add anything.
Description (displayed with the short name and URL.)
||My preference is sequencing it into the mitochondrial DNA of female goats. Ideal for archiving data, with a nanny-mother goat-kid strategy.
||Worst comes to the worst, at least you can eat the data.
||Interesting concept, but I think you missed a factor of 1000. 33MB * 36 frames is 1.2GB, not 1.2TB. That makes this significantly less compelling if I need 75 roll of film to back up my 90GB archive of digital photos. You might still be able to market it for longevity, but if it had actually been 1.2TB that would have been mass-market worthy.
||[not_morrison_rm] - clever, especially given estimates of the amount of 'junk' DNA which is essentially meaningless and doesn't contribute to the development of the DNA's host body. Replacing this with your email archive, or treasured collection of 'specialist' pictures should cause no adverse effects to the goat. Furthermore, if the goats breed and the goat population grows, your backup and business continuity problems are solved. Finally, data densities are high as you can code in quarternary rather than binary.
||You are correct, [sm]. I was thinking gigabytes but I typed TB.
Thanks for the correction; it's been fixed.
||Consider, though, that this is just the amount of data that can be
stored in a single 36 exposure roll, which is 4.5 feet of film. If
you're really serious about data archival, you can upgrade to the
Pro version of our product, which can accomodate a 1000'
magazine that allows you to store a whopping 250GB of data for
nearly all eternity.
||Or, a single 220 roll film (144cm x 6cm) should be about 25GB of storage.
||Incidentally, in case anyone's interested, but the
concept of storing data digitally on film isn't really
new. In fact, that's how certain types of digital
soundtracks for motion pictures are distributed. The
digital data is printed either in the margins outside
the perforations, or between the individual
perforations. So the actual data density of 35mm
film is somewhat higher, perhaps as much as a third
to half again more than specified in this idea. But a
lot of that space would probably be devoted to error
correction data, since reliability is the main selling
point of the product.
||An interesting problem would arise with this strategy. If
you keep your archives in the form of goats, I could easily
give you a document which encoded some harmful genetic
element, thereby erasing your entire archive.
||Moreover, I could encode a retrovirus, thereby wiping out
the archives of anyone else who shared the same data-
||//Or, a single 220 roll film (144cm x 6cm) should be
about 25GB of storage.//
||Well, 25 gigabits, so only 3.3 gigabytes. Still it'd be
enough to encode an entire feature length movie in
HD, albeit pretty highly compressed.
||//I could encode a retrovirus,
||Sorry I don't do hosted goats, these are my goats, you go get your own.
||Anyway all the really vital stuff is stored on a stand-alone goat, so not much chance of infection. I'm also on the lookout for feed bags, marked "Stuxnet". Also the EULA on my goats comes with a strict "no trolling" policy, I kid you not.
||Someone might break into your garden in the night and RAID your goats though.
||Ahh, but they have special security collars, accessorised by Jean-Paul Goatier.
||//all the really vital stuff is stored on a stand-alone
||Would one goat have sufficient capacity? You'd need
at least a couple of biggergoats.
||I think your goats are really WORMs.
||"We're gonna need a bigger goat."
||//I think your goats are really WORMs.
||I think you're thinking of the Diet of Worms, but I could be wrong.
||Use one of the X chromosomes - humans can survive
(barely) with only one of them, so there's spare
||Ok so I maybe way off base here, but could you not use 32 colours
to code 32 bits in each pix or 64 for a much higher storage capacity?
I'm maintance in a printing house and I have sensors that can see
256 colours easly.In a 32 exposer roll using 8bit to a byte and
1024bytes in a megabyte and 256 colour you get 2.6 gigabytes a cell
or 83.2 gigabytes a roll. Or if you moved to [ytk]'s 1000' roll
||You could, but it would complicate things
considerably, and you'd pay a huge price in terms of
reliability. Color film consists of three layers that
respond to red, green, and blue light. So for each
pixel, you have either a 1, i.e. exposed, or a 0, not
exposed, for each color. In order to use more colors,
you'd need to control your exposure time carefully
and consistently, so the density for each color is
adjusted. You'd also need to have calibration data
for the film stock you're using, since film responds
nonlinearly to light, and different film stocks respond
in different ways. And if your film fades over time, it
could alter the data on the film unless you have some
way of controlling for it when you decode the data.
||Basically, by using more colors, you're turning it into
an analog system. But by limiting it to 3 bits per
pixel, you keep things in the digital world, and thus
keep the advantages of digital over analogsimple
implementation, more reliable decoding, and far
superior noise rejection.
||You still have the problem with achieving consistent
density across film stocks. If you just make a 1 "a lot
of light" and a 0 "no light", you don't need to know
anything about the response curve of the film you're
using, whether it's daylight or tungsten stock,
negative or reversal, and so on. Also, it's entirely
possible that different parts of the film will fade
differently, in which case your calibration data is
useless. And you still have the problem of dealing
with noise that's inherent to analog systems.
||But most importantly, one of the advantages of this
system is that it's inherently human readable, which
means it's inherently machine readable by any
conceivable computer that might exist in the future.
||Well, you two'll have to go into a dark room and see what develops.
||//Basically, by using more colors, you're turning it into an analog system.
||Goats have variance in fur colour, dots. patterns etc. They even have self-rebooting ones in the sub-breed, the fainting goats. See very strange link,
||Well, the link itself is perfectly standard, but the goats themselves are a bit..differently normal.
||There was the ICI Creo optical tape drive from the early 1990s. It had a 1 TB capacity, but with only a SCSI-1 interface, it would take about 3 days to fill one tape.
||//it would take about 3 days to fill one tape.
||Quicker to take the tape out and do the zeros and ones yourself. If you wanted the lazy way, just do the ones.
||Speaking of underdeveloped, where's [Ubie]?
||You're right. I should f-stop, though I'm tempted to
stick a finger aperture friend [Ubie]. Also, this topic
lens itself to puns.
||Did you draw the pentagram with half-baked cheesecake, I always forget that bit.
||//Ilford. Ilford. Ilford. //
||An optional attachment allows you to store the data in punchcard
format on stiff plastic tape. Less data, but longer storage duration.
What would we guess, 500+ years?
||Eh, not sure we'd offer that. Screw my great great
great great great great great grandkids. What have
they ever done for me?
||Although that being said, it /would/ be useful for
certain applications. Like archiving embarrassing
photographs of your enemies. You could send them
a picture of the backup tape, and they'd have to live
the rest of their lives with the knowledge that the
picture of them passed out in a pool of their own
vomit with their pants around their ankles ain't
NEVER going away.