19 hours of print time, and half an hour of removing supports later (multi-material/multi-color supports are horrible):

The main problems with this print were that the red turned out horribly. I printed it at 205 instead of my usual 210 to try and compromise between the black and red filaments on the right nozzle without having to issue M109s and slow down the process. Definitely a tradeoff there. Some of the red didn't adhere properly, evidenced by the missing bit right in the middle of the frog's back.

The other problem was that the supports, as I mentioned, were TERRIBLE to remove. I think I have a small (0.1-0.2mm) adjustment to make in the hotend offsets, because the supports were extra dense where the green met the red. The red tended to fill in between the walls of the green support, making it a solid block that was really stuck on and really hard to break up.

The other issue is that I under-extruded slightly. That obviously didn't help the red situation either. I had been playing with my extrusion multiplier a little and I think I had it too low for this print. That seems to be why the seams between different colors are so obvious (though it gives it an interesting "bumpy" quality, it also means it's extra fragile -- I nearly broke off a front foot trying to remove the supports). I'm pretty sure this is the main reason for the wide seams, and it's not just a nozzle offset problem, because you can see the space between the green and white clearly, when both those colors were being extruded from the same nozzle (so there should be absolutely no offset issues).

I did get one designed for the left side, and it printed perfectly and fit well -- sort of. But I made the cup too tall by accident, and there just isn't enough room to get that blade past the hotend without it hitting the bed mounts. The cup idea is pretty silly anyway, since the amount of purged material I'm dealing with far exceeds what a cup this size could hold.

Anyway, I think it was a neat attempt, and the clip-on part for the Z carriage could be useful elsewhere. If I ever redesign my bed mounts I might revisit this idea.

In other news, I attempted to print a 4-color frog based on the fun 2-color frog on thingiverse: https://www.thingiverse.com/thing:329436

I printed this at 0.2mm layer height by just purging into thin air and adding an ooze wall to keep the strings off the print. It's a messy process, and there are still some bugs to work out, but the results weren't completely disastrous. I had a lot of trouble removing the supports and ooze wall -- lessons learned below -- so it didn't come out that well, but you can still see basically what's possible here.

Things I learned with this print:

• Purging into thin air is fine, but it's good to clear out the mess once in a while so it doesn't build up too bad. It wasn't an issue for me, but in theory you could end up with the purge strings interfering with your bed, gumming up the works, or just getting dragged into your print

• I used S3D's "waterfall" ooze shield at a 2mm offset. With the messiness of the purge, the supports and ooze shield at this distance just ended up fusing with the print, and it was very, very hard to remove. I had to chip away at it gradually with wire cutters and needle nose pliers.

• All those tool changes, especially for a piece this small where the actual time spent per color per layer is short, dramatically increases the print time. This small ~3x3cm piece (I reduced its size 50%) took 9 hours to print. It's definitely not a very scalable process.
  

• Simplify3D, despite its $150 price tag, is really about $150 worth of bugs. If you don't set its ooze shield and supports to use All Extruders, it will happily attempt to cold-extrude from the inactive nozzles without using your toolchange gcode routines. I figured this out the hard way when I tried to set all the supports and ooze shield to print in white, which probably would have made removing those additions much easier.
  

You can see the interesting-looking ooze shield in this shot:

I'm attempting a new print of the same piece, this time with a larger offset for the ooze shield, and I'm trying out using a consistent temperature for both filaments in each hotend without cooling the inactive nozzle. Since I'm retracting the inactive filaments a whole 110mm, there's really not much chance for ooze anyway, so I'm trying to reduce the print time by keeping the empty nozzles hot. I'm also keeping the left nozzle at 190 and the right at 205 so that I don't need to perform any M109s during the toolchanges. That alone seems to be reducing the print time to about 6 hours from the original 9.

A rose is still a rose ....

or is called   potpourri  if dried and mixed with other flowers hmm..
Anthems Monster ??

So for the most part, this is just a home built open source printer that XYZ parts where used on? I mean if you get rid of the case, firmware, and so fourth it really is no longer right to call it a JR. anymore is it?

Good job by the way.

That's why, for lack of a better name, I'm calling it my Not-so Junior But yes, you're basically right. The frame, Z and X movements are stock, but practically nothing else remains of the original.

Good job by the way.

https://cad.onshape.com/documents/f2aa3 … 7e53c899a4

Once I get it all printed out and fully tested I'll post it on Thingiverse. For now I'll post the bracket and blade STLs here.

The neatest thing about this is I'm using a clip-on attachment for the left side Z carriage. I wasn't sure at first how best to attach this to the Z axis, but this clip actually works quite well, and it would be easy to modify to add other attachments/tools that need to follow Z.

The blade itself I print in TPU. It can be a little tricky to wedge it into the slot, but it fits snugly, and just barely grazes the nozzle which is exactly what I wanted. The really tricky part is that there's hardly any clearance between the center of my nozzle and my Y axis bed levelers, so the tip of the blade is just a mm or two away from the nozzle center, and my levelers just barely clear the end of the bracket when Y moves. Very tight spaces

The bracket has to be printed with the brace flat against the bed, and you have to use supports for the clip parts since they're then floating. S3D's supports are pretty nice and pop out very easily and cleanly. However, because of the length and thinness of the bracket parts, you almost have to have a heated bed to print this properly. Fortunately I just picked up an MP Select Mini last week, which is still both large enough and hot enough to print this bracket perfectly. Printing it on my Not-so-Junior did not work well. No heated bed there, and it's currently living in a chilly basement The bracket lifted off the bed by a good 3mm at the far end, making it not fit properly. You could probably manage it with a very wide brim.

Pics to come soon.

For a CMYK+white setup you'd of course need five extruders. That could be done on a RAMPS 1.4 board with a Stepper Expander x3, and you'd still have enough pins. Marlin, though, would take quite a bit more editing to handle that since I don't think it has defines set up in its configuration to handle more than 4 extruders.

Now, how the color mixing itself works, I don't have experience with that. I've read a few howtos on it, but I haven't grasped the configuration and firmware setup yet.

The biggest limitation I see at the moment is software support -- until we have software that will literally let you "paint" a model either with a virtual paintbrush or an overlaid texture (and somehow add a sense of "depth" to that texture for the solid nature of the model), I'm not sure a full-color setup is all that useful. Don't get me wrong, I'm not dismissing the idea or the concept of it. I just think it would be really hard to make it do exactly what you want it to.

Maybe such software already exists somewhere, but I don't know about it. I'd love to, if anyone is aware of any.

Anthem IMHO you deserve the mod of the year award for this .

I am not a davinci fan although I was tempted to by one as a first printer.
I did however get bit by the build bug and did a kit build and just ordered a second kit build. I know in the beginning I thought you crazy  for doing this extreme mod.
What can I say it works . you did things others including myself would have not attempted. Congratulations are in order and I hope this thread has led others to follow their dreams.
Tin

Wow, thanks, Tin Falcon.

I do this mainly because I'm a hobbyist and jack-of-all-trades at heart. When I see the possibility for improvement, and it's within my means, I become obsessed with making it happen. I also embrace the open source spirit. If I know how to do something, I like to spread the knowledge and make sure anyone who has the interest is able to as well, and then hopefully they too can help improve on it. I see 3D printing as a prime example of an area where that's as important as ever -- there's so much open source hardware and software for it out there already, and working together as a community, there's so much room for new ideas and methods to take root to take even at-home printing to new levels.

Anyway, thanks. No awards are necessary when I get feedback like that.

But since I'm posting anyway, I'll add another update. I tweaked my gcode to use the M218 command, and it works! By purging 25mm on tool changes I can even get a clean red. Pics or it didn't happen:

(I'm not quite sure why it looks a little over-extruded; probably because I just re-leveled the bed after installing the hardened nozzles and it might be a little too high.)

In that photo you can see part of the red and black bird's nest of purges. I don't have a proper purge bucket, so I'm relying on the low height of this print and my print bed to catch the purged strings as the hotend moves over the print. But as you can see, no need for an ooze shield or prime tower here, and it produces much more reliable colors. Purging 25mm of filament with a 0.4mm nozzle produces a string about 230mm long, and while you can still see a little bit of shading on the red perimeters, I'd call a 25mm good enough for most purposes (on my 230mm string, the red starts to look "clean" in the last 10mm or so, so I'm hopefully wasting as little filament as possible).

The downside to purging is that if you do waste 25mm (or however much you decide is good enough) of filament with every tool change. Say you're printing a 10cm tall piece at 0.1mm layer height, that's going to waste 25 meters of every color, and that number is solely based on layers (and exact number of toolchanges), not actual print volume. Not only is that a lot of wasted filament, but you'll have to empty your purge bucket several times mid-print. That, of course, is perhaps an extreme example, but it does highlight part of the problem inherent in sharing nozzles between colors.

Of course, it's a trade off. You can waste filament in purging/priming by sharing nozzles, or you can deal with the frustration of getting all your offsets just right and losing print volume (and increasing your out-of-pocket expenses) by actually installing four different nozzles.

At the moment, I'm going forward with the purge idea since it seems to work so well. I'm now designing some clip-on brackets that can be used for purge/wipe buckets

I am not a davinci fan although I was tempted to by one as a first printer.
I did however get bit by the build bug and did a kit build and just ordered a second kit build. I know in the beginning I thought you crazy  for doing this extreme mod.
What can I say it works . you did things others including myself would have not attempted. Congratulations are in order and I hope this thread has led others to follow their dreams.
Tin

I tried printing with TPU this morning, and everything seems to be working fine (at 25mm/s at least), which I think is pretty impressive. That means I haven't lost any obvious functionality by adding in the y-splitters.

Also, since I've been printing in a lot of exotic filaments lately, I decided to invest in some Micro-Swiss hardened nozzles, and I got those installed and my bed re-leveled as well (the new nozzles are about 0.2mm longer than the stock E3D brass nozzles).

And, finally, I completed another "test" print with four colors. This one was a slightly simpler road cone, where it only changed colors on layers. Unfortunately I broke it between the black and green layers while trying to remove it from the bed (slipped while trying to remove the ooze shield), but it otherwise turned out quite well! I'm still having an issue with the black bleeding into the red on the same nozzle for some reason, but I still suspect it has to do with the characteristics of the black filament at that temperature. I'll be trying some different filaments soon to see if they behave a bit better.

Also, I think I've figured out how to solve my tool offset/purge on the right nozzle problem. Gcode M218 can programmatically reset the hotend offsets, and I found that if I reset my T2/T3 offsets to 0 before switching tools and performing the extrusion G1, I can get it to not move the X axis. After the purge, I can reset the tool and reset the T2/T3 offset back to the correct settings.

Unfortunately this means a little more complication in the gcode, and I have to duplicate my hotend offset values in my toolchange gcode (I usually prefer to have these measured values set only in firmware), but I think it's workable. Now I just need to design a catch or slide for the purged filament, and I'm thinking I'll create a thin TPU "blade" that can wipe the hotend as it returns to the bed, as well. That should eliminate the need for prime towers or excessive ooze shields entirely. Maybe, if I perform a standard retraction after the purge, I also can mostly eliminate the need for an ooze shield as well.

This has been a fun but frustrating ride. I've had the hardware set up for several days now, but getting Simplify3D to work well with four colors has been tricky. It likes to make a lot of assumptions.

This is before "cleanup" with the 3-layer/4-color ooze shield still in place, and it highlights some of the problems and solutions to them I've been solving:

(And yes, it looks like I need to tighten up some belts)

Surprisingly, I haven't had too many issues with the filament not reloading into the adapters properly. I gave everything a good reaming with drills and files to eliminate some of the early issues, but I haven't hit any major issues with stuck filament lately.

Most of the problems have been software related. First off, while S3D is pretty powerful and useful for a lot of things (and I like the surface finish I get with it), configuring it is really, really hard. I hate its configuration model. Give me Slic3r-like settings any day.

Anyhow, getting the retraction settings coordinated is the first major hurdle, and it wasn't too bad, but S3D likes to issue a toolchange on the first tool. That means my start gcode has to reload filament, only to have it retracted and reloaded again before anything even starts printing. Weird and annoying, but not a big deal.

Getting a clean print is tougher. With dual extruders, I already was decreasing the temp of the inactive extruder in my toolchange decode (S3D doesn't have built-in support for that, even though with a $150 price tag it seems like it should -- even Slic3r can do it with a prefab setting). That helps a lot. And with the filament retracted 100mm above the adapter, that means it's virtually impossible to get leakage from your inactive extruder.

The bigger problems have to do with purging colors using the same nozzle. S3D, again, fails it here. While they offer prime towers, the prime tower only appears to work with two colors, and there can only ever be ONE prime tower. Seems like for best results there should be one tower per nozzle, but there's apparently no way to achieve that, and S3D assumes I want the prime tower to use the first two colors (which for my setup are on the same nozzle, meaning the other nozzle never gets primed). So, S3D prime towers are stupid and useless with my setup.

I tried using a "purge bucket" technique, by moving the hotend away from the bed, then purging ~30mm filament after changing tools. That produces great colors, but there's a fatal flaw in my setup, where one of my nozzles is always over the bed. Purging over the bed is bad.

So, I tried a technique where the nozzle to be purged gets moved off the bed first. That means purging the left nozzle moves the carriage all the way left, and vice versa. This works... mostly. Problem is, because I have to change the active tool in order to purge the right nozzle, Marlin applies the tool offset as well. So as soon as the G1 command to extrude gets processed, it extrudes while slowly moving the carriage back over the bed. That turns into a big mess at the end of the bed that ends up dragging goop all over the print.

I'd really love to go with a purge bucket technique, since that eliminates the need for any other sort of priming and produces really clean colors, but unless I can get this offset problem figured out (where somehow I can purge T2/T3's filament without applying T2/T3's X offset), it's off the table.

So, my interim solution is to use an excessive ooze shield. This, at least, S3D does well. You can make the ooze shield print with all of the colors after a tool change, and you can set the offset and number of walls. So I create a very heavy ooze shield (in the piece shown, each loop is about 10cm in length. At 0.1 layer height, 0.4 extrusion width, each 100mm of extrusion counts for about 4mm of filament length, so three loops purges about 3cm of filament.

With that setup, I'm still getting some unclean color in the red (tool 4) after printing black (tool 3), making the red muddy and dark, but I'm thinking this has to do with my filament temperatures. This red prints at 210, while the black prints at 190. I'm thinking what's left of the black that's still in the nozzle when the red primes is basically liquifying at that temperature and mixing with the red. A longer purge of the red would probably help there, but without the purge bucket idea working, that's hard to achieve cleanly. You can see more evidence of this in the "pre-cleanup" shot where there's also a big black streak where the filament started to purge (or leak) on the reload.

Anyhow, here are some more photos. This one's of the adapters attached to my two E3D Lite6s:

And a shot of my four extruders:

They're all different colors for easy identification. In tool order, it goes red, blue (it's in the background attached to the other side of the frame), black, then white.

All my settings at the moment are S3D specific. I configure four extruders with normal settings, then I set up two temperature controllers, one for the left nozzle, one for the right. I then define a new process for each color I'm using, and in that process configure the temperature controllers appropriately for the active/inactive tool temperatures I want. Unfortunately, though, S3D only uses the temperature controllers between layers. I need temperatures to change for the same nozzle within the same layer. So that part is handled with my toolchange gcode.

The Start, Toolchange and End gcodes work together to manage the retraction. When off, all filament needs to be retracted above the adapters. The start gcode reloads the filament (100mm in my case) up to the nozzle after heating, and uses G92 to make sure the other tool E positions are set correctly. S3D issues a G92 E0 right at the beginning of the print, after the start gcode, so this only applies to T0 which has been loaded to the nozzle.

The toolchange script moves the X carriage all the way left (so that if there's any leakage during retraction or heating, it doesn't hit the print) and retracts the filament to -100 before switching tools (this happens after any layer change or normal retraction is done - I use 6mm for my bowden setup). I set S3D's own toolchange retraction setting to 0 so that I can override its behavior without surprises in my gcode (otherwise, it would force me to use the same retraction and reload feed rates, which I want to customize -- retraction can be much faster than the reload, which needs to be slower toward the end in case of accidental purge). After fully retracting, it readjusts temperatures, switches tools, and reloads the filament before continuing the print.

This setup requires an ooze shield, not just to avoid leaks from the nozzle when moving X back into place, but also to  prime the colors as I explained above.

In the photos above you can see a few artifacts of the X carriage "idle" position while reloading and reheating -- there's a couple of streaks where there was some accidental purge and the ooze shield didn't catch it all. I'm still playing with retraction settings to see if I can avoid this (another "normal" length retraction just before moving X back over the print might help).

Start gcode:

M42 P58 S255 ; turn on the lights
M104 T0 S[extruder0_temperature]
M104 T2 S[extruder2_temperature]
G28
M109 T0 S[extruder0_temperature]
M109 T2 S[extruder2_temperature]

T0 ; reset tool to T0
G92 E0
G1 E110 F1000 ; reload T0 filament at a medium pace

; record other filaments as retracted
T1
G92 E-110 
T2
G92 E-110 
T3
G92 E-110 
T0

Toolchange gcode:

; toolchange script runs post-retraction old_tool's E is <= 0 right now
; don't use S3D's toolchange retraction with this or distances will get messed up

; move new hotend out of the way for cleaning purge
T0 ; move relative to T0
G1 X-32 F6000; move X left for reload

; if we want to purge
 do this...
;{IF NEWTOOL=0} G1 X-32 F6000; move X left for T0 purge
;{IF NEWTOOL=1} G1 X-32 F6000; move X left for T1 purge
;{IF NEWTOOL=2} G1 X155 F6000; move X right for T2 purge
;{IF NEWTOOL=3} G1 X155 F6000; move X right for T3 purge

T[old_tool] ; switch back to old/current tool for retraction past splitter
G1 E-100 F6000

; change temperatures
{IF NEWTOOL=0}M104 S[extruder2_temperature] T2 ; set T2 inactive extruder to cooldown temp
{IF NEWTOOL=0}M109 S[extruder0_temperature] T0; Set T0 new active extruder to normal temp and wait for it to reach temperature before proceeding.
{IF NEWTOOL=1}M104 S[extruder2_temperature] T2 ; set T2 inactive extruder to cooldown temp
{IF NEWTOOL=1}M109 S[extruder0_temperature] T0; Set T0 new active extruder to normal temp and wait for it to reach temperature before proceeding.
{IF NEWTOOL=2}M104 S[extruder0_temperature] T0 ; set T0 inactive extruder to cooldown temp
{IF NEWTOOL=2}M109 S[extruder2_temperature] T2; Set T2 new active extruder to normal temp and wait for it to reach temperature before proceeding.
{IF NEWTOOL=3}M104 S[extruder0_temperature] T0 ; set T0 inactive extruder to cooldown temp
{IF NEWTOOL=3}M109 S[extruder2_temperature] T2; Set T2 new active extruder to normal temp and wait for it to reach temperature before proceeding.

; switch to new tool and issue our own restart gcode. S3D's restart length if set will extrude beyond this point

T[new_tool]
G1 E-30 F6000 ; restart (fast most of the way)
G1 E0 F300 ; restart slow last 30mm

; skip purge for now
; G1 E25 F100 ; purge 25mm
; G92 E0 ; reset position to ignore purge length

End gcode:

; end gcode
; E is at <=0 (retracted) at this point

G28 X ; home X to get it out of the way
G1 E-100 F600 ; retract filament into Y-splitter
T0 ; Switch back to T0

M104 S0 T0; turn off extruder 1
M104 S0 T2; turn off extruder 2
M140 S0 ; turn off bed
M106 S0 ; turn off cooling fan

G90 ; absolute positioning

; G1 Y400 F5000 ; present part

M84     ; disable motors
M42 P58 S0 ; turn off the lights

This weekend I might try translating my settings into Slic3r to see how that goes. I worry about priming, though, because Slic3r's ooze shield is only a single color, and it has no prime towers to speak of (though there's a new post-processing prime tower script I found on reprap.org that I'm eager to try out).

I also haven't tried printing TPU in this setup yet. I was able to print TPU at <30mmps with these extruders previously, despite being a bowden configuration, but I'm wondering if the extra long retractions and the physical features of the Y-adapter construction might make that much more difficult. I'll definitely post here once I've tried it.

It appears to be better to share a nozzle with consecutive extruders, so that left = (0, 1) and right = (2, 3). This is because slicing software typically prints with each extruder in numerical order, so if you interleave the numbers like I have, then *every* tool change requires a nozzle reheat. If you map them consecutively, however, then you halve the number of reheats needed.

Should be a simple change -- with RAMPS I can just move the stepper cables to switch extruders and then update the extruder/nozzle mappings I put into Marlin.

Lessons learned:

• This Y-splitter works fine: https://www.thingiverse.com/thing:1680034 however, it can be finicky and you have to do some cleaning up of the print. I used a round 150-grit needle file to clean out the inside of it to make sure the filament wouldn't catch going through. This was actually my third time printing these devices, not being happy with my first two results. This last best print came out of my new MP Select Mini using Inland White PLA (of which either I got a bad batch, or it just always really sucks -- it has terrible layer adhesion and 210c is a minimum for a print that doesn't just fall apart)
  

• PC4-M5 fittings can also be really finicky. The last batch I ordered had a really sharp ridge on the inside of the M5 threaded end. I again had to use a round needle file to smooth down these edges on all the PC4-M5 fittings I used for the Y-adapter to make sure the filament didn't catch during retraction/reinsertion.

• I also had to drill out the M5 end of the "exit" fitting (hotend side) because the hole was smaller than the one in the Y-adapter, so the filament was getting caught on that edge too.

• Related to the above, when screwing in the PC4-M5 fittings (which self-tap just fine into the adapter), resist any temptation to NOT screw it in all the way (for fear of breaking the print). Kind of like the hotend barrel, if you leave any gap between the fitting and the Y-adapter, the filament will find a way to get stuck in that gap.

• Don't forget to add a prime tower (I forgot to in the print above -- you might notice some green/white blending if you look really close)

The gcode I used is actually pretty simple. My start gcode assumes that the filament has already been retracted into the Y-adapter (the full retraction distance I use is 110mm), so it starts by heating, then slowly re-inserting 110mm of filament.

The toolchange gcode then quickly retracts 10mm (to start the filament cooling), moves the X carriage back to 0 (to get it off the print), and then quickly retracts the remaining 100mm to clear the Y-adapter. It then switches temps (I cool the inactive hotend to 160), switches tools, and slowly reinserts 110mm of the new color.

I'll try to get my Marlin patches and all that posted up soon. I did find out that setting the per-nozzle temperature via the LCD isn't fixed by my patch the way temp settings via gcode are fixed, so I might either find a way to fix that, or I might live with it for now (the workaround is to set the OTHER nozzle's temp to the same value before Marlin times out with a "heating error" -- certainly possible to do if you remember to do it).

I've ordered more PC4-M5 fittings... I ran out But once I get them I'll be setting up my fourth extruder and the Y-adapter for four glorious colors.

Next extruder upgrade... mixing hotend (maybe a diamond?) with Y-splitters for full six-extruder CMYKWT color?

I've had two additional extruders printed and assembled for some months now, and I've had the extra stepper drivers on hand, but naturally as RAMPS 1.4 only provides two extruder driver slots, I haven't been able to hook them up.

I had figured out the pinouts on the drivers and had a breadboard implementation kind of working, but let's be honest -- I didn't have the time or motivation to "do it right" and solder my own board to match the schematics for that, and I wasn't about to hook up the breadboard implementation to my RAMPS board (I've learned my lesson about leaving things "loose" like that too many times).

So I bit the bullet and bought a couple of these beauties: http://www.reprap.me/stepper-expander-x2.html

I hooked one up to my Not-So-Jr and have two more steppers controllable as extruders now. This weekend I plan to screw everything together and mount the new extruders.

I'm going to be setting this up in a 2x2 sort of configuration, with two extruders each leading to two Y-splitters on each of two hotends. I'll need some custom gcode to control that, but it's a problem that's been solved before.

Now, one thing I had to solve, is that Marlin does not have support for "sharing" hotends between multiple extruders. It's easy enough to set the TEMP pins the same and see the right temps associated with each extruder, but I found that setting the temps is tricky -- to actually get a hotend to heat, because Marlin thinks each of the four extruders to has a unique heater, you have to set the same temperature on BOTH extruders at the same time, otherwise the PID heating algorithm just keeps setting and unsetting the heating pin and basically nothing happens.

So I had to fix that -- and fix it I did. I just found the "setTargetHotend" function and added some new settings so that any time the temperature gets set for EITHER of the extruders with a shared hotend, it actually sets them both simultaneously. Not too hard.

This actually seems to be working. I've done a few single-extrusion prints since setting that up and things are working fine. The only thing standing in my way now (besides me) is the fact that I'm running low on bowden fittings (I've had to replace two or three in the last couple months), and I need to re-print the Y-splitters I'll be using, since one of the first designs I tried didn't work so well.

Be sure I'll keep you posted. Hopefully pretty soon my hacked-up Da Vinci Jr will have two-bit color