Possibility one. Micro mixing.

If you have bought paint anytime in the recent past, you have seen your color mixed right in front of you. Micro mixing is doing that on a drop by drop basis.

Lets say the system is going to be based on airbrush fine tips. This gives you about a 1/64" or 1/32" on center pixel resolution. 60 million pixels on a 10' x 10' wall possible.

How could you do this?
My first thought is that a finely ribbed plastic ribbon is run through the pigment. That ribbon is pulled through the loading chamber far enough each cycle to load the right portion to the base. The base is then forced through the channels in the ribbon picking up the pigment. That pigment ribbon is then pulled through enough for the next pass and the process repeats. If the pigment is not needed the ribbon is not pulled as the base passing through the channels has cleared out the previous load. In a CMYK system there would be four ribbons and you would be mixing four loads at any one time. It is equally possible that each actual pixel on the wall would require more then one fire on the airbrush that would mean that the pigment delivery could be adjusted so a portion of the final total is delivered for each fire.

My second thought is to hack up some inkjet cartridges replacing the ink with pigment and letting that fire into the mixing chamber. When the air atomizes the base it will mix in the pigment.

Why don't I just hack the print cartridges print directly on the wall? Because paint is not ink. Printing directly onto the wall will look just like printing directly onto the wall. There is a major aesthetic difference and so a major level of customer satisfaction which leads to more work and better prices.

Anybody feel free to chime in.
One of the considerations when I am thinking about this project is can I do this. Creating the ribbon and loading chamber is just on the edge of what I think I could do. Others of you will have different thoughts and capabilities.

Hmmm. Are you alluding to texture, thickness, and the 3D quality of man-made paintings? Simulated paint or charcoal strokes, such as that done with Photoshop filters, but in a single fine layer, is not what you want?

Personally, I kind of like the pseudo-texture canvas paintings that are done nowadays, with a rough base layer (usually white) laid down quickly with strokes that roughly approximate those in the original painting (of course not needing the precision of colored strokes) followed by an automated printing process. At typical viewing distances there is the very real perception of an authentic work. I'm sure you're very familiar with this, but do you favor or rule out this kind of thing?

Just adding to the thought experiment here; I realize this isn't directly addressing micro-mixing. 'Tis getting late, gotta sleep on that one. I might have something to say about it tomorrow.

PS: Oh, I was thinking that with an embossing filter, one could take any image and create a pseudo (and synthetic) texture and lay it down with automation, in several passes using the base coat. An obvious idea. I haven't checked but I'm sure this is already a common and widely-practiced technique for art reproduction.

Not really I was alluding to the depth and transparency differences of the respective media.

All of the stuff you mention is cool and could be used later in more complex and expensive designs. Indeed I'm creating a tool that will simply add an enhanced path to expression. If as I currently expect the system will be based on airbrushing there is nothing that prevents the system from mimicking that store of gestures. You could create an interface that allows the artist to airbrush a model and translate that to the wall (a digital airbrush pantograph).

Awesome concept! The painting process could be recorded digitally and then "replayed" on the wall. Would be much, much more interesting to watch than a raster type scanning pattern. Again, I s'pose this has been done before. I'm sure though it's quite a technical challenge when you add the z dimension and scale differences. If the original artwork was done on a smaller scale, the way an airbrush pattern is sprayed at larger distances and different velocities on the larger canvas would likely introduce noticeable differences. Worth a try though!

Suggestion: Build a smallish model of the robotic painter to test the technology; say, 4x4 feet. Try recording a painting done at 1x1 and 2x2 feet and see if the airbrushing scales up satisfactorily. Recording the artist's movements could be done fairly easily nowadays, with motion capture technology that's used to create CG character movements for movies. There are several methods to do this.

A search for "robotic painter" turns up some interesting results!

Nozzles!

The job today went really well so I had the afternoon off.

The possibility of micro mixing has been on my mind so I decided to attempt a test.

I molded a plug of plastic around a small machine screw and let it harden. I then took a strand from some #14 wire I had an melted a small hole at the front to the face of the screw. Backing out the screw I had a small nozzle. Needless to say the aerosol pattern it created left a lot to be desired. Using a small brad nail to open out the tip of the hole helped that a bit. (more research needed).

Loading the screw chamber with paint and then reinserting the screw allowed me a surprising amount of control over the size of the drop I could create. And it cleaned up really well.

Next I drilled out a little pocket on the top and heating that brad melted a hole into the front of the threaded cavity. Now when the nozzle is full the brad is inserted sealing this hole. Turning the screw still creates the drop. But if I fill this pocket with paint remove the brad and backup the screw I can refill the chamber. The original hole at the front doesn't appear to be large enough in relation to the hole created by the brad to interject much if any air.

Next I used a pigment instead of paint. There was no noticeable difference in behavior even though the pigment was a lot less viscous.

So I think I have the beginnings of a system here. Small diameter finely threaded screws turned in small measured increments providing pigment to the air mix atomization chamber.
By closing the supply feed I can close the system eliminate the siphon effect of the air flow passing the pigment/paint supply. The atomization should also mix the pigment and the paint quite well and even incomplete mixing will be compensated by the random aerosol pattern.

However robustness and speed are concerns as I mentioned before we are talking of tens of millions actions per wall.

Interesting! Can you get any pictures of your nozzle prototype? Also, referring to my previous post, do you know a mural artist you'd get to do a mural to be replicated? Or do you do murals yourself?

Pictures of a blob of plastic with a screw and a nail sticking out?
Nah.

I think I'll wait on pictures after I've built a mold.

i think you're looking at multiple systems here unless you're intending to do a true fresco.

the "prep" work of priming, is just a mass coating, which aside from the geekery would be faster applied via commercial spray or roll technologies, which are designed to do just that.

the most functional wall marking bots iv'e seen are 2 line suspended bots on wheels that rely on differs in the tension of their two leads to cover an entire wall. you mount a pulley on each of the two top corners, and by reeling in and out those you move the print head around. using multiple print heads for multiple colors with a high enough resolution, the applied dots will blend ala pointalist work into a uniform piece, and with the proper mist would even bleed across closely located nozzles into locally mixed colors. the trick of course is wheels that won't mar the applied paint... which could be accomplished by rotating the support wheels about the print head depending on the motion of the head, which via the two wire method can be in almost any xy direction and will stay a uniform depth off the wall.

Multiple tools certainly.

Age and the current economic system is catching up with me. I just got underbid on two decks by $400 on each. The thing is I bid them at materials + $200. They guys who won the bids can only do this because they will be able to do both decks on the same day. I need an assistant, but work is so sporadic I cannot keep a helper on line. Indeed most of the work I am doing as of late I don't need an assistant for. Hence my desire to create some mechanical ones.

Now, as the housing market recovers and the foreclosure inventory works itself out there will be lots of gut and replace jobs. So if I can get a robot to even do some of the bulk work while I handle the detail a one man crew can make money. (Plus the geekery will be a huge advert). Since I have more time then I want to have I figured I'd at least make the attempt.