Digital Printed Masks for Autochromes

Has anyone made an attempt at printing a bayer filter pattern onto a film base with an inkjet printer? Would an inkjet even produce fine enough dots to make an autochrome pattern? I'm interested in giving it a shot, but I don't really know the best way to get started. Has anyone on here actually tried making autochromes before? I know that there was a series of them made by Frederic Mocellin, but other than that, work on them seems pretty sparse.

Unless you're going to expose through the base, your Bayer pattern would need to be on the emulsion -- original Autochrome had the dyed starch granule filter layer on the emulsion side. Further, you'll find a major problem in that ink jet inks are generally water soluble -- and they need to remain in place through processing, in order for each spot of emulsion to be viewed or projected through the same filter dot it was exposed through.

Also, a Bayer isn't the best pattern -- it'll become visible at much smaller pitch than a random dot pattern. It's used in digital sensors because the fabrication process for conventional CCD and CMOS sensor chips produces square or rectangular elements which don't lend themselves well to non-regular or non-rectangular color filter array layouts.

Donald Qualls said:

Further, you'll find a major problem in that ink jet inks are generally water soluble --.

Click to expand...

Actually, they are quite waterfast once dry. I’m speaking of the Epson pigment inks. I know first hand that you can print them and then lay platinum/palladium over the top of them - a water based process - and the pigment inks are quite durable.

If they're pigment, they probably aren't very translucent. And they still need to be on the emulsion side.

Donald Qualls said:

If they're pigment, they probably aren't very translucent. And they still need to be on the emulsion side.

Click to expand...

That seems like a known problem with autochromes, you're expecting a major speed loss as starch isn't particularly translucent either.
You're right about the gelatin side, I was reading about lipmann plates and had some wires crossed in my head.

I was lying in the bath the other day and had just that thought. Why not inkjet the filter matrix? I'd love to try it. Great minds think alike - but fools seldom differ!

If you could get it to work there would be many advantages of inkjet printing the matrix. e.g. you could finely adjust your matrix to match the emulsion's spectral response. Not only could you change the proportion of RGBK "grains" but you could use other colours.

The main reason I feel it is totally doable is that this approach was more or less taken (sans inkjet) into the 21st Century in the form of Polachrome. I remember using this in the 1980s to produce rush job slide presentations for people. I'm not sure how the colour matrix was laid down though.

https://en.wikipedia.org/wiki/Polavision

There was Dufaycolor in the early 20th Century that used a similar technique with a regular colour matrix on film.

But the number of things that need to come together to create any autochrome-like additive process are enormous. The matrix is only one part of the puzzle and the other parts may be harder.

1. You need a near panchromatic (ideally somewhat isochromatic) emulsion or you won't capture the colours or roughly balance the colours.
2. You need the emulsion to be relatively fast because the light has to pass through the filter before it hits the emulsion and the filter can pass a theoretical maximum of 1/3 of the light (you're splitting it into 3) but practically much less because you need black between the grains. Also reversal processing typically makes it even slower.
3. You need to be able to coat in total darkness (the emulsion is pan) or sensitise the plate after coating by soaking it in a red dye - but that dye can't be allowed to effect the colour matrix.
4. You need to develop and reverse the image without damaging the matrix. The reversal processes I know of use strong oxidising agents (H2O2 or toxic chromiums) to bleach out the silver this may be tricky. (You could just make negs and scan them though but that wouldn't feel like autochromes more like 19th Century C41!)

I didn't jump out the bath and start work on autochromes because I don't have the skills to make a descent orthochromatic emulsion yet let alone a panchromatic one of any speed. If I could make one I might try my luck at Lippmann colour plates first as they have fewer complexities. Autochrome are a distant dream.

Go for it though. I'd love to hear how you get on.

RogerHyam said:

I was lying in the bath the other day and had just that thought. Why not inkjet the filter matrix? I'd love to try it. Great minds think alike - but fools seldom differ!

If you could get it to work there would be many advantages of inkjet printing the matrix. e.g. you could finely adjust your matrix to match the emulsion's spectral response. Not only could you change the proportion of RGBK "grains" but you could use other colours.

The main reason I feel it is totally doable is that this approach was more or less taken (sans inkjet) into the 21st Century in the form of Polachrome. I remember using this in the 1980s to produce rush job slide presentations for people. I'm not sure how the colour matrix was laid down though.

https://en.wikipedia.org/wiki/Polavision

There was Dufaycolor in the early 20th Century that used a similar technique with a regular colour matrix on film.

But the number of things that need to come together to create any autochrome-like additive process are enormous. The matrix is only one part of the puzzle and the other parts may be harder.

1. You need a near panchromatic (ideally somewhat isochromatic) emulsion or you won't capture the colours or roughly balance the colours.
2. You need the emulsion to be relatively fast because the light has to pass through the filter before it hits the emulsion and the filter can pass a theoretical maximum of 1/3 of the light (you're splitting it into 3) but practically much less because you need black between the grains. Also reversal processing typically makes it even slower.
3. You need to be able to coat in total darkness (the emulsion is pan) or sensitise the plate after coating by soaking it in a red dye - but that dye can't be allowed to effect the colour matrix.
4. You need to develop and reverse the image without damaging the matrix. The reversal processes I know of use strong oxidising agents (H2O2 or toxic chromiums) to bleach out the silver this may be tricky. (You could just make negs and scan them though but that wouldn't feel like autochromes more like 19th Century C41!)

I didn't jump out the bath and start work on autochromes because I don't have the skills to make a descent orthochromatic emulsion yet let alone a panchromatic one of any speed. If I could make one I might try my luck at Lippmann colour plates first as they have fewer complexities. Autochrome are a distant dream.

Go for it though. I'd love to hear how you get on.

Click to expand...

The idea I've been tossing around in my head is to modify an inkjet printer with a feed from a standard bulk loaded 35mm reel, then process it as a negative, but thinking about it more and looking at the polachrome pattern, it might make sense to just make 3 pdms stamps and push whatever ink you want through

First: Polachrome had the color applied in fine stripes, much like the color phosphors on a Sony Trinitron television CRT. The stripes were very fine, so not very visible in projection (unless the projector had higher than usual optical quality), but they were obvious under a microscope, and Polaroid weren't shy about saying how they produced a color slide that could be processed in minutes, at room temperature.

Second: you don't really need black if you can make your color filter spots touch -- that is, if there's no space between the filter spots, there's no need for a black matrix. The black bitumen matrix in original autochrome was there because the starch granules were irregularly shaped and would leave gaps passing unfiltered light -- and unfiltered light would wash out the color saturation.

There was an experimenter ten-fifteen years ago who did some work toward autochrome and posted either here (APUG, then) or on Photo.net about it. Seems to me he formed a filter gelatin layer separately and then applied that to the emulsion. This has the advantage that you can verify the filter (for color distribution and consistency) before you expose and process the film. It might even be possible to use registration pins to expose through a separate filter layer, process without it (producing a fairly normal-looking B&W transparency or negative) and then return to register for scanning, projection, printing, or rephotography to a positive if a negative process was used originally. You'll have trouble printing to RA-4, since you won't have the orange mask, but if you process a positive, that's very much an optional step.

Worth noting that when Autochrome originally entered the market, even panchromatic emulsions were uncommon -- not unheard of, there's a fairly large body of tricolor photography from as far back as the mid-1890s -- but the level of panchromic response apparently doesn't have to be exceptionally high; as with tricolor negatives, the low response to red at worst requires adjusting the red dye (Autochrome didn't have a true red, presumably for this reason -- the "red" was actually pretty orange, but if you have a film with reduced red response, like "orthopanchromatic" stock, you could also include a higher proportion of red in your filter).

Of course, the main issue with Lippman process is that you have to handle mercury (I suppose you could use gallium if the camera were very mildly heated -- more expensive than mercury, but less of a hazmat panic if there's a spill). You also need a material with no antihalation, since you must expose through the base in order to have the emulsion in direct contact with the reflecting liquid. I suspect J. Lane dry plates would be good candidates.

RogerHyam said:

Dufaycolor

Click to expand...

The Dufaycolour screen ruling was in the order of about 600lpi from what I recall - which is just about the finest line screen used even today for various ink on paper processes (offset, rotogravure etc) - even then it's extremely challenging to run halftones in. I think most inkjets would struggle to lay down lines that fine.

Donald Qualls said:

Of course, the main issue with Lippman process is that you have to handle mercury (I suppose you could use gallium if the camera were very mildly heated -- more expensive than mercury, but less of a hazmat panic if there's a spill).

Click to expand...

There are "modern" Lippmann methods that use an air gap as the reflector.

I don't think inkjet printing onto a commercial film would work.

Option 1 is to print on the emulsion. Gelatine is "designed" (possibly by God) to expand and absorb chemicals when wet often by as much as ten fold. The chances that the inks won't bleed during processing are remote. This could be tested quickly just by drawing on the emulsion with intended ink/pigment and processing the film. Bleed of a few microns and it fails especially in 35mm.

Option 2 is to coat the back of the film and expose through the back. If you can get the ink to stick and not wash off this is an option but... commercial black and white film has multiple layers including anti-halation layers designed to absorb light. I know people do red scale C41 (expose through the back of regular negative film) I've never heard of it done with B&W film. This could be tested just by loading a candidate film in a camera backwards and seeing what effective ISO and colour you got. I think the image would be pretty blurred. i.e. it would be blurred after it passes through the colour matrix thus messing up the

Lachlan is right in that original Autochrome starch grains were very small compared with what could be done on an inkjet printer but if you worked large format and just wanted an autochrome/pointillist effect you weren't going to enlarge then you could still get images.

The idea of having a separate screen that is somehow pin registered seems a non starter. Registration would need to be perfect across the whole frame and deformation of materials would make this impossible. We're talking a screen in the region of 1/1000th of an inch pitch at least. Screen and emulsion need to be bound to the same substrate.

Sorry but it looks like everyone is being a downer on this. I love the idea though.

A commercial film manufacture could make an additive colour film though. Adox or someone could do a reversal film with a matrix in the film support but whether it would be commercially viable or desirable is another thing.

RogerHyam said:

I don't think inkjet printing onto a commercial film would work.

Option 1 is to print on the emulsion. Gelatine is "designed" (possibly by God) to expand and absorb chemicals when wet often by as much as ten fold. The chances that the inks won't bleed during processing are remote. This could be tested quickly just by drawing on the emulsion with intended ink/pigment and processing the film. Bleed of a few microns and it fails especially in 35mm.

Click to expand...

That's a good point, I didn't think of that, the gelatin swelling would almost certainly cause bleeding. Exposing through the back of a black and white image yields an odd, out of focus and hazy image. Likewise coating a starch or similar layer on top of the gelatin would probably effect development by messing with the water imbibing into the gelatin. Likewise the starch would probably wash out in development if you were to try to coat starch directly on the gelatin itself. I suppose you might be able to lock it in place by going:
---hard gelatin ----
-- starch dyes-----
--- emulsion ------
--- base-----------

But I'm almost positive you're going to have problems with getting the developer into the gelatin

What could be done with commercial film potentially would be to coat some clear acetate leader with colored starch, then a receiver layer from a silver diffusion system and then process it with a monobath and rollers, that would give you a one step positive, but then you have all of the problems that people have been trying to work out with building diffusion transfer receivers. You'd also still have a registration problem, but I don't know exactly how close the color layer needs to be to the same position as it was when the picture was taken. My naïve guess is that it would have to be at the very least approximately +-10um in the x and y, as iirc the starch granules would be approximately 20um, and you'd likely want each starch granule's center position (approximating they are round) to be within 1 radius away from their starting position when the photo was taken.

Unfortunately as it stands right now, I don't have access to a large format camera, so I'm limited to at most a 1/4 plate camera, which might work if I hack together some sort of plate holder.

@RogerHyam the screen ruling for the pattern on Dufaycolor is more likely feasible at industrial scale - via rotogravure. I would not be surprised if that was how the resist on the original was applied, followed by the dye bath for the relevant colour. The limitation on inkjet for replicating the effect is the size of ink droplet and the dot gain of the substrate used. I'd use something like the Grafix wet media film (which can accept aqueous coating on both sides) and work form there - though you may have to be careful about ink loading.

RogerHyam said:

Option 2 is to coat the back of the film and expose through the back. If you can get the ink to stick and not wash off this is an option but... commercial black and white film has multiple layers including anti-halation layers designed to absorb light. I know people do red scale C41 (expose through the back of regular negative film) I've never heard of it done with B&W film. This could be tested just by loading a candidate film in a camera backwards and seeing what effective ISO and colour you got. I think the image would be pretty blurred. i.e. it would be blurred after it passes through the colour matrix thus messing up the [image quality.]

Click to expand...

I have (accidentally) exposed B&W sheet film through the base (when you're newish at large format, you make all kinds of fairly predictable mistakes). Apparently, the antihalation features in sheet film add up to about five stops of neutral density (early oughties vintage 320TXP); I was able to save the images on the base-exposed sheets by developing to completion in the hottest multi-agent developer I could throw together on short notice:



Zeiss Ideal 9x12 cm, 13.5 cm f/4.5 Tessar, exposed through base, developed in "Super Soup"

With the lens stopped down for a sunlit exposure and EI 320, defocus wasn't noticeable (your mileage may vary).

laingsoft said:

That's a good point, I didn't think of that, the gelatin swelling would almost certainly cause bleeding.

Click to expand...

I don't think there would be much swelling -- the amount of water delivered to the film by picoliter ink drops is, well, pretty small. Also, if your printed matrix elements are several printer pixels in size, any bleed would simply result in mixed color over a short distance, which the Autochrome process will correct for (by viewing/projecting through the same mixed dyes you expose through). Further, pigment ink will show little if any color bleed, because the gelatin won't carry the pigment particles the way it does the liquid carrier. The element size needing to be 3-6 dots across at printer resolution of around 600 ppi (~23 pix/mm, or about 0.4 mm per matrix element) would limit the process to medium format or larger film, however.