Experiments with RGB-colored screens... a la Dufaycolor & Autochrome

holmburgers · Jul 30, 2010

erikg, I think that should work, but personally never having seen an autochrome plate in person I can't say for certain. Indeed, there's a pretty sweet lot on eBay right now, but I'm not giving a link.... I want them!

But I'm sure it could be done, pending an appropriate plate holder. Please report back if you do it.

holmburgers · Aug 9, 2010

Ok, so I got back my latest screen-plates. These I made by photographing a flat-screen CPU monitor. Both good news and bad.

Since my earlier tests turned out blue, I made a custom color that was a light yellow. What I got back were distinctly green slides. At first I couldn't wrap my head around it; I thought maybe it was a processing error, I thought maybe the film was too old, maybe the heat encountered in mail-order shifted the color.... etc.

HOWEVER, the lightbulb over my head lit up and I realized that this color shift is due to the fluorescent illumination used in flat-screen monitors. Duh!

The good news is that under the loupe I can see definition to the RGB elements, however I can't make out the colors. But the lens can render this fine detail and the resolution promises to be excellent with CPU monitors.

Meanwhile I've taken a picture with my first bluish screen-plate under tungsten light (to hopefully balance it out). That film will be going to dr5 soon.

So any suggestions on dialing in this color? I've decided that I can't continue using/wasting 4x5" film until I have the right color balance, so I think I'll be running a roll of 35mm with many different source colors to see what comes out the most neutral on the slides. But since I'm dealing with fluorescent illumination, aren't I kind of shooting in the dark? If I add magenta to my source, can I expect that to eliminate the green, or is it going to be more complicated than that?

holmburgers · Aug 10, 2010

http://www.kodak.com/global/en/professional/support/techPubs/e113/e113.pdf

Kodak talks about using EPP with fluorescent lighting. There's a table in the tech pup attached here. It gives different types of fluorescents and the appropriate CC filter plus exposure comps.

They say "When you don't know the type of fluorescent lamp, try a 30M filter and increase exposure by 2/3 stop; color rendition will probably be less than optimum."

Well, at least they're honest.

This gives me a starting point at the very least

Photo Engineer · Aug 10, 2010

You might try a Fuji film with the 4th layer which is adjusted for fluorescent lighting.

PE

holmburgers · Aug 10, 2010

Ahhh, actually this occured to me while reading the thread that's going on just now. Mind you, I need a slide film; does Fuji use the 4th layer in their slide films? I didn't think that they did, but IDK.

Photo Engineer · Aug 10, 2010

IDK. Look at their web site and the data sheets there.

PE

Photo Engineer · Aug 10, 2010

Go here: http://www.fujifilm.us/shared/bin/AF3-150E_Sensia100_Data_Sheet_2003.pdf page 5. Even though the sensitivity does not show 4th layer technology, they claim that it is present in Fujichrome Sensia. I did not look any others up.

"Not only has the spectral sensitivity of the three lightsensitive
layers (R, G, B) been optimized, 4th Color
Layer (Green-Color-Correction Layer) has been incorporated
that performs the negative spectral sensitivity
function of the human eye. Additionally, in order to enable
more appropriate color reproduction with a higher
level of fidelity, Sensia 100 has incorporated MCCL
Technology, as an extension of 4th Color Layer Technology
. This new technology has provided Red-Color-
Correction Layer (5th Color Layer) to enable more faithful
color reproduction and excellent skin tone reproduction."

PE

holmburgers · Aug 11, 2010

Thanks PE I'll definitely look into the other films; it appears that Sensia doesn't come in 4x5".

Ultimately though, I think I'll just try to dial it in with a test roll on 35mm.

Now, I'm not up to snuff on my color theory, and especially not today :eek:..... but even with a spiky/discontinuous source shouldn't it be possible to balance red, green & blue to any desired proportion? Or do the lacking parts of the spectral sensitivity make this impossible?

Any suggestions on a testing regime? I've got 36 frames to work with, and I'd like to go about it systematically so that I can be assured to get a least 1 hit, sunk your battleship

Also, this just occured to me. Do I need to be working with CC filters here? Because no matter what color I make on the screen, the RGB elements will be rendered identically unless there's filtration somewhere. I wonder if changing the color balance on the monitor actually changes the hue of the RGB elements or just the proportion... SO MANY QUESTIONS! The Cathode-Ray-Tube seems so simle at this point...

Q.G. · Aug 11, 2010

holmburgers said:
Also, this just occured to me. Do I need to be working with CC filters here? Because no matter what color I make on the screen, the RGB elements will be rendered identically unless there's filtration somewhere. I wonder if changing the color balance on the monitor actually changes the hue of the RGB elements or just the proportion... SO MANY QUESTIONS! The Cathode-Ray-Tube seems so simle at this point...

The RGB elements on the screen can only get brighter or dimmer. So no change in hue, only in proportion.
Same as on a CRT, by the way.

holmburgers · Aug 11, 2010

ok, honestly that's what I thought, but i appreciate the clarification

holmburgers · Aug 25, 2010

http://www.minsocam.org/ammin/AM23/AM23_20.pdf

Cool link with some rare (read: I've never seen them) Finlay plate photographs.

As for an update, the 1st attempt is at or on its way to dr5 for reversal processing. Once I get it back I'll be sure to report back... I'm pretty anxious to be honest!

Also, I'm going to give the CPU monitor screen-plate to a friend with access to a microscope because at this point I can't see the individual RGB elements with a loupe. I'm curious if it even resolved the detail and if there's any hope for getting true color with the fluorescent backlighting. I may resort to buying a flat-screen CRT CPU monitor if this proves to be too troublesome. Damn fluorescents....

michaelbsc · Aug 30, 2010

holmburgers said:
Does anyone have a process lens optimized for flat field & close focus? My 180mm Symmar (old school, convertible type) might not be the most ideal lens for these kind of "tolerances"

Do you still need this? I have some from process cameras that I'll lend you one (if you promise to make a few matrices for ME too). I'd want them back eventually, but they're in my "pie in the sky" project box doing nothing right now. You'll be on your own about how to actually get them mounted to your camera. They're still mounted on the OEM hardware, and I'm not even sure what camera they're from.

In fact, if there were a "good working set" produced I don't see why they couldn't be easy to duplicate by contact.

PM me if you want to talk about it.

michaelbsc · Aug 31, 2010

Making the RGB color screen with CYMK inks?

OK gang, Chris and I have been kicking around how one might go about actually making the RGB mask other than photographing a computer screen.

One idea has come up, and neither of us knows enough about color to know if this will work properly or not. So all you color guru's jump right in and set us straight.

I commented that at my office is a 34 in plotter, which is basically what Kinko's uses as a large format printer. Leaving a margin on the sides, it seems one could print a matrix of red, blue, and green dots that is 32x40 inches at a pretty high resolution. The plotter at my office is old, but the ones at Kinko's claim OMG resolution. (Never mind that the only cartridge that ever gets used in my office is black for making electrical drawings. Who knows if the others even work, but one at a commercial outfit should be in good shape.)

This is a lot of color dots, and if you could get it onto a piece of film it ought to make a nice mask.

But the question arose whether the fact that it is CYMK reflected instead of RGB transmissive would matter.

My gut feeling is that if the colors look right to my eye, then they'll record correctly on the film. Does anyone know if this is correct or not? How does this work?

And is this idea of taking a picture of a large print consisting of color dots reasonable?

Any suggestions?

holmburgers · Aug 31, 2010

We realize that it has been suggested before, to print the screen-plate on transparency film using an ink-jet, but indeed we couldn't reason how the CMYK would pan out. This time, the difference is photographing the paper with an RGB-grid printed on it.

I didn't get enough sleep last night, so perhaps another day it would make more sense...

*some time elapsed* Actually... thinking about a CMY (let's forget about K for the moment) screen-plate, perhaps it would just reduce efficiency. That is, the cyan screen element would let in both blue & yellow, magenta - both blue & red and so forth. But again, can't draw any conclusions....

One other thing with printing on paper is, unless you printed K (key/black) in all the blank spaces, you'd be left w/ transparent places in the screen-plate which could be bad. That was one of my initial attractions to the TV screen.... built-in lampblack.

Photo Engineer · Aug 31, 2010

It can be done either way with the right combination of dot density and positioning. It would really be best by transmission. I have seen it done and it replicates Dufay color, as this is what they used, but printed by a different method.

PE

Photo Engineer · Aug 31, 2010

You would print a C+M, a C+Y, and a M+Y dot set to mimic the R/G/B of the additive system. Interstices would be filled with black "Y" dots to increase density.

PE

holmburgers · Aug 31, 2010

:wizard:

That makes sense

michaelbsc · Aug 31, 2010

Photo Engineer said:
You would print a C+M, a C+Y, and a M+Y dot set to mimic the R/G/B of the additive system. Interstices would be filled with black "Y" dots to increase density.

PE

Being a color neophyte myself, can you point me to an explanation of how this works? I'll do some Google on my own, but if there is succinct explanation that I can go to instead of hunt down that would help.

Michael

Photo Engineer · Aug 31, 2010

Try this for a visual representation.

http://www.google.com/images?q=rgb+...tle&resnum=4&ved=0CD0QsAQwAw&biw=1184&bih=692

This might be better:

http://en.wikipedia.org/wiki/CMYK_color_model

michaelbsc · Aug 31, 2010

Photo Engineer said:
Try this for a visual representation.

http://www.google.com/images?q=rgb+...tle&resnum=4&ved=0CD0QsAQwAw&biw=1184&bih=692

This might be better:

http://en.wikipedia.org/wiki/CMYK_color_model

Thanks.

holmburgers · Aug 31, 2010

How would one produce a file like this suitable for printing? I wonder if there are some patterns floating around out there...

Photo Engineer · Aug 31, 2010

You could use the color, point and line functions in VBASIC and then the Printer Print function. This is how the "Dufay" example was apparently made. IDK for sure. But a simple print or plot routine would do the job if you used transparent film support for the image.

PE

David Grenet · Aug 31, 2010

It's easy enough if you have access to Matlab too. I have attached a script that I wrote for this purpose (although I never got around to actually using the result!). Just remove the .txt to use it with Matlab. It generates a regular RGB matrix and saves it to a tiff file. It should be adaptable for Scilab easily enough which is an open-source program similar to Matlab if you don't have access to a copy of Matlab.

happyjam64 · Sep 1, 2010

I made a screen pattern quite a while back. I haven't printed it yet, but i'll probably try this coming fall (i can print it at school). Anyway, what i did was just make a pattern with photoshop. Start with a 2x2 grid with R, G, and B, then just use the stamp tool to cover an 8x10 document at 300 dpi.

michaelbsc · Sep 1, 2010

happyjam64 said:
I made a screen pattern quite a while back. I haven't printed it yet, but i'll probably try this coming fall (i can print it at school). Anyway, what i did was just make a pattern with photoshop. Start with a 2x2 grid with R, G, and B, then just use the stamp tool to cover an 8x10 document at 300 dpi.

So, thinking about taking a picture of an 8x10 @ 300dpi on 4x5 film will give us a grid at 600 dpi.

Back to one of Chris's original questions, what size do you really want the grid to be? How small were the flour grains, and where do you hit the lpm resolving limit of the taking lens?

Experiments with RGB-colored screens... a la Dufaycolor & Autochrome

Attachments

Privacy & Transparency

Privacy & Transparency