h a l f b a k e r yFlaky rehab
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
Please log in.
Before you can vote, you need to register.
Please log in or create an account.
|
The problem with film, esp. old film, is that it degrades with time. Making copies introduces noise into the image. So, you have problems if you are trying to preserve photography.
With digital media, you can make perfect copies. Until recently, however, digital photography could not match the resolution
of film and be reasonably priced.
I propose that one use a high resolution scanner to examine photographic film, and make a map of each silver halide crystal on the film.
I assume that these crystals are roughly spherical, so the only data you need to record are the position and what kind of crystal it is (thus telling you the color).
This would reduce the amount of data for digitizing film, compared to high-resolution scans.
If need be, you could even recreate the photo by planting cyrstals according to the map!
Silver Halide Crystals
http://images.googl...safe%3Doff%26sa%3DN
This is not TAB. [lawpoop, Oct 04 2004]
[link]
|
| |
Sounds like a good way to increase filesize and processing time, at least. |
|
| |
[FloridaManatee] Filesize would decrease because you're only storing the co-ordinates and type of each grain, as opposed an entire bitmap of the thing. |
|
| |
Typically these are secondary concerns in archiving anyway. |
|
| |
Except for the newer T-Grain stocks from Eastman. The grain crystals on these stocks are flat tab shapes. |
|
| |
Maybe I don't understand the mechanism of your idea: |
|
| |
How many cystals are there on an exposure of silm stock? Are these crystals layered one deep, or is there overlap? Are they uniformly shaped? How many types of crystal are there? How many shades of exposure do they come in (or you sample in)? What is the resolution of the x-y coordinates >> Therefore: How many bits of information are required per exposure? |
|
| |
Next, how do you propose to measure the position, type, location and exposure of the crystals and how long can would this take per exposure? |
|
| |
And finally, what kind of process would you use to place crystals individually on the substrate? |
|
| |
FloridaManatee - i don't think lawpoop has gone into thinking about that much technical detail about the mechanism of his idea. Like you i'll be waiting to see his thesis before I believe anything he says.
:D
Anyway...
Each crystal isn't necessarily a pixel on standard film. The nature of lenses, light and film means each pixel is really a blob of hundreds and thousands of cyrstals, the more the exposure then the larger the blobs. It'd be better to record blobs and save the few gigabytes per image.
So there is little point in mapping individual crystals,
scanners can already scan more detail than most film actually has. So just a scan at high resolution and storage in a coventional lossless format should do just fine for what your suggesting, individual crystals is going a bit far and your probably wrong that it would take less size than just a regular scan. Basicly a map of millions of dots would be very slow to render on a PC screen for example and you'd need a complex algorithm to convert it to any other usefull digital format. |
|
| |
Digital cameras are getting better and cheaper all the time, eventually nobody will use film because there will be no point. In 2-4 years they will be cheap as dirt with crazy megapixels. |
|
| |
The sane megapixel models will be a little more expensive. |
|
| |
Let me start my saying I really don't understand photography all that well. |
|
| |
This idea is for archiving old photographs, like in musuems. I understand it will be expensive and time-consuming, which is why I aim it for musuem/archive type work. Typically they place preservation above all other concerns. |
|
| |
And I guess I'm not talking about film (or negatives) but about the actual exposed image, the photograph. |
|
| |
You're right, it is cheaper and easier to make high resolution scans, or to just take digital photos in the first place. |
|
| |
But what we're talking about is making sure photos can last into the future. |
|
| |
My understanding is basically that a high resolution scan is a full color bitmap. At a high enough resolution, what you're looking at is a really detailed image of a film grain (whatever crystal molecule that is). |
|
| |
So, I thought, instead of bitmapping the entire grain, just record what type it is, and where it lies. That would reduce the amount of information. If the atoms are rougly symmetrical, then all you need to do is record the centerpoint. |
|
| |
It may not be 100% accurate, but I guess it wouldn't be more off than another print made from the same negatives. |
|
| |
And in fact, if you *really wanted to* (I know it would be expensive and all that) you could re-create the print by placing the molecules appropirately on photographic paper and exposing it. |
|
| |
So if you have a really degraded old photo, you could map it, and re-create it in fresher chemicals. |
|
| |
I deleted that other link because it just seemed to be an ad for Tab. |
|
| |
So, this link shows silver halide crystals. They seem to look like 2D convex polygons no matter what angle you look at them, so you would need to store a little more information about thier geometry than just their center point. |
|
| |
So let's pretend that this link above is actually part of a huge high resolution scan of some famous photo that some rich museum wants to preserve. |
|
| |
All you need to record for these buggers is their center, a few points to describe the polygon, and their location. |
|
| |
I'm thinking that if you did another print of this same photo from the negative, you wouldn't get the exact same crystal layout. So, if you just record the centerpoint of each polygon that the crystal makes, it's kind of the same as having a digital negative. Then you just place the crystals appropriately when you want to 'print' it. |
|
| |
The grains are much smaller than you think. An 8 inch by 10 inch photo would just begin to resolve the larger crystals at a quarter million megapixels. Converting that from a raster framework to vectored graphics would be simply insane in terms of computation time and storage space. |
|
| |
// If need be, you could even recreate the photo by planting cyrstals according to the map! // Oh, yeah. A few thousand million crystals, each one custom grown for color and shape, and each placed by hand, I'm sure. Job security, I guess. |
|
| |
Fundamentally, the PURPOSE of film is to record an image. Because film grains have finite size, film does not do an absolutely perfect job of recording an image. Nonetheless, while film is not ideal, it is intended to record a continuous-tone image rather than one composed of discrete dots. |
|
| |
Except in very rare circumstances, the tones represented by the combination of exposed and unexposed grains are fare more relevant than the actual combinations of grains themselves. Because the arrangement of grains is essentially random, an effort to record film by recording the locations of the individual grains would require storing very large amounts of essentially uncompressable data. Even if it were feasible, I'm not sure I see much point. |
|
| |