Anders Conbere on Nostr: nprofile1q…fcvt4 This will be too many words but I'll do my best! I've been ...
nprofile1qy2hwumn8ghj7un9d3shjtnddaehgu3wwp6kyqpqqdmvd6xtmfm6w0f7rnqkkq2477q8uxpq345p55vu8emf8zmqsytspfcvt4 (nprofile…cvt4) This will be too many words but I'll do my best!
I've been printing with UV processes (Cyanotype, Kalotype, etc.). Because of the absorbancy of UV to glass these methods are contact printed rather than printed with a photographic enlarger.
There are two issues with contact printing: 1. Large prints require large negatives and large negatives are expensive to create. 2. Every process has it's own unique tonal curve and a negative that prints well in one process is unlikely to print well in another.
To get around these limitations modern printers now print digital negatives onto transparency allowing cheap reproduction of digital files at all sorts of sizes as well as applying tone correction curves in the digital process.
The question then becomes: What is the tone correction curve one should apply? There are too many uncontrolled variables in our home labs for there to be standardized curves that work for everyone. Whether it's small differences in the sourced chemicals, lenses, water, etc. So in order to determine our correction curve we test the process by printing a calibrated sequence of grey scale steps onto transparency, printing the transparency in our process, scanning the resulting analog print and observing the difference from the digital linear tone progression.
You can do this by hand with some small effort, and there is also some software to help automate the analysis. Chartthrob is a popular choice but it only exists as a photoshop plugin. I have been working on a tool to do this that isn't locked to any platform.
I've gotten to a point where I can generate my calibrated step wedge (read sequence of grey squares from white to black in a linear tonal progression on the digital side). And I'm able to ingest a digital scan, find those squares and read back out their tones.
Now what I want to be able to do is pass in a digital image and /apply/ the correction curve to it!
The result of this process is two collections of 101 points. The first is my 1:1 tonal scale progression from (0,0) to (65535, 65535). The second is the observed values from the print (normalized to the same scale).
By observing what input densities resulted what output densities we can reconstruct a new set of input densities that when printed onto transparency should result in a linear tonal curve. The result is a sequence of 101 2D points bounded by (0,0) and (65535, 65535) that represent a mapping from the original digital tone X to the tone it should be to print correctly Y.
The job now is to find a function such that f(X) = Y for all of my points above and where X is extended to the full range of u16 so that I can apply it to a full digital file and get the right outputs for printing.
I've been printing with UV processes (Cyanotype, Kalotype, etc.). Because of the absorbancy of UV to glass these methods are contact printed rather than printed with a photographic enlarger.
There are two issues with contact printing: 1. Large prints require large negatives and large negatives are expensive to create. 2. Every process has it's own unique tonal curve and a negative that prints well in one process is unlikely to print well in another.
To get around these limitations modern printers now print digital negatives onto transparency allowing cheap reproduction of digital files at all sorts of sizes as well as applying tone correction curves in the digital process.
The question then becomes: What is the tone correction curve one should apply? There are too many uncontrolled variables in our home labs for there to be standardized curves that work for everyone. Whether it's small differences in the sourced chemicals, lenses, water, etc. So in order to determine our correction curve we test the process by printing a calibrated sequence of grey scale steps onto transparency, printing the transparency in our process, scanning the resulting analog print and observing the difference from the digital linear tone progression.
You can do this by hand with some small effort, and there is also some software to help automate the analysis. Chartthrob is a popular choice but it only exists as a photoshop plugin. I have been working on a tool to do this that isn't locked to any platform.
I've gotten to a point where I can generate my calibrated step wedge (read sequence of grey squares from white to black in a linear tonal progression on the digital side). And I'm able to ingest a digital scan, find those squares and read back out their tones.
Now what I want to be able to do is pass in a digital image and /apply/ the correction curve to it!
The result of this process is two collections of 101 points. The first is my 1:1 tonal scale progression from (0,0) to (65535, 65535). The second is the observed values from the print (normalized to the same scale).
By observing what input densities resulted what output densities we can reconstruct a new set of input densities that when printed onto transparency should result in a linear tonal curve. The result is a sequence of 101 2D points bounded by (0,0) and (65535, 65535) that represent a mapping from the original digital tone X to the tone it should be to print correctly Y.
The job now is to find a function such that f(X) = Y for all of my points above and where X is extended to the full range of u16 so that I can apply it to a full digital file and get the right outputs for printing.