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

Ray Rogers · Nov 3, 2010

<Perhaps this should go in a new thread- feel free to move it if so moved.>

RGB filters are I think, not graded like YMC filters are, for density... I am thinking of tricolor separation filters in particular, and wondering how different densities affect their use in pratice...

I am quite sleepy now but I think given a mixed blue and green light source, the blue can be removed by one of two methods... a green filter or a yellow ("minus blue") Is this correct?

Can the same result be achieved both methods... tie removal of the blue?

If all this is good so far, then here comes my real question:

How can I equate the blue removing power of a yellow filter with that of a green one?
I want to know eg. the equivalent in green to say a K12...

or better, to one of the standard tricolor separation filters....

Thanks,

Ray

Photo Engineer · Nov 3, 2010

Ray;

Good questions.

Yes, R/G/B filters do come in grades just like C/M/Y filters. Thus there are 10R, 20R, 30R and etc filters and so on. I have 6 complete sets of the two types of filters.

As for measuring the effectiveness of any given filter, it is done using a spectrophotometer. Kodak had published the complete spectra of all of these filters and it was also posted on their web site. IDK if it is still there. There are three qualities to look for in a spectrogram of a filter, namely lambda max, bandpass and density.

PE

Ray Rogers · Nov 3, 2010

PE,

Yes, I was thinking mostly about the tricolor filter sets for making separations... AFAIK their is only one density(?) offered per color per group, and so that got me thinking why... for printing we have as you pointed out a series, so why only one for the separation tricolors... then I tried to figure out what difference using a nonstandard density tricolor would make and that is where I lost consciousness.

OK, I probably have the books and data you mentioned for both Kodak and Fuji,
So, assuming I do have the filters lambda max, bandpass and density data,
How can I find "equilivent" filters (equilivent as mentioned in my previous post) between whatever the number is of the blue tricolor in either set (R=25 or 29=) Was it 47 ? I can't remenber off hand, and a y of cmy?

What would my goal be, to insure equality of blue stopping power between green and yellow filters?

One being green the other yellow it seems like i might not be looking for the same 3 values...

Photo Engineer · Nov 3, 2010

Tricolor separation filters vary mainly in bandpass or "cut".

The way to compare is to place the curves on top of each other at the same x and y scale sizes and compare those 3 features that I gave. Depending on whether you are looking at transmission or density, you may have to invert the curves for a real match.

In no case do the tricolor filters match the cc or cp filter sets though. The tricolor filter sets have much narrower bandpass and the comparison is difficult.

PE

Ray Rogers · Nov 4, 2010

OK.

That sounds logical.
Once I actually try to do it I might run into trouble, but that just means I'll get another chance to learn.

Thanks.

Ray Rogers · Nov 7, 2010

Are any tricolor seperation filters available in tri-acetate or other "soft" version, or are they available only in glass?

I can't find nor recall ever having seen any....

Photo Engineer · Nov 7, 2010

I have both Ray.

PE

Ray Rogers · Nov 7, 2010

Photo Engineer said:
I have both Ray.

PE

Can you ID the soft ones
(by cat # or other ID printed on the pkg.) ?

Photo Engineer · Nov 7, 2010

By no means. I have had the glass plate filters since about 1950, and the packing material is long gone. They are wrapped in foam in a 4x5 box. The film based filters came as 8x10 sheets with no ID at all. I have some small 2" square ones that came in envelopes but they would probably have no usable ID now either if I even have the envelopes, having been bought in the 50s as well.

Sorry.

PE

Ray Rogers · Nov 7, 2010

So separation tricolors are most likely no longer avaliable in flexible form?

Photo Engineer · Nov 7, 2010

I didn't say that. In fact, that is the easiest and most cost effective way to make them. Kodak made them using a coating machine of special design, but any coater could be used. Glass filters are harder to make.

PE

Mustafa Umut Sarac · Nov 7, 2010

Some companies do such coatings on glass for remote sensing or military. And schott have extensive filter glass catalog and they cut to your specifications. If you must have these , this is an option.

And if you contact with American Cinematographer magazine , they forward you to gel filter advertisers. I dont know if there remains who use them but I saw them 10 years ago.

Ray Rogers · Nov 7, 2010

Initially I was planning on using glass, but a soft film version would make the unit more compact and light weight.

Mustafa Umut Sarac · Nov 7, 2010

Gel filters tends to lose their colors in time. And they easy to strach. I think cinematographer filters are for skin colors or for a more romantic mood.
Yes all is good but depends lots of money.

And all depends on panchromatic emulsions. Its a factory business and lots of research and development.
You must create a emulsion and color grid for it. And result will be no different than holga. Thats why museums documents the process but never attemps to recreate it.

Umut

Ray Rogers · Nov 8, 2010

I found some Fuji filters,
there are 3 sets of 3 so there are 9 of interest...

They varry in bandwith and % transmission.
They also varry in transmission ratio between members of the same set.

One set is for additive printing on color paper
Two others are for "TV etc."

How does one choose the most suitable set?

?

holmburgers · Nov 8, 2010

Hey Ray,

The classic Wratten separation filters are...

25 red, 58 green & 47B blue - Convenient because all filters share a filter factor of 8 (3-stops)

29 red, 61 green & 47B blue - Purer color separations, larger color gamut (the 25/58 set has over 6% overlap while this set has less than 1% overlap)

Check eBay lots, or you can look for the individual fitlers.

Ray Rogers · Nov 8, 2010

Hey There!

Now that info was useful...
I had not noticed the first set all had the same filter factors!

i have several 25s and 58s in glass btw

However (and this is where I get sleepy...)

What difference does the overlap make?

When are overlaps of value?

Why do both sets exist?

When is the set that over laps better than the set w/no ovelap?

Is "tricolor" separation better
if done with narrow band pass
or
with wide (but w/precsion placed) bands?

I think what I am trying to ask is... is there a criticism of the tricolor method?

Is their a dissenting view?

I seem to recall there are limitations and some others have found 4 or 5 to give more robust results.

---
Ray

Photo Engineer · Nov 8, 2010

Tricolor works best if there is no overlap in transmission bands. Overlap leads to crosstalk between the images.

PE

holmburgers · Nov 8, 2010

Separation filters are a topic which I would love some clarification on myself. Reading some old books, I learned that separation filters should match the color sensation curve of that color... but that would imply some overlap.

I also read that they should be matched to your pigment.

What result would incredibly narrow-band achieve? Let's assume a theoretical infinitely-narrow separation band for each color.

The real question is... why are you discussing separation filters in a screen-plate thread?! Ok, just kidding, but what do you have in mind?

Photo Engineer · Nov 8, 2010

Well, IMHO the bandpass of a separation filter has a lot to do with the design of a good screen plate. Working with "confetti" of the filters, pasted onto film in a random pattern would actually make a fair screen plate.

As for an infinitely narrow filter, well it would probably require an infinite exposure to get enough light through! I have seen very very narrow cutting filters at just about every wavelength. We have even coated them experimentally to improve color separation in films and papers. The filters should be R/G/B and if the images are to be dyed then the dyes should be C/M/Y, or the opposite of the filters.

PE

holmburgers · Nov 8, 2010

Confetti.... not a bad idea!

So, infinite exposures aside, the narrower the better, is that what you're saying?

And by matching the pigments, I meant the obvious application... that your red sep. filter should completely darken the cyan pigment and make the magenta & yellow pigments appear equal in luminance when viewing the pigments thru the filter. And so on/so forth around the color wheel.

In A handbook of photography in colours by Bolas, Tallent and Senior, this method of testing pigments/filters was suggested. They had to be clever in 1900...

Mustafa Umut Sarac · Nov 8, 2010

Holmes ,

Really great idea .

Umut

Photo Engineer · Nov 8, 2010

The narrower the better within limits. In fact, I have advocated the WR 98,99 and 70 filters as they are narrower than the ones mentioned in the usual texts, but still allow enough light through to give reasonable exposures.

PE

Mustafa Umut Sarac · Nov 8, 2010

Wider the cut or overlapping , lower the color saturation.
Narrower filters are important for drum scanners also. Printing press inks have such a failure and cheaper the ink , worse the overlapping. For example , you have cyan ink , less the yellow inside , better the result

Ray Rogers · Nov 10, 2010

Has anyone seen a "white" created with only 3 wavelengths ?

Is it possible that color synthesis
done with only 3 selected, isolated representatives of the R, G, B tricolors
is better or somehow different, than when
there are a multiplicity of wavelengths that fall into the R, G, B groups?

Ray

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

Privacy & Transparency

Privacy & Transparency