Autochrome

[el french]

Perhaps a small random orbit sander attached to the build plate would distribute the starch grains evenly. A paint spray gun is capable of putting down a precise thickness of paint so it may work for the emulsion.

[Dustin McAmera]

What about a half-cheat: you could step past the starch part of the project (at least for now). It ought to be possible to prepare an image consisting of random coloured dots in three colours, and print that onto a flexible plastic backing (I'm thinking of an OHP slide). Maybe you have to varnish that to protect the colour-screen image; then coat it with the gelatin emulsion (again, at least at first, I'd buy emulsion in a bottle).

This way, you would have sheet film, not plates, and an emulsion that ought to work. You can concentrate on the practicalities of exposure and reversal development, and developing a colour set that gives a good result. You can always add back the extra difficulties of glass plates and starch later.

[Steven Tribe]

There were patented processes which used micro dot random colours (Dufay?) on plastic after lumiere.

My system of coating was clear spray lacquer with the mixed grains applied instanteously and excess removed straightaway (gravity and brushing)

[Drew Wiley]

There were dozens of different color processes. Autochrome was one of the first, but not the very first by any means. Otherwise, wood dyes are one of the worst
choices possible because they are mostly fugitive analine. If you want to see a gritty slightly off-color but gorgeous print process, look at color Fresson. It's a well
kept trade secret, true pigment screen process in the direct-carbon category (versus carbon transfer). As primitive as Autochrome was, it was still an industrial
process that required not only its on R&D phase, but skilled workmen and standardized industrial ingredients and equipment. Another once-popular process that is
difficult to reproduce on a do-it-yourself basis is color Carbro. Dye transfer is more realistic.