Factorio Color Printer v3.0, 125 colors, 2.16 megapixels

[travvo]

Factorio Printer v 3.0

Get the save file [HERE](https://drive.google.com/file/d/1XFU56b ... sp=sharing).

Get some pre-made image blueprints [HERE](https://drive.google.com/file/d/1ufyGm8 ... sp=sharing).

Mods needed: Editor Extensions, Color Coding, Whats a Spoilage, Pushbutton, all 3 Space Age mods.

Input image is processed with python, which is then dithered and color-matched to the palette of 125 available items. A blueprint with 720 constant combinators, each with up to 750 signals is exported, named and ready to be imported into Factorio.

Previous version of printer [HERE](https://www.reddit.com/r/factorio/comme ... 80_pixels/). The conversion to 2.0 and Space Age was marvelous. Legendary stack inserters can move fast adjacent inserters can all swing every ten ticks, and fill a belt. This is much easier than in previous attempts where the timing of each inserter had to be staggered. In addition, the ability to store hundreds of signals in each combinator is awesome, and selector combinators are game-changers. By contrast, my version 2 required lots and lots of decider combinators, look at the size of the construction in [map view](https://imgur.com/kGznhSF).

Feel free to ask any questions, and send pictures if you want me to convert to blueprint/take a screenshot. I will have my code up on github by the end of the weekend. Long term goal is creation of custom supporting mod, increasing color depth, and creating working website a-la piebro's blueprint vis website.

Thank you devs, 5500 hours in and I love this game <3

EDIT: the github https://github.com/the-travvo/Factorio-Printer-v3

printer_run_2.gif (45.21 MiB) Viewed 1341 times

printer_run_1.gif (22.46 MiB) Viewed 1341 times

screenshot-tick-49546319_cr.jpg (3.67 MiB) Viewed 1341 times

screenshot-tick-49184605_cr.jpg (2.55 MiB) Viewed 1341 times

screenshot-tick-49146605_cr.jpg (2.69 MiB) Viewed 1341 times

screenshot-tick-49126975_cr.jpg (3.76 MiB) Viewed 1341 times

screenshot-tick-49123670_cr.jpg (3.47 MiB) Viewed 1341 times

screenshot-tick-49086344_cr.jpg (4.26 MiB) Viewed 1341 times

screenshot-tick-49082658_cr.jpg (4.05 MiB) Viewed 1341 times

screenshot-tick-49079727_cr.jpg (3.12 MiB) Viewed 1341 times

[eugenekay]

Wow!

I am not brave enough to download this (yet), but I am very intrigued.

It looks like this operates by generating a list/ordering of Items for each Inserter/Belt (translating into a Horizontal scan line)? That seems pretty neat, but it requires Belts to actually hold the display on the ground.

Can you do a version which creates Vertical scan lines instead? I am thinking about how to use this mechanism (or a derivative) to create a Nyan Cat style display ribbon as my Platform travels through space.

[travvo]

Wow!

I am not brave enough to download this (yet), but I am very intrigued.

It looks like this operates by generating a list/ordering of Items for each Inserter/Belt (translating into a Horizontal scan line)? That seems pretty neat, but it requires Belts to actually hold the display on the ground.

Can you do a version which creates Vertical scan lines instead? I am thinking about how to use this mechanism (or a derivative) to create a Nyan Cat style display ribbon as my Platform travels through space.

Correct, I started from the assumption that items would be moving from left to right. I built palette colors by taking screenshots of belts full of in-game items, and because sprites have different shapes and are not a uniform color, you'd want to create a new palette with the new orientation. Or, you could assume colors are similar enough, and rotate and go.

Are you interested in trying to match colors, or are you planning to just use (for example) all the bright sciences for the rainbow? One other consideration - once you have moving belts full of items, you have to deal with all those items. I use infinity requester chests and cheaty-deletey-underneathies to avoid this. In space, with no chests/bots, you will need a robust plan to handle that. I suppose you could load items via launched rockets, and dump them in space to get rid of them.

[eugenekay]

Correct, I started from the assumption that items would be moving from left to right. I built palette colors by taking screenshots of belts full of in-game items, and because sprites have different shapes and are not a uniform color, you'd want to create a new palette with the new orientation. Or, you could assume colors are similar enough, and rotate and go.

Are you interested in trying to match colors, or are you planning to just use (for example) all the bright sciences for the rainbow? One other consideration - once you have moving belts full of items, you have to deal with all those items. I use infinity requester chests and cheaty-deletey-underneathies to avoid this. In space, with no chests/bots, you will need a robust plan to handle that. I suppose you could load items via launched rockets, and dump them in space to get rid of them.

I'm not sure yet! I would like to avoid using an Infinity Chest, which means incorporating "Ink" production to the mechanism. The initial version will probably be very simple rainbow, but I think it would be awesome to recreate the "Star Wars" opening scroll text - or an arbitrarily-addressable "screen" without relying upon blueprints.

The screen itself will be dumping the items overboard, and relying upon the despawn fade effect, instead of using belts at all. It should be possible to adjust the scan rate (and thus - vertical resolution) by in/decreasing the platform speed.

[travvo]

Correct, I started from the assumption that items would be moving from left to right. I built palette colors by taking screenshots of belts full of in-game items, and because sprites have different shapes and are not a uniform color, you'd want to create a new palette with the new orientation. Or, you could assume colors are similar enough, and rotate and go.

Are you interested in trying to match colors, or are you planning to just use (for example) all the bright sciences for the rainbow? One other consideration - once you have moving belts full of items, you have to deal with all those items. I use infinity requester chests and cheaty-deletey-underneathies to avoid this. In space, with no chests/bots, you will need a robust plan to handle that. I suppose you could load items via launched rockets, and dump them in space to get rid of them.

I'm not sure yet! I would like to avoid using an Infinity Chest, which means incorporating "Ink" production to the mechanism. The initial version will probably be very simple rainbow, but I think it would be awesome to recreate the "Star Wars" opening scroll text - or an arbitrarily-addressable "screen" without relying upon blueprints.

The screen itself will be dumping the items overboard, and relying upon the despawn fade effect, instead of using belts at all. It should be possible to adjust the scan rate (and thus - vertical resolution) by in/decreasing the platform speed.

oooh, very fun idea. I've thought about trying to make images like that as well. The good news is that sulfur, space science, and carbon are some of the most saturated colors, along with EM science and Metallurgic science. All produced or transported in space. Let me know if you are interested in the data I gathered on item colors on belts, won't be exactly the same but would be close.

[eugenekay]

oooh, very fun idea. I've thought about trying to make images like that as well. The good news is that sulfur, space science, and carbon are some of the most saturated colors, along with EM science and Metallurgic science. All produced or transported in space. Let me know if you are interested in the data I gathered on item colors on belts, won't be exactly the same but would be close.

The Science Packs do look like good candidates - I am expecting to use a lot of Iron & Copper ore since it is so easy to produce in bulk (and I already throw a lot of Iron & Carbon overboard….).

Yes, please! There are a few collections of “item colors” for Factorio, but they seem to be focused on how the raw icon look or setting Train color masks; ignoring backgrounds and neighbors.

[travvo]

The Science Packs do look like good candidates - I am expecting to use a lot of Iron & Copper ore since it is so easy to produce in bulk (and I already throw a lot of Iron & Carbon overboard….).

Yes, please! There are a few collections of “item colors” for Factorio, but they seem to be focused on how the raw icon look or setting Train color masks; ignoring backgrounds and neighbors.

Here you go!

SA_Palette_original.csv

(17.93 KiB) Downloaded 25 times

I don't have a lot of the military/equipment tab mapped...by the time I've gone through three tabs I'm really sick of taking screenshots of belts full of items. Let me know if there's something you want to see that's not in the list, not a big deal to add it. You'll also notice I don't always use the correct internal name (blasphemy) because I was typing the names on the fly for my reference - they aren't needed by the code for blueprints. It also looks like this has the 'super' and 'infinity' items from Editor Extensions mod mixed in.

[mmmPI]

I don't have a lot of the military/equipment tab mapped...by the time I've gone through three tabs I'm really sick of taking screenshots of belts full of items.

When reading this i thought : " how come the person who made the awesome printer hasn't automated this ? " sounds like it would be easier, and i was puzzled by the csv in previous post too because i saw no way to use it, thanks for adding the github link in the OP it was not missed !!

Now i'm puzzled because there's a lot i need to document on to understand even the first part where you describe the process but it's fascinating !

[travvo]

When reading this i thought : " how come the person who made the awesome printer hasn't automated this ? " sounds like it would be easier, and i was puzzled by the csv in previous post too because i saw no way to use it, thanks for adding the github link in the OP it was not missed !!

Now i'm puzzled because there's a lot i need to document on to understand even the first part where you describe the process but it's fascinating !

Cheers and thanks! Let me know if I can answer any questions

[mmmPI]

Cheers and thanks! Let me know if I can answer any questions

Yeah so it's hard to ask question on such impressive project when you not sure you understood something without sounding like repeating something already said, get it right it's useless, get it wrong and you sound like an idiot or insulting the effort already made into the well detailed explanations.

I tried to understand what it would take to automate the creation of the screenshot you mentionned, so it's the palette creation part, which i saw in the description was using some image analysis done in R, and it's too bad because i had never heard of it, only a little python with Gimp. So i imagine the R part is aiming at analyzing what is the resulting color when a full belt of item is given as input as screenshot. Maybe removing some parts that are not actually covered by items, or offsetting the resulting color.I thought with the timelapse tools existing already and a specific "good blueprint" you may be able to automate the "taking of the screenshot" ? ( to feed the image analysis ).
( although i understand in practice "making a good blueprint" may require manually inputing all items signals that are supposed to be put on belts or similar chore that wouldn't really make the automation qualify as a success in terms of efficient use of time, more like "finally it's automated" "tm").

And that doesn't yield a palette, the iterating process from black and white i had never heard of something like this although several similar things that reminded me of it like music notes and scale or choosing the white balance settings but for every color (and notes), i thought it could be "cheating" but if you make modded item that are just square, and you define their color, you may be able to skip the image analysis part for the calibration ?( no boring screenshot), considering you'll know what the resulting color would be maybe you could go straight to the palette part and associating each item to its proper color without having to analyse how it would look in game ? At this point i'm feeling like i must be missing something x).

[travvo]

I tried to understand what it would take to automate the creation of the screenshot you mentionned, so it's the palette creation part, which i saw in the description was using some image analysis done in R, and it's too bad because i had never heard of it, only a little python with Gimp. So i imagine the R part is aiming at analyzing what is the resulting color when a full belt of item is given as input as screenshot. Maybe removing some parts that are not actually covered by items, or offsetting the resulting color.I thought with the timelapse tools existing already and a specific "good blueprint" you may be able to automate the "taking of the screenshot" ? ( to feed the image analysis ).

Yeah, that's a real solid idea, actually. You are correct that the R portion is analyzing the resulting color of the screenshot, which could easily be done in python. I already have a small setup that allows me to change a signal and many belts get filled with the objects, and then I use

Code: Select all

/screenshot 100 100

and it takes just a small 100 x 100 .jpg of the belts full. My palette folders look like:

palette_snip.PNG (367.87 KiB) Viewed 887 times

and R loops through, finds the average RGB, and does some adjustment to boost saturation and contrast. From there I have a table of item name, R, G, B, that I can use to find the 'nearest' color to each pixel in an image.

I could set up a timer and combinator with all the signals, so that it advances the signal every n ticks, and use the step forward function in editor like you mention, take all the screenshots, and then decode the blueprint for the combinator and use the list to programatically rename the image files.

In practice, it doesn't take me too long to name them by hand, so it never really seemed worth the time to make it more automated (than I already had it). But maybe I'll try this. There's probably a way to program through the console to go forward by n ticks and take a screenshot, but I don't know anything about console coding.

[eugenekay]

There's probably a way to program through the console to go forward by n ticks and take a screenshot, but I don't know anything about console coding.

--load-game and --until-tick work great for automating this sort of thing. Programmatically loading blueprints requires a Helper mod, which moves the task to the gray area of the graph.

[mmmPI]

Hey thanks for the confirmation and additionnal infos ! It clarifies things a lot, the palette folder look familiar to many things x). I wasn't sure what kind of screenshot it would be, as i imagined it could also be large screenshots with different items and something in R to map different area of the picture to different item when calculating the average R G B or yet another new thing.

I could set up a timer and combinator with all the signals, so that it advances the signal every n ticks, and use the step forward function in editor like you mention, take all the screenshots, and then decode the blueprint for the combinator and use the list to programatically rename the image files.

I'm glad that confirm i was not completly off when trying to understand the printer x) , i wasn't too serious about, i'm talking more from "theory standpoint" eugenekay raised a good point about the time needed to automate something you may not need do that many times x) .
But just in case, I made a mod where an AI read the name of all the items in the game, and had to rename all the sound sample (multiple times due to attempt at improving the quality) and made this renamer script as 2 files, a .txt with the old name and new name in front and a .bat file, you place them in a folder, double click the .bat, and it rename all the file in the folder.

renamer.zip

(1.29 KiB) Downloaded 28 times

It would only leaves this part :

There's probably a way to program through the console to go forward by n ticks and take a screenshot, but I don't know anything about console coding.

--load-game and --until-tick work great for automating this sort of thing. Programmatically loading blueprints requires a Helper mod, which moves the task to the gray area of the graph.

Ok so maybe this idea is based on my ignorance of the difficulty, but i thought you could only need 1 blueprint. And not need to progammatically load them, the blueprint need to be able to put some item on belts in a controlled manner so that if you don't move your character and some screenshot are taken automatically by a mod or command or autoclicker x), it would contain a different item. That sound like a modified version of the current printer to me, one where you "could" set up on the left manually, a test run for a printing machine, where it just print a single small square/rectangle of every single color in a row, then you walk your character in the position where you can run your command or activate the mod that takes a picture every XXX ticks. Printing a palette ?

[travvo]

Ok so maybe this idea is based on my ignorance of the difficulty, but i thought you could only need 1 blueprint. And not need to progammatically load them, the blueprint need to be able to put some item on belts in a controlled manner so that if you don't move your character and some screenshot are taken automatically by a mod or command or autoclicker x), it would contain a different item. That sound like a modified version of the current printer to me, one where you "could" set up on the left manually, a test run for a printing machine, where it just print a single small square/rectangle of every single color in a row, then you walk your character in the position where you can run your command or activate the mod that takes a picture every XXX ticks. Printing a palette ?

That part would be easy. Set up a constant combinator with T = 1, and an arithmetic combinator that outputs T, and attach the constant to the input of the arithmetic, and loop the arithmetic to itself. This will start counting from T = 1 and increase. Wire this signal into another arithmetic combinator with K = T // 300. K will be the index of the item we want to see on the belts, so K changes every 5 seconds. Make a constant combinator with signal 1 for every item you want to see, and make sure you put them in the order they appear in the item GUI. Wire the K signal and item signal into a selector combinator, programmed to sort signals based on K. Then, every 5 seconds, the selector will be outputting a new signal from your set. Wire this signal to creative chests with loaders, and you will have belts that switch being full of items every five seconds. Then, either advance the time by 300 ticks and use the screenshot command, or script something a loop that does that.

When it's time to get the items, you only blueprint the signal combinator. Extract the indexed list of items, and use it to rename all the screenshots.