Adding onto this - if it can't recognize a foil or any card for that matter an "undetected" pile might be a good idea instead of potentially holding up the machine waiting for user input.
Glad to see these videos taking off. And VERY glad the algorithm recommended this channel. Love MTG and love engineering. Literally, two of my favorite things rolled into one.
I'm interested to see how well the card-swiping mechanism works. I think there's a reason all the commercial systems use a suction cup and bend the cards slightly after picking them up - the cards really like to stick together.
Maybe having another rubber component (some sort of lip for the cards to pass over) touching the underside of the cards could keep any extra cards from sliding in with the intended card?
@@paulkelemen4199 yeah that’s the kind of thing you’d have to do. Perhaps even a set of constant turning rollers pushing cards back onto the stack from the underside, so that only the card being directly grabbed by the sweeper can overcome it
If you're sorting by color, you may want multicolor piles and lands. Also tokens/other pile for anything that is either not applicable to those 8 or as an unknown pile when confidence is below 99 or 98%
Hi Jack, thanks for sharing this info about your process, I'm very invested in this! I would love the ability to sort bulk with a machine that I made myself or with a cheap machine available commercially. You have a background in engineering so you may have already considered this, but I'd like to recommend you take a look at the Voron 3d printer project. It's an open source 3d printer with a focus on being as 3d printable as possible, and I think it would be great inspiration for this project! Keep up the great work!
Thanks for the support and info Jacob! The Voron project is super cool and a great example, I appreciate you sharing! I'll definitely take a closer look at them for some inspiration.
I'm concerned the convergent chute design is going to ding the corners of cards. Why not go with 2 straight chutes since you are already motorizing the end stacks. Just program it to alternate between the 2 chutes as required.
You can make a suction device just as easily as the flicking mechanism by using a big plastic syringe and a a servo to push or pull the plunger. Flicking the cards has potential to ding up the edges, scratch against the aluminum extrusion, or just flat out complicate the parallel processing because now you've got to deal with stacking tolerances between cards.
You're right I could engineer a pretty cheap suction style device, but then it has to move and deposit the card into a pile, which is another set of motors. It's certainly a idea worth keeping in the back of my mind, I appreciate the input!
That's the problem with "mulling things over", is that it's just kicking the can down the road. If you have an idea, work on it before someone else makes it.
I foresee the biggest issue in the design is the chute and swiper. Cards will 1) get jammed in the chute sideways. 2) Not singulate properly, as they may bend a little when coming up the stack 3) May not have enough velocity to slide into the hoppers. Super cool design, though. I was thinking of making the same thing!
@@jackbaumgartel Came here just to say this. I've already tried doing a similar design with chute, and all three problems were present. The suction + slight bend to suctioned card fixes the problem with picking up exactly 1 card. Chutes + stacks + relying on gravity always produce inconsistent results. Placing the cards with a gantry-style placer is the best way to go. You can see my attempts to do this (albeit with LEGO Mindstorms) on my channel. I did also try the gantry-style suction picker (also with LEGO Mindstorms) but at the time I couldn't get it to reliably place the cards into each pile (LEGO Servos aren't intended for millimeter accuracy) and I couldn't get it to reliably determine which card was which (lighting and camera focus made this incredibly difficult). I eventually abandoned the idea as I didn't feel that the original design of using something modular, familiar (like LEGO), and easily customizable led to a "good enough" product. I like what I see here so I wish you luck!
Regarding number of piles, I sort my cards based on set then collector number so ideally would like the machine to be able to sort through a complete set in no more than 3 passes. Given that sets normally have
Thanks for the input! I'm not sure if you've seen the video I did about my software plans for the project yet, but I've linked it here if you're interested! th-cam.com/video/T63tQijJzC4/w-d-xo.html
Love the idea of this and look forward to seeing how this proceeds! One things I'd suggest though, as someone currently sorting through their bulk by hand, would be to not worry so much about that sorting speed. As long as it's relatively quiet and safe to leave running, having something that can just keep cranking away when I'm at work or sleeping would be awesome. Compared to the established machines that are meant more for game stores where time is money for me just being able to better use the time I can't currently matters far more than efficiency.
you could skip the card merger and just use 2 seperate slides if you make the sorter slightly larger i dont think a cheap motor will be able to move the sorted card piles fast enough to get 1card/s, too much force needed for the acceleration
Great idea! 2 Separate slides is a much better solution. And did you get a chance to see the math I laid out in the video description? I ran some crude calculations to show that a 42Ncm Nema 17 stepper can easily whip around 8lbs of cards. (Unless I made an error in the calculations haha)
Keen to see where this goes, will definitely build my own to tinker with if you need testers. Thankfully for hobbyists it doesn't need the degree of accuracy and resiliency a card shop needs, as long as there's no vectors for damaging cards we can run it through and diff the catalogue to make sure it's correct. I think the commercial machines use a suction mechanism because singularly moving cards is the crux of the design. Any system of legs/wheels/belts trying to slide the top card is going to have a chance of taking the card beneath it, and unless your steppers are very accurate and there's a "ledge" to keep the stack trapped it'll run into issues. That said, stepper motors are plenty accurate for 3D printing so maybe it is viable... Best of luck, keen to hear more and participate.
Custom sort criteria would be a really good extension option, once you get a functional system (gotta avoid that feature creep). Could get to a very high level of specificity-eg: pull out every "goblin" creature type card-by planning multiple run sorts, too. Like in your present design, a colorless pile could be fed through again, to sort further into lands, multicolored, etc.
@@jackbaumgartel I don't know how do you intend to program it or what, but it would be ideal if the program expects a "sorting function", that receives all the data obtained from scanning the card (name, cost, power, toughness, types, etc), and returns a "number" to indicate which pile should go to. With this, a user could design its own algorithm for sorting with whatever criteria.
If you are going for a lower price point, then this is made to be more for consumer markets. If that's the case, I would say processing time is somewhat irrelevant and what's more important is the cost to purchase, energy costs, and noise. If it's cheap to buy and run and it's quiet I can just have it running the background over many days. A store has to process a lot of cards quickly a household person with a back pile of lots bulk has plenty of time.
Thanks for the input. A few others have expressed similar views and I appreciate the perspective. I'll be prioritizing noise more than I showed in the video.
If you go for 8 piles it may save space if the mixed pile is in the middle with the 8 sorted piles surround it in a ring. This way either the mixed pile or the ring of piles could rotate to the required position and deposit the card. This idea is in response to the potential that the chute clogging and/or how overly long 8 piles in a row would be on your desk
Thanks for the input and I appreciate you reading through some of the other responses! You're right the rotating style of piles would likely save desk space compared to 8 linear piles.
Don't forget that you can do iterative runs of the sorting process. You could have multiple programmes, one that sorts first by colour and then rarity and then cmc etc
This seems like a neat challenge. On the topic of number of piles, I think you lightly touched upon it. A 6 pile design works for sorting all your colors in one go and as you mentioned in the comments, "colorless" being a catch all. An alternative if you want to save desk space at the cost of overall sorting speed is to have the output hopper types programmatically assigned. So say once you do your first run, you switch the software mode and then sort through the stack of catch-all that your first run left. This also would be useful if in the future you change the criteria you sort by. It would just be a matter of updating the detection algorithm to your new criteria and assigning the results to their proper bins. Either way, I will love seeing how far you go with it. Good luck!
I would add a 7th chute for multicolor (mc) cards; as mc's and colorless cards usually are the least represented cards in sets I would put these piles at the front and back to center the weight
Thanks for the input, a few other people have mentioned another pile as well and I think it's a good idea, centering the weight based on predicted occurrences is a great idea!! I appreciate you sharing!
Looks super cool, i'm very excited to see your building process. Never thought about a sorting machine like that, so seeing all the other options in the other video was also really nice. I'd be interested in seeing how easy it would be to adapt a machine like that to work with different card sizes (notably yu gi oh cards are smaller than magic cards). Subscribed !
Thanks for the support and I'm glad you're enjoying the content! It shouldn't be too hard to adapt the machine for close card sizes as well, but that's a feature I'll tackle after I've got a working prototype.
This project is so cool! I think tour approach is really professional, and follows good design criteria. I personally think the tricky part will be in the stepper finger, because cards sometime stick to eachother, so maybe a roller/blocker might be needed, imo.
Making the receiving containers modules so you could add 1 at a time seems like a good design idea, I think I would setup a rack and pinion for shifting them left and right making it so you only need to add track and containers to expand. I don't know enough about these cards to know why you can't pull cards off the pile like printers pull paper, possibly making things even simpler for your design and dual side scanning easier.
I recently sorted my collection and scanned it with the dragonshield app. fron there i later imported the result to manabox. i have like 7 full binders one for each color, mc and lands and about 20 boxes each holding around 1200 cards of bulk. in all my binders cards from all kind of sets are found. i had to scan each card separately. this took me ages to do. once i finished it and reached the bulk filled with tons of cards from limited play i have changed to bulk scanning. for bulk scanning you sort your cards per set and tell the app which set your are going to scan. this greatly improves the scanning since it only has to check 250 possibilities instead of a few thousand. that was also the point where i no longer cared for the color or type of the card. i gave every box a unique letter like a,b,… and subdibided it into 6 segments. each segments holds arround 200 cards. so after scanning 200 cards of lets say ammonketh i marked that list as A1 and put the cards into is corresponsing place. so if i need any card from my bulk boxes i open the app and search in which segment the card is contained. after that i only have to search it in 200 instend off n thousands of cards. i'm now at the point where whenever i visit a limited event and end up with cards from one set i quickly scan them and put them in a new segment. My tip: don't sort the cards by color or type if you end up putting them into a box with thousand of other cards. make smaller piles sorted by set. much easier to scan and to find the cards later.
Absolutely signing up for the email list & hitting the bell. While I find organizing cards relaxing, I don't have the time I'd like for it compared to how many cards end up adding to the pile... While I know this device isn't anywhere near completion, I'm very excited to see the process and how it evolves. Couple of things I might add to the discussion: - The act of sliding the cards down a chute could cause harm to the edges as it bumps them to align with the slot/pile destination. While this arguably feels like me being overprotective, I would like to see if there's not another solution to prevent that. Maybe two chutes, removing any horizontal movement? Would also prevent the possible case of a card sliding sideways, like if it gets caught on some grime or has some defect with the card on the corner making it "swing out" while sliding down. - What say we remove the chute entirely and have it deposit pretty much straight into the piles, as the chute seems only used for aligning the two stacks into one output. This could save on costs to remove a custom part entirely. Might even give more control for their final orientation. This does impact the speed as you would have to move the gantry for every card every time. Even if both were greens, the two stacks are never lined up to the same slot. But I mean, with the chute system you had to account for the speed they're falling at or have some way to confirm they're in the slot else the movement of the gantry could seriously bend cards that were halfway in place. - Is there any planned way to make the final slots lower the pile of cards as it gets filled? I am concerned about overly long drops into the slots causing cards to angle themselves down and get jammed. Basically standing tall on the pile vs flat against it. If there was some way to keep them mostly raised that would be ideal. However then it's a cost issue of more parts per slot. Even a simple spring might work but that would have to be calibrated to maintain a good distance at the start with only a few cards. - I see people concerned about cards sticking, and yeah that's actually a really valid one. Especially with the taco cards. My first thought was to have the top of the stack be nearly in line with the chute but that doesn't really help with tacos or edge damage. What's the chance of a rubber bar on the bottom of the chute, right against the stack? It could raise up with less force than the swinging arms above and catch cards that might be sticking? Then the swinging arms on top can pull any cards back into place. It would be a balance to make the lower bar able to grab a stuck card while not scraping a non-stuck card however. As well as timing the lower bar to come up after the swinging arm has started, so it doesn't catch the front of the card that we're trying to move.
Thank you for the support and for all of your input, I really appreciate the time you've taken to share your thoughts here. First off a few other people have expressed similar concern over the chute, so I am planning on having two straight chutes instead of funneling the stacks into one! I do think some kind of chute is still a good idea though, as I want the cards to be moving at a more uniform speed when they enter the piles. The swiper will likely have some variation that I want to be smoothed out. I have thought about a moving bottom for the piles, so that cards don't have to fall as far, and that's something I'll be testing on the prototype. Lastly yes I didn't include it in the model but there will be a little rubber bumper to help prevent two cards from entering the cute on a single swipe!
@@jackbaumgartel I tried to see if other people had touched on em. Super happy that it's all been covered and acknowledged! Really looking forwards to this project as a whole and the future updates. I didn't think about the speed of the cards from the arms being variable. Curious to see a real test build and see what all the parts do in conjunction.
I think that push arm might be the biggest problem. First, it may get stuck as the end of the rod (where it touches the card) will not move in straight line - it will move in arc so you`ll probably end up overcomplicating the mechanism. Maybe something like printer`s paper release mechanism would be better? Second that finger may be pushing multiple cards at a time if the cards get vacuumed together. Additionally the card feeder should be at a slight angle back - cards stacked like this, especially a 1000 cards, do not like to stay in one place when touched.
Very elegant design for such a complex problem. Though something I didn't see addressed is the nature of real-world card swiping when faced with sticky, chipped, or curled (even non-foil) cards. One of my concerns about throwing a stack of unsorted bulk into a machine like this is what will happen to the various mechanisms when 2-4 cards are stuck together or when a bent corner catches and jams the whole thing. Excited to see how this project evolves.
I feel as if the chute combining the output of both of the feeder stacks is a potential problem where something could jam. Maybe if you just had independent output chutes for each pile it would be better? You certainly wouldn't have to accommodate a rectangular card rotating to slide down the chute and then rotating back to align with the collection pile. Also, you might want to consider having some safety sensor to ensure that your collection gantry never slides laterally while a card is still partially in the output slide. Potentially an optical sensor at the end of the slide? I'm stoked to see how this project progresses no matter what though! Good luck!
Yes a few others have suggested similar design changes and I do plan on reworking the convergent chute, it was not the best solution admittedly. I had thought about a sensor there too, but I fear the extra cost & complexity. I think I can ensure that issue doesn't occur with properly timed software commands, but if during testing things go south, an optical sensor is a great idea! Thank you for the input and support, I appreciate it!
TH-cam algorithm suggested this to me, love the project idea and while I'm not a huge retailer or anything something like this appeals to my lazier sensibilities
I dunno MTG but I did build my own laser cutter, I feel like this could be easily improved. You've done a few rookie mistakes and all I'm going to say is "keep it simple" Only sort one pile at a time, two piles means two sets of parts and two sets of places of failure and twice the price. Counter weight the steppers lifting plate externally. and that flicking arm is going to be trouble, I can feel it but my only suggestion right now is compressed air and that would mean starting again. Love the code too btw
My recommendation is to separate your two piles further and have your grabbers sweep them sideways into the center. It would reduce the travel time for the cards before the next is uncovered, speeding up scanning acquisition, and the chute design could be simplified as a simple ramp without the need for the angled edges to funnel the cards inwards. There might also be a way to connect the two sweepers into a single motivator with both arms attached, as it could be changed to a synchronized left-right motion of both arms. Of course you may prefer to keep the independent motion if your design allows you to sweep from the same pile twice in a row.
For the problem of raising the unsorted stack to the correct height initially, I think there's a really simple solution. You could slowly raise the stack, while using the camera you already have in the design to see when the edge of the topmost card is in the frame. Once the card reaches a specific height (I.e. a specific spot on the chute frame gets obscured by the pile), you stop the stepper motor. It would take some testing to make sure the latency from the camera, software, and motor controls doesn't cause the stack to over shoot. And it may be worth marking the inside of the chute, perhaps with some sort of solid colour sticker or tape to make it easier to see where the stack has reached. Also, you probably won't want to rely on the user making 2 exactly even sized stacks, since that takes a lot of fiddling. So you'd probably want this sort of system anyway, to tell when the system has been fed uneven stacks.
Thanks for the input! That's a pretty clever solution. I like that it doesn't require any additional hardware, just some testing and software adjustments.
I'm a new sub. Part 1 came across my feed and I was intrigued. I'm very interested in this. Main thing I'm looking for is an algorithm that will pick my Commander "staples" and separate them first. Than another algorithm to further sort based on card function.
If we are sorting one card while scanning another, then maybe it would be better to have two separate shoots instead of one. The difference would just be that the bin motor would need to shift one to the left or right by one depending on shoot being processed. But it‘a probably more reliable and slightly less stressful on the cards.
Maybe this is something you've already considered, but rather than hard coding WUBRG sorting into the project, it may be a good idea to set it up to pick a sorting type via software. For example if you have already color sorted and you then want to sort by mana cost or price. If it is not more versatile than color sorting it seems of limited use
Card sorting is cool, but I would love for a machine that is similar to this in design with the sole purpose of cart cataloging. Feed Card, scan card, input card into DB, move color pile, rinse repeat.
Thanks for the input, a few other people have expressed a similar wish and I like the idea too! I may have to do a 'base model' that just does cataloging, it would be cheaper to produce so it would be a more affordable version of the full sorting machine.
Having a circular end destination for the cards instead of a line would be harder to switch out sizes but make selecting compartments faster (since it can go all the way around) and also probably more stable as it doesnt have the same inertia of moving in a straight line? Might lead to some issues with the space it takes up however
Agreed, with the current layout, the machine could tip or jitter if the piles move too quickly. With a circular layout, the machine would spin in a similar situation. I think the larger space and higher cost is a disadvantage though for the circle design.
Have you done any testing of your card chute design? Looks like the cards might be free to wander about, and take varying amounts of time to reach their destination
I haven't yet, and thank you for the feedback. Software wise I'll make sure the pile is aligned before the card could possibly exit the chute, and I'll be sure to test the chute thoroughly once I've got a working prototype.
Very exciting stuff! Personally I would prefer an eight pile solution but even if there were three piles it would be useful. As long as affordability is alright then it could be very successful.
The thing that will slow this down, is you can’t really eject the first card until the proper pile is in place. This means you have to slide the whole set of piles, at worst from one end to the other. Depending on how long that takes, you could have both piles with cards scanned and waiting. You may need to have a better algorithm that is more like en elevator, where it doesn’t just go left pile, right pile, but tries to find the closest pile to one of the two cards and moves to that first.
That's a really cool idea! Assuming both cards are scanned moving the piles based on proximity rather than original order. Also take a peek at some crude math I did out in the video description, a cheap stepper motor should be able to move the (full) piles around pretty quick!
@@jackbaumgartel My initial thought of this is simply adding another servo at the exit of the merger, a simple gate (or even "ensure card ends up in pile cleanly" sweeper style?) such that you can have a one-card deep queue waiting for the bin/pile mechanism to be in the final placement for receive. If you are extra clever, you could make it a N-length queue that just progresses one card at a time once pile/bin is in place but that depends more on mechanism and code complexities.
@@jackbaumgartel I think you may be starting something that could become like the old maker-bot challenge. I would love to see competitive open-source card sorting machines. Especially after you showed us the cost of similar proprietary machines?! The companies can only charge such high rates because they don't have healthy competition. You are singlehandedly the most competition these companies have right now. As you showed, each machine on the market is astonishingly expensive, and they seem to try to be cordial in their designs. They try to be application specific via each machine to hopefully hit different levels of consumers. You're making a machine that will be a fraction of the cost of any competitor, and be adjustable to each application. Meaning it would become economically advantageous to build or buy, 4-20+ of your machines instead of ONE competitor machine. You're going to make alot of consumers very happy, and some companies very angry. I love it all!!! Thanks again! Keep up the good work! I love this project so much.
I have recently bought EPS32 camera modules, I paid 25€ for 2. It should have enough pins left to control all motors. At the highest resolution for the cheapest module it can take a little over 3 pictures per second which should be plenty. 1600x1200 should also be plenty to just find the color of 2 cards. Wifi control also means you could offload image processing to another computer of some cloud service.
@@jackbaumgartel Further clarification: The modules should be much cheaper. I basically bought sets on Amazon that include an USB programmer. So just the boards with wifi and cam should be available around 4-7€. Just did a quick check: probably the exact same product is around 8€ on Aliexpress.
I think the way of sorting really depends on further usage: deck building, collecting, trading and even on the format played. In my case (EDH) I have worked this way of sorting: rarity, colour, A-Z where rarity mixes rare&mythic, then uncommon, then common. Non-basic lands are mixed together without rarity dependence, basic lands are separate (treated as bulk). Also, A-Z sorting is just by 1st letter only, because it would be overkill for me to go with full A-Z every time I have to add cards (usualy dozens).
I'm eagerly waiting to purchase a functional model. Next project: a small battery operated automatic shuffler for Commander decks across a wide range of sleeve types.
This idea and engineering process walkthrough are both right up my alley, excited to see where this goes. Out of curiosity, how do you intend to protect the card corners on the chute? or protect the cards when they fall into the piles? I would hate to have all my corners knicked up from the machine or for my cards to fall at an angle and fold from the weigh of more card being put on top of them should I leave it unattended.
Great to hear! And I'm planning on redesigning the chute after some solid feedback in the comments here, the new design should be more gentle on card corners. For the falling into the piles properly, I'm going to see how testing fares in the real world and go from there!
I see a problem with the speed. You have the acces to scan fast and move to next scan fast by double slider finger. But your sorting pile area can not move as fast as you can scan. I had once a idea of a sorter, where not the storage area move but the slide ramp where the card slide on. The ramp can be mived quickly because it doesnt depend on the weight of the cards already sorted
@@jackbaumgartel awesome, I do aswell, could be awesome if you release the stl files, would definitely drive your price figure down. Maybe even open source it, could really help make this thing competitive with the big brands. Of course I understand if that's not something you'd want to do. I'd definitely be very interested in trying to help out, I'm sure other could be aswell.
Just a thought, I know there is a lot of the work done for the coding based on the other videos, but what about seeing if somehow you could integrate it with the mtgstand site. It already has a card recognizing script, and uses a webcame or any camera that can stay open.
I don't want the project to rely heavily on any third party processing like you mention, even though it definitely would make things easier for me! However, MTGstand accepts CSV imports, so once the machine scans cards, you'll be able to easily import the collection!
Really great starting point. Can't wait to see where this project goes. I'm a software dev with experience in computer vision. If you need help with the programing for the camera I would love to collaborate!
that arm with the rubber arm can quickly go from soft to scour texture if you cheap out on the rubber. imagine those pencils with erasers that become stones or charging cables that become stiff. imagine that rubbing across the face of every card. the orientation shouldn't matter. most AI card scanners I've used can also recognize the image being upside down
Thanks for the input. Yes all rubbers eventually wear out so material selection there will be important. For the orientation, in the video I more meant is the card face up or face down, I agree that software can flip a card rightside up if need be
@@jackbaumgartelWould not be posting about something as insipid as "wear" unless we are talking about what happens to the card when it meets a dry eraser/rubber band. One can imagine that it would be rough waking up and finding small vertical scratches on an entire stack of cards. Easily bump a card from near-mint to played because of a scratch.
if the flick method of pushing cards does not work i can show you a cheap way to make suction work. the only downside is the cost in motors more then anything. lastly i think you can easily make it 10 piles with little extra room. you can do a 2x5 config of piles and make the first set go under the shoot to save space. where both of these ideas might be an expansion they might be worth thinking about. o ya are you making this a open project or one you plan to profit from? i would like to build or buy one depending on how it turns out and what your plans are.
Thanks for the input and ideas! Definitely worth considering and I think you're right they sound like expansions from the current design. I'm not sure if the design will be open source or not yet, that's something I've been considering though, and great to hear!
@@jackbaumgartel if you wanted, you should look at something like the Voron style of open source. basically the code and files and BOM are open and free but people can sell kits. so you could do that to get the power of crowed sourcing and refining but still make some money for your time and effort.
Jack, could you elaborate on why you don't feel comfortable using a friction feeder mechanism? It's technically how the sheets were fed in the manufacturing process in the first place and properly tuned wouldn't cause any damage to cards. Additionally, I don't think the wipers are a bad idea I just think you might run into issues without some sort of gate or bottom side wiper preventing double feeds. Edit: also I just thought of this, do you have any idea how you're going to handle the increasing weight of the sorted piles when trying to move them at speed? I guess if you go slow enough?
I agree a properly engineered friction feeder would work great here, but it would necessitate a taller machine if the piles are still to be gravity-fed. That's the main reason I prefer an upward facing stack honestly. You're right about the wipers though, I haven't added it to the model yet but I'll be including a lower bumper/rubber pad to help prevent double feeds. And yes, I did out some crude math in the video's description showing how quickly the motor would theoretically be able to move full piles around! I appreciate the feedback and good questions!
Seeking knowledge of the active machines, and perhaps future suggestions. Which of these can alphabetize stacks of cards? Which of these is best at sorting by set? When a mixed box of bulk comes in, is there a better machine for that, or do we still need excessive manual sorting first? Thanks.
While I don't have access to any machines other than my own, I think it's safe to say they are each capable of what you're asking. I went through each machine's capabilities in detail here if you are interested: th-cam.com/video/sy7dtW8CvQ4/w-d-xo.html
Couldn't you also as a second shoot on the back of the machine to another gantry and 6 more slots. you could make it where the gantry is easy to detach ignorer to load cards. Then you could be sorting two cards at once cause lets say you have a red and a blue at the top it could push the red backwards to go in that slot and then the blue forward then you would not have to wait for the gantry to move back and fourth.
Unfortunately you need a good program that can set up customized organization which means lots of slot and piles. If I could I would organize everything down from besides what the card such as control, blocker, whip, fog, etc with sub cats for the abilities. Ex a hydra that a mana dork but also makes tokens
Oh boy you sort by specific card use? That's the first I've heard of such a method. How do you handle cards that belong in multiple categories? I'd love to hear more about how this organizational strategy!
Thanks for the question, it would just take a few passes through the machine! 1st pass perhaps sorts all cards A-E into pile 1, all cards E-I into pile 2 etc. Then you can put those piles back in afterwards to fully alphabetize them!
Personally, I believe the number of piles isn't much of a concern as long as the device is easy to load. Through software, the machine can sub-divide piles with multiple run-throughs. For instance: Let's say you're trying to sort by Converted Mana Cost but you only have 5 piles. The machine would be able to sort these cards with 3 run-throughs. The first run would separate the cards into two piles. [0-4] cmc and [5-9+] cmc. then, put the [0-4] cmc pile back in, then reassign the sorting parameters to 0,1,2,3,4 cmc (One for each pile). Then you do the same thing with the other pile. I guess this is a long-winded way of saying if you change your mind about the amount of piles late into the design process, then it's no problem, just subdivide! This design looks really good, but I have doubts about the reliability of a chute to funnel the 2 card piles into one. This seems very prone to possible jams due to inconsistent sliding of the cards. Perhaps you could have each card go straight down a path, and simply reposition the output piles on the rails to correspond to whether it is receiving from the left or right input pile.
Thanks for the response! You’re certainly right about subdivision and the number of piles, the more piles you have the less you’ll have to subdivide. And yes a few others have expressed concern with the funneling chute design as well, I’ll likely be doing exactly as you said and instead having two chutes, where the cards don’t have to turn or rotate at all
With cards really liking to stick together, I'd be worried that the "swiping arm" might occasionally move two cards at once, which could cause obvious problems.
Agreed, I haven't added it to the model yet but I plan on having a rubber bumper at the chute edge, which should hopefully help prevent doubles. I'll definitely be doing lots of testing to ensure it's reliable!
I fear you'll have a lot of flipped over and side clinging cards if you're just relying on it falling into the stack neatly. Maybe light spring pressure plates unter the card piles would help reduce the fall distance and prevent flipping. Don't know how accurate those weights would be at that height.
Thanks for the input! A few others have expressed similar concerns and I agree that could be a failure point. I'll be doing some testing on the pile & chute design to ensure it's reliable. If that testing goes south then your solution is a great one!
The dual-actuator arm setup is amazing. Once the machine is complete, and can successfully sort cards by color, would the software driving this thing be customizable? With 6 piles, you essentially have a very easy time coding wise, as you have (and I am dumbing this down, not to be obtuse, but just conversationally) 6 if-then statements. Could say, the colorless pile be configured to catch cards outside of bulk price (>=$3)? Obviously it complicates things in the API department, since color can be retrieved from the Gatherer API, but not market price.
Thank you! Yes I apologize for not making it too clear in the video but I plan on making the piles programmable so that you can sort by a wide variety of metrics. Eg. sort by market price, card type, set...
Not an engineer or even adjacent, but this looks awesome! My only concern is mitigating damage. I don’t imagine cards would take too kindly to being slid on a plastic surface or potentially impacting the sides of the chute. But it may just be a non-issue since if memory serves me right, none of the designs in your last video had these issues. Would some sort of soft fabric lining the chute and gantry be a solution if it’s even a problem? Or even a tighter chute, if that makes any sense.
A few others have expressed similar concern so I'll be redesigning the chute to avoid too much bumping/rotating of the cards. And yes, a soft fabric lining is a great idea there!
I love this concept! However wouldn't the moving tray of sorted cards make for a large area on both sides of the machine that needs to be kept clear of obstructions?
I like the design so far but I'd be a little bit concerned about dinged corners after the cards are removed from the pile and slide into the wall of the chute.
This was also my initial thought. If you go to a 2 chute design will the machine wait to fire the second off ? Or will the gantry be moving quick with all the weight of the sorted cards? Seems like a potential part failure or loss of sorting speed. Following for updates, looks super cool.
@@245ti thanks! and yeah I’ll make sure on the software side that the gantry has enough time to get in position before a card is swiped into the chute!
Bet the cards wont fall into their respective slots cleanly. Might be even trickier than getting the sweeper to consistently pull cards without cards sticking together. Very much looking forward to seeing your finished product. If this works well enough, are you thinking about producing more than one and selling them?
Yeah I'm eager to get testing and work out those potential issue areas in particular! And yes, assuming I can get it working well I will likely launch a kickstarter campaign! I appreciate your input & support!
Id be interested to see if this project could be adapted to smaller card game sorting (yugioh perhaps?) And if that would pose a significant redesign. I'd also be interested if your design only sorts, or if it could scan your collection for cataloging purposes, perhaps in a later software update as a setting
Yes, a few others have asked both of those as well and cataloging will be an easy feature to add, while, like you mention, adding support for card games with different size cards will likely require some redesign work! Definitely on the road map though assuming all goes well!
@@corestyler Did you see where I drew out the circular design in the video? It did have 12 piles, not 6, but it was drawn to scale and it was not very compact. Also I had trouble sourcing cheap turntable components of appropriate size online. Where could I find such parts at a reasonable cost?
Any ideas on how to sort by both alphabetically and by color? I noticed that's one of the things that was mentioned at the 0:17 mark, and I think that would be a really great feature to have. Sorted by color isn't all that helpful on its own, as you still have to look through massive piles to see if you have a card or not. And even if you have the software side log each card, so you can search through your collection quickly, you still have to manually look through massive unorganized color piles to _find_ the card. You could make 2 machines, one to sort by color, one to sort alphabetically. Or you could run your cards through the same machine multiple times with a different algorithm to sort by first letter of the card name. So: First Pass: Color Second Pass: First letter of Card name into 5 piles (with a 6th pile for anything that isn't the first 5 letters of the alphabet) Third Pass: Sort the 6th pile into the next 5 letters of the alphabet. Fourth Pass: Sort the new 6th pile into _the next_ 5 letters of the alphabet. Continue making passes until all cards of a given color are sorted by the first letter of their name. Of course, that means you're making 6 passes for each color, totaling 40-ish passes. (5 colors + Multicolored + Colorless + Lands) So, that's not ideal, but it's the best I can come up with.
Yeah! It wasn't super clear in the video so my apologies but you'll be able to sort however you want! My preference is color but you don't have to do it that way!
Will there be any concern about the cards falling into the piles maintaining their orientation? From the simple chute/pile design it seems like there might be a risk of the card flipping over as it falls into the pile, especially with the first portion when they have a longer distance to fall. Since the end looks to be open for ease of accessing the card, there could be a possibility of cards actually falling out of the sorting piles if something like this were to occur.
Nice project, and good ideas - here are my 2 cents: Edit: I just read a few of the comments under the video - most of the things I wrote are already said in them, sorry for that, I didn't read them before I wrote my comment... You said you like to sort by color, thats why you use 6 piles. That's white, red, blue, black, green, and colorless. What pile do you want to use for multi-colored cards? Or for cards that are for some reason not identified/scan-able? I know you said it is easy to extend the amount of piles to sort into, but you can only go so far till you need to reconstruct the lower part of your machine to accommodate more sorting piles and the mechanism that moves it. Regarding the moving arm: part of me thinks: that's a good idea. Sadly, another part of me knows, that magic cards like to stick together - that's the reason most sorters pick them up and bend them slightly to drop off cards sticking to it. The aluminum extrusions you use: you might want to look at 30x30 instead of 20x20, they are far more stable - it stands on a table, so there is a high chance that people bump against the table, or even the sorter itself, with a small Profil like 20x20 that might become a problem, and 30x30 is not that much more expensive... Regarding the speed you want: in your other video you said you want to be able to sleep in the same room - while you are sleeping, you can't refill the machine, so either you can reduce the speed (maybe around 400-500 cards per hour) or you drastically increase the card feed size to hold maybe 3000 cards each, that would allow it to run for the entire night. To hold a pile that size, you might have to add guide rails (or linear rails) on the side of the plate the cards sit on. Software. If you only plan to use it yourself, hey, no need to go overboard. But if you might want to extend your collection of cards, or plan to maybe sell the machine to others, you have to look into cards with extended or altered arts, cards in other languages, promotion cards, and so on... - you will need an entire database (that's also permanently up-to-date) just to identify the cards correctly. (you might be able to do something like that with a picture-identification tool like the one used in Googles search engine, but I don't know how fast that is...) Enough criticism from my side - hope you don't take it bad... As I said, it is an interesting project, and I would really like to see it work someday - good luck! (P.s: if you plan to sell those machines some day, count me as definitely interested...)
Wow thank you for such a well thought out response! A few other people have mentioned multicolor cards as well, I didn't do a great job conveying this in the video but my plan was to lump those all into my colorless pile, then just run that pile through the machine again to differentiate from there. For the swiper arm, I agree sticky cards are of concern, but I'll be adding a lower rubber bumper to hopefully combat the issue. As for the stability, good idea, if the machine is too flimsy or unstable I'll go ahead and upgrade to wider stock. And good memory! I do plan on having a sleep mode on the software side, as the motors running full speed will likely be noisy. That mode would just slow things down that way you don't have to refill it and it will stay quiet. Speaking on the software side, I did put out a few videos earlier on some initial card recognition code if you'd like to check those out! Again I appreciate your interest and feedback!
Just wondering if you are looking for a partial backer on this project? Very interested in building my own, and yours is the most affordable/reliable idea I have been able to find.
Thanks for the input! Agreed there's potential for card damage there and that's an area I'll be reworking. And I didn't model it for the video but I plan to have a lower rubber bumper to help prevent any double cards!
I think it's possible to just add another bucket for known DFCs to just sort manually. DFCs are relatively rare and all non-token DFCs will be clearly labeled as such. With the suggestion to add a pile for multicolor and lands, that's a total of ten different boxes, which seems like a lot to me. Probably easier to just sort into the six piles in the original spec and then re-sort the sixth pile into 1) Multicolor cards containing white 2) Multicolor cards not containing white 3) Non-artifact colorless cards 4) Colorless artifacts 5) Tokens, tip cards, etc. 6) DFCs (I may be missing a few piles.) This requires more human input once the machine is built, but saves a lot of time and money engineering for a large number of buckets that won't necessarily be similar weights. If a poker card somehow wound up in your box, it might be a good idea to figure out what to do with it. I guess it's most similar to token, so maybe you sort it into pile 6, then into tokens and remove it from the collection manually when finished.
Thanks for the input and thoughts! You're right that DFCs make up a pretty small portion of printed cards, and you nailed the process for more fine-tuned sorting. I'll try to go through what I'm envisioning for sorting more clearly in my next video, again thanks for the comment!
I really enjoy this series, do you have a patreon to help with possible costs? it would really be cool to release the blueprints / mock up after crowdfunding.
I'm glad to hear you're enjoying the content and I really appreciate the generosity. I don't have a patreon or anything like that but I do plan on launching the project on Kickstarter if my prototyping and testing goes well!
Hello! This is really interesting, but one thing has been bugging me. From what I've seen in the video, this seems like more of a classifier than a sorter. How does it sort the cards into a specified order, to where the same cards are next to one another in the stack? For example, a Llanowar Elves from Eternal Masters being separate from one from Dominaria (based on their IDs), or if you're sorting alphabetically, every Llanowar Elves being next to one another (maybe with a secondary sort of their IDs). Or, is that outside the scope of sorting machines? Either way, I think the engineering problems you're trying to solve are really great, and you have amazing solutions! I really hope that everything comes together for you!
Thank you! And true this machine is more of a classifier, but you can achieve true sorting by just running the each pile through the machine again. Let's say the first pass sorts your cards into pile A-E, pile F-K ... Just pass the A-E pile through again to get piles of cards starting with just A, just B, just C and so on. It would take a while but you could truly alphabetize this way if you wanted to!
@@jackbaumgartel Hello Jack, thank you for the response! Do you think there may be any way, such as an optional extension to the machine, to accomplish this automatically? I could very well see it storing the values on the initial run, automatically reloading the piles into the machine via some mechanism, then using the empty piles to sort? I think there are some algorithms which do such a thing. Bucket sort could be a starting point. However, this would also introduce several issues, such as severely increased cost and mechanical complexity. I feel though that this could be an interesting future extension for the machine, and may be worth some thought
@@FoxLivestreams Perhaps! I know the Cardbot, instead of focussing on physically sorting the cards, just digitally notes the location of each card within the piles. That may be a good alternative as you can still easily find cards by querying the digital database. But yeah assuming the project goes well an extension like you mention here could be a good add-on.
I understand foil detection, but what's the advantage of dual-sided card scanning? I thought all double-sided cards had both sides unique, so you can still identify them with only one.
@@jackbaumgartel Maybe they were introduced after I stopped playing. I guess I shouldn't have assumed there weren't other card types. Thanks for the response. I'm looking forward to your next videos!
@@LolaliciousSmiley They're pretty few an far between, I didn't know they existed until starting this project! And of course, I appreciate your support!
I'd suggest maybe adding 1 or 2 more piles for multicolor cards and/or Lands. It's a good design either way though.
Adding onto this - if it can't recognize a foil or any card for that matter an "undetected" pile might be a good idea instead of potentially holding up the machine waiting for user input.
@@valkaerion I like the Misc Pile Option as it feels like a must, especially if it encounters a poke'mon card, MTG Art Card, ect..
Appreciate the feedback! In my head the 'colorless' pile shown would be a catch-all for multicoloreds, lands etc, but I see the value in another pile.
@@valkaerion Yes, handling unrecognized/error cards is definitely a necessary consideration. I appreciate the feedback and input!
@@damontarloth3774 Agreed, thanks for the input!
Glad to see these videos taking off. And VERY glad the algorithm recommended this channel. Love MTG and love engineering. Literally, two of my favorite things rolled into one.
Heck yeah, thanks for the support! If you can't tell those are two of my favorite things as well haha
I'm interested to see how well the card-swiping mechanism works. I think there's a reason all the commercial systems use a suction cup and bend the cards slightly after picking them up - the cards really like to stick together.
Maybe having another rubber component (some sort of lip for the cards to pass over) touching the underside of the cards could keep any extra cards from sliding in with the intended card?
@@paulkelemen4199 yeah that’s the kind of thing you’d have to do. Perhaps even a set of constant turning rollers pushing cards back onto the stack from the underside, so that only the card being directly grabbed by the sweeper can overcome it
Especially when you stack them 1000 high
Thanks for the input, you may be right! I'm excited as well to see how the mechanism fares in some real world testing.
@@paulkelemen4199 Thanks for the input! That is something I had planned to add but I was too lazy to model it before the video. Great idea!
If you're sorting by color, you may want multicolor piles and lands. Also tokens/other pile for anything that is either not applicable to those 8 or as an unknown pile when confidence is below 99 or 98%
Thanks for the input, you're very right there needs to be an overflow/unsorted pile as well.
Hi Jack, thanks for sharing this info about your process, I'm very invested in this! I would love the ability to sort bulk with a machine that I made myself or with a cheap machine available commercially. You have a background in engineering so you may have already considered this, but I'd like to recommend you take a look at the Voron 3d printer project. It's an open source 3d printer with a focus on being as 3d printable as possible, and I think it would be great inspiration for this project!
Keep up the great work!
Thanks for the support and info Jacob! The Voron project is super cool and a great example, I appreciate you sharing! I'll definitely take a closer look at them for some inspiration.
I'm concerned the convergent chute design is going to ding the corners of cards. Why not go with 2 straight chutes since you are already motorizing the end stacks. Just program it to alternate between the 2 chutes as required.
Or a card will get turned and jam up the chute, as 2
chutes would fix this.
Brilliant! I can't believe this never occurred to me. Thank you for the suggestion!
@@damontarloth3774 You are very right, thanks for the suggestion
Good luck on the project, I'm looking forward to seeing your process.
Thank you, I appreciate the support!
This is currently one of my favorite projects on youtube. Thanks for sharing.
Awe glad to hear it! Thank you for the support!
You can make a suction device just as easily as the flicking mechanism by using a big plastic syringe and a a servo to push or pull the plunger. Flicking the cards has potential to ding up the edges, scratch against the aluminum extrusion, or just flat out complicate the parallel processing because now you've got to deal with stacking tolerances between cards.
You're right I could engineer a pretty cheap suction style device, but then it has to move and deposit the card into a pile, which is another set of motors. It's certainly a idea worth keeping in the back of my mind, I appreciate the input!
I have been mulling over the idea of making one of these for a while now! Really excited to see your progress!!
That's the problem with "mulling things over", is that it's just kicking the can down the road. If you have an idea, work on it before someone else makes it.
Great to hear, thanks for the support!
I foresee the biggest issue in the design is the chute and swiper. Cards will 1) get jammed in the chute sideways. 2) Not singulate properly, as they may bend a little when coming up the stack 3) May not have enough velocity to slide into the hoppers. Super cool design, though. I was thinking of making the same thing!
Thanks for the feedback, and yeah I've realized the chutes style needs some improvement
@@jackbaumgartel Came here just to say this. I've already tried doing a similar design with chute, and all three problems were present. The suction + slight bend to suctioned card fixes the problem with picking up exactly 1 card. Chutes + stacks + relying on gravity always produce inconsistent results. Placing the cards with a gantry-style placer is the best way to go.
You can see my attempts to do this (albeit with LEGO Mindstorms) on my channel. I did also try the gantry-style suction picker (also with LEGO Mindstorms) but at the time I couldn't get it to reliably place the cards into each pile (LEGO Servos aren't intended for millimeter accuracy) and I couldn't get it to reliably determine which card was which (lighting and camera focus made this incredibly difficult).
I eventually abandoned the idea as I didn't feel that the original design of using something modular, familiar (like LEGO), and easily customizable led to a "good enough" product. I like what I see here so I wish you luck!
No rush man, you clearly put a lot of time into this!
Thank you, I appreciate it.
Can't wait to see this project come to fruition. Bell icon clicked.
Great to hear, thanks for the support!
this is so cool man. I cant wait to see how it all goes
Thank you, I appreciate it!
Regarding number of piles, I sort my cards based on set then collector number so ideally would like the machine to be able to sort through a complete set in no more than 3 passes. Given that sets normally have
Thanks for the input! I'm not sure if you've seen the video I did about my software plans for the project yet, but I've linked it here if you're interested! th-cam.com/video/T63tQijJzC4/w-d-xo.html
@@jackbaumgartel Yes, I have. Thank you!
Love the idea of this and look forward to seeing how this proceeds! One things I'd suggest though, as someone currently sorting through their bulk by hand, would be to not worry so much about that sorting speed. As long as it's relatively quiet and safe to leave running, having something that can just keep cranking away when I'm at work or sleeping would be awesome. Compared to the established machines that are meant more for game stores where time is money for me just being able to better use the time I can't currently matters far more than efficiency.
Thanks for the input and perspective, I appreciate it!
I've been wanting one of these since to exist for YEARS
Haha hopefully I can deliver one for you!
you could skip the card merger and just use 2 seperate slides if you make the sorter slightly larger
i dont think a cheap motor will be able to move the sorted card piles fast enough to get 1card/s, too much force needed for the acceleration
Great idea! 2 Separate slides is a much better solution. And did you get a chance to see the math I laid out in the video description? I ran some crude calculations to show that a 42Ncm Nema 17 stepper can easily whip around 8lbs of cards. (Unless I made an error in the calculations haha)
Keen to see where this goes, will definitely build my own to tinker with if you need testers. Thankfully for hobbyists it doesn't need the degree of accuracy and resiliency a card shop needs, as long as there's no vectors for damaging cards we can run it through and diff the catalogue to make sure it's correct.
I think the commercial machines use a suction mechanism because singularly moving cards is the crux of the design. Any system of legs/wheels/belts trying to slide the top card is going to have a chance of taking the card beneath it, and unless your steppers are very accurate and there's a "ledge" to keep the stack trapped it'll run into issues. That said, stepper motors are plenty accurate for 3D printing so maybe it is viable...
Best of luck, keen to hear more and participate.
Thanks for the input and advice! I appreciate the support
Just subscribed from your last video! Good to see a follow up so soon.
Haha great to hear, thanks for the support!
Custom sort criteria would be a really good extension option, once you get a functional system (gotta avoid that feature creep).
Could get to a very high level of specificity-eg: pull out every "goblin" creature type card-by planning multiple run sorts, too. Like in your present design, a colorless pile could be fed through again, to sort further into lands, multicolored, etc.
I appreciate the mention of feature creep haha Once I've got things up and functional custom sorting metrics will definitely be a big priority for me!
@@jackbaumgartel I don't know how do you intend to program it or what, but it would be ideal if the program expects a "sorting function", that receives all the data obtained from scanning the card (name, cost, power, toughness, types, etc), and returns a "number" to indicate which pile should go to. With this, a user could design its own algorithm for sorting with whatever criteria.
@@Dsiefus Yep! The plan is to allow for a whole range of custom sorting functionality within the software.
If you are going for a lower price point, then this is made to be more for consumer markets. If that's the case, I would say processing time is somewhat irrelevant and what's more important is the cost to purchase, energy costs, and noise. If it's cheap to buy and run and it's quiet I can just have it running the background over many days. A store has to process a lot of cards quickly a household person with a back pile of lots bulk has plenty of time.
Thanks for the input. A few others have expressed similar views and I appreciate the perspective. I'll be prioritizing noise more than I showed in the video.
@@jackbaumgartel I am so excited I just can't hide it! Hope things are going well with the project
@@RocketRaven Great to hear, thanks for the support! And yes I'll be publishing the next update video shortly!
If you go for 8 piles it may save space if the mixed pile is in the middle with the 8 sorted piles surround it in a ring. This way either the mixed pile or the ring of piles could rotate to the required position and deposit the card. This idea is in response to the potential that the chute clogging and/or how overly long 8 piles in a row would be on your desk
Thanks for the input and I appreciate you reading through some of the other responses! You're right the rotating style of piles would likely save desk space compared to 8 linear piles.
Don't forget that you can do iterative runs of the sorting process. You could have multiple programmes, one that sorts first by colour and then rarity and then cmc etc
Exactly! You'll be able to sort by whatever metrics you choose using the just the piles provided.
Machine was cool but I like how you climbed over that rock
haha thank you!
This seems like a neat challenge. On the topic of number of piles, I think you lightly touched upon it. A 6 pile design works for sorting all your colors in one go and as you mentioned in the comments, "colorless" being a catch all. An alternative if you want to save desk space at the cost of overall sorting speed is to have the output hopper types programmatically assigned. So say once you do your first run, you switch the software mode and then sort through the stack of catch-all that your first run left. This also would be useful if in the future you change the criteria you sort by. It would just be a matter of updating the detection algorithm to your new criteria and assigning the results to their proper bins.
Either way, I will love seeing how far you go with it. Good luck!
Thanks for the input, I appreciate it! That would definitely be a fun addition to the software feature list, and would save time as you said.
I would add a 7th chute for multicolor (mc) cards; as mc's and colorless cards usually are the least represented cards in sets I would put these piles at the front and back to center the weight
Thanks for the input, a few other people have mentioned another pile as well and I think it's a good idea, centering the weight based on predicted occurrences is a great idea!! I appreciate you sharing!
Looks super cool, i'm very excited to see your building process. Never thought about a sorting machine like that, so seeing all the other options in the other video was also really nice. I'd be interested in seeing how easy it would be to adapt a machine like that to work with different card sizes (notably yu gi oh cards are smaller than magic cards). Subscribed !
Thanks for the support and I'm glad you're enjoying the content! It shouldn't be too hard to adapt the machine for close card sizes as well, but that's a feature I'll tackle after I've got a working prototype.
This looks really neat! I'll have to follow and see where it goes. :)
Glad to hear it, thanks for the support!
This project is so cool! I think tour approach is really professional, and follows good design criteria. I personally think the tricky part will be in the stepper finger, because cards sometime stick to eachother, so maybe a roller/blocker might be needed, imo.
Thank you! And yes I hadn't included it in the model yet but I've got a lower rubber bumper to help prevent those sticky cards from moving as one!
Making the receiving containers modules so you could add 1 at a time seems like a good design idea, I think I would setup a rack and pinion for shifting them left and right making it so you only need to add track and containers to expand.
I don't know enough about these cards to know why you can't pull cards off the pile like printers pull paper, possibly making things even simpler for your design and dual side scanning easier.
Thank you for the input!
I recently sorted my collection and scanned it with the dragonshield app. fron there i later imported the result to manabox. i have like 7 full binders one for each color, mc and lands and about 20 boxes each holding around 1200 cards of bulk. in all my binders cards from all kind of sets are found. i had to scan each card separately. this took me ages to do. once i finished it and reached the bulk filled with tons of cards from limited play i have changed to bulk scanning. for bulk scanning you sort your cards per set and tell the app which set your are going to scan. this greatly improves the scanning since it only has to check 250 possibilities instead of a few thousand. that was also the point where i no longer cared for the color or type of the card. i gave every box a unique letter like a,b,… and subdibided it into 6 segments. each segments holds arround 200 cards. so after scanning 200 cards of lets say ammonketh i marked that list as A1 and put the cards into is corresponsing place. so if i need any card from my bulk boxes i open the app and search in which segment the card is contained. after that i only have to search it in 200 instend off n thousands of cards. i'm now at the point where whenever i visit a limited event and end up with cards from one set i quickly scan them and put them in a new segment.
My tip: don't sort the cards by color or type if you end up putting them into a box with thousand of other cards. make smaller piles sorted by set. much easier to scan and to find the cards later.
Thanks for sharing!
Absolutely signing up for the email list & hitting the bell. While I find organizing cards relaxing, I don't have the time I'd like for it compared to how many cards end up adding to the pile... While I know this device isn't anywhere near completion, I'm very excited to see the process and how it evolves.
Couple of things I might add to the discussion:
- The act of sliding the cards down a chute could cause harm to the edges as it bumps them to align with the slot/pile destination. While this arguably feels like me being overprotective, I would like to see if there's not another solution to prevent that. Maybe two chutes, removing any horizontal movement? Would also prevent the possible case of a card sliding sideways, like if it gets caught on some grime or has some defect with the card on the corner making it "swing out" while sliding down.
- What say we remove the chute entirely and have it deposit pretty much straight into the piles, as the chute seems only used for aligning the two stacks into one output. This could save on costs to remove a custom part entirely. Might even give more control for their final orientation. This does impact the speed as you would have to move the gantry for every card every time. Even if both were greens, the two stacks are never lined up to the same slot. But I mean, with the chute system you had to account for the speed they're falling at or have some way to confirm they're in the slot else the movement of the gantry could seriously bend cards that were halfway in place.
- Is there any planned way to make the final slots lower the pile of cards as it gets filled? I am concerned about overly long drops into the slots causing cards to angle themselves down and get jammed. Basically standing tall on the pile vs flat against it. If there was some way to keep them mostly raised that would be ideal. However then it's a cost issue of more parts per slot. Even a simple spring might work but that would have to be calibrated to maintain a good distance at the start with only a few cards.
- I see people concerned about cards sticking, and yeah that's actually a really valid one. Especially with the taco cards. My first thought was to have the top of the stack be nearly in line with the chute but that doesn't really help with tacos or edge damage. What's the chance of a rubber bar on the bottom of the chute, right against the stack? It could raise up with less force than the swinging arms above and catch cards that might be sticking? Then the swinging arms on top can pull any cards back into place. It would be a balance to make the lower bar able to grab a stuck card while not scraping a non-stuck card however. As well as timing the lower bar to come up after the swinging arm has started, so it doesn't catch the front of the card that we're trying to move.
Thank you for the support and for all of your input, I really appreciate the time you've taken to share your thoughts here. First off a few other people have expressed similar concern over the chute, so I am planning on having two straight chutes instead of funneling the stacks into one! I do think some kind of chute is still a good idea though, as I want the cards to be moving at a more uniform speed when they enter the piles. The swiper will likely have some variation that I want to be smoothed out. I have thought about a moving bottom for the piles, so that cards don't have to fall as far, and that's something I'll be testing on the prototype. Lastly yes I didn't include it in the model but there will be a little rubber bumper to help prevent two cards from entering the cute on a single swipe!
@@jackbaumgartel I tried to see if other people had touched on em. Super happy that it's all been covered and acknowledged! Really looking forwards to this project as a whole and the future updates.
I didn't think about the speed of the cards from the arms being variable. Curious to see a real test build and see what all the parts do in conjunction.
@@Zaezar Great to hear, again I appreciate the support and I too am looking forward to some real world tests!
I think that push arm might be the biggest problem. First, it may get stuck as the end of the rod (where it touches the card) will not move in straight line - it will move in arc so you`ll probably end up overcomplicating the mechanism. Maybe something like printer`s paper release mechanism would be better? Second that finger may be pushing multiple cards at a time if the cards get vacuumed together.
Additionally the card feeder should be at a slight angle back - cards stacked like this, especially a 1000 cards, do not like to stay in one place when touched.
Very elegant design for such a complex problem. Though something I didn't see addressed is the nature of real-world card swiping when faced with sticky, chipped, or curled (even non-foil) cards. One of my concerns about throwing a stack of unsorted bulk into a machine like this is what will happen to the various mechanisms when 2-4 cards are stuck together or when a bent corner catches and jams the whole thing.
Excited to see how this project evolves.
Yeah that's something I'll be diving into with the real world testing! Glad to hear you're excited and I appreciate the support
I feel as if the chute combining the output of both of the feeder stacks is a potential problem where something could jam. Maybe if you just had independent output chutes for each pile it would be better? You certainly wouldn't have to accommodate a rectangular card rotating to slide down the chute and then rotating back to align with the collection pile.
Also, you might want to consider having some safety sensor to ensure that your collection gantry never slides laterally while a card is still partially in the output slide. Potentially an optical sensor at the end of the slide?
I'm stoked to see how this project progresses no matter what though! Good luck!
Yes a few others have suggested similar design changes and I do plan on reworking the convergent chute, it was not the best solution admittedly. I had thought about a sensor there too, but I fear the extra cost & complexity. I think I can ensure that issue doesn't occur with properly timed software commands, but if during testing things go south, an optical sensor is a great idea! Thank you for the input and support, I appreciate it!
TH-cam algorithm suggested this to me, love the project idea and while I'm not a huge retailer or anything something like this appeals to my lazier sensibilities
Great to hear! The reason I'm building this thing is so I don't have to sift through my own bulk so I sympathize with you
I dunno MTG but I did build my own laser cutter, I feel like this could be easily improved. You've done a few rookie mistakes and all I'm going to say is "keep it simple"
Only sort one pile at a time, two piles means two sets of parts and two sets of places of failure and twice the price. Counter weight the steppers lifting plate externally. and that flicking arm is going to be trouble, I can feel it but my only suggestion right now is compressed air and that would mean starting again.
Love the code too btw
Thanks for all the input! I appreciate the wisdom but I may have to learn through these mistakes on my own before I make any changes!
My recommendation is to separate your two piles further and have your grabbers sweep them sideways into the center. It would reduce the travel time for the cards before the next is uncovered, speeding up scanning acquisition, and the chute design could be simplified as a simple ramp without the need for the angled edges to funnel the cards inwards.
There might also be a way to connect the two sweepers into a single motivator with both arms attached, as it could be changed to a synchronized left-right motion of both arms. Of course you may prefer to keep the independent motion if your design allows you to sweep from the same pile twice in a row.
Thanks for the ideas, I intentionally want the sweepers to be independent, but swiping centrally is a pretty interesting thought!
For the problem of raising the unsorted stack to the correct height initially, I think there's a really simple solution. You could slowly raise the stack, while using the camera you already have in the design to see when the edge of the topmost card is in the frame. Once the card reaches a specific height (I.e. a specific spot on the chute frame gets obscured by the pile), you stop the stepper motor.
It would take some testing to make sure the latency from the camera, software, and motor controls doesn't cause the stack to over shoot. And it may be worth marking the inside of the chute, perhaps with some sort of solid colour sticker or tape to make it easier to see where the stack has reached.
Also, you probably won't want to rely on the user making 2 exactly even sized stacks, since that takes a lot of fiddling. So you'd probably want this sort of system anyway, to tell when the system has been fed uneven stacks.
Thanks for the input! That's a pretty clever solution. I like that it doesn't require any additional hardware, just some testing and software adjustments.
Really cool project! Signed up for email, very interested and excited to see how it goes!
Thank you, I appreciate the support!
I'm a new sub. Part 1 came across my feed and I was intrigued. I'm very interested in this. Main thing I'm looking for is an algorithm that will pick my Commander "staples" and separate them first. Than another algorithm to further sort based on card function.
Thanks for the support! And yeah there will definitely be custom sorting features in the software
@@jackbaumgartel also. Imo, don't be afraid of using 3d printed parts of needed.
I'm going to watch this project like a hawk!
Heck yeah great to hear!
Sitting on 500k Yugioh cards and really wanting a Yugioh sorting machine. Hope your progress goes well!
Great to hear, thank you for the support!
If we are sorting one card while scanning another, then maybe it would be better to have two separate shoots instead of one. The difference would just be that the bin motor would need to shift one to the left or right by one depending on shoot being processed. But it‘a probably more reliable and slightly less stressful on the cards.
Thanks for the input! A few other people have expressed similar ideas and I agree it’s a much better solution than mine!
Maybe this is something you've already considered, but rather than hard coding WUBRG sorting into the project, it may be a good idea to set it up to pick a sorting type via software. For example if you have already color sorted and you then want to sort by mana cost or price. If it is not more versatile than color sorting it seems of limited use
Yeah I didn't clarify very well but you'll be able to sort however you'd like in the software! Color was just my example for the video haha
Card sorting is cool, but I would love for a machine that is similar to this in design with the sole purpose of cart cataloging.
Feed Card, scan card, input card into DB, move color pile, rinse repeat.
Thanks for the input, a few other people have expressed a similar wish and I like the idea too! I may have to do a 'base model' that just does cataloging, it would be cheaper to produce so it would be a more affordable version of the full sorting machine.
@@jackbaumgartel Good luck on your journey with this project!
@@alex_mula Thank you!
Having a circular end destination for the cards instead of a line would be harder to switch out sizes but make selecting compartments faster (since it can go all the way around) and also probably more stable as it doesnt have the same inertia of moving in a straight line? Might lead to some issues with the space it takes up however
Agreed, with the current layout, the machine could tip or jitter if the piles move too quickly. With a circular layout, the machine would spin in a similar situation. I think the larger space and higher cost is a disadvantage though for the circle design.
Have you done any testing of your card chute design? Looks like the cards might be free to wander about, and take varying amounts of time to reach their destination
I haven't yet, and thank you for the feedback. Software wise I'll make sure the pile is aligned before the card could possibly exit the chute, and I'll be sure to test the chute thoroughly once I've got a working prototype.
I would love to get one. Kickstart it after you get the bug out
Great to hear, and yes that's the plan!
I graduated Bachelor's comp science last year, and this is inspirational stuff
Thank you, I appreciate the kind words :)
Very exciting stuff! Personally I would prefer an eight pile solution but even if there were three piles it would be useful. As long as affordability is alright then it could be very successful.
Thank you! I do think I’ll actually be adding a pile or two based on the responses here so you may be in luck!
The thing that will slow this down, is you can’t really eject the first card until the proper pile is in place. This means you have to slide the whole set of piles, at worst from one end to the other. Depending on how long that takes, you could have both piles with cards scanned and waiting. You may need to have a better algorithm that is more like en elevator, where it doesn’t just go left pile, right pile, but tries to find the closest pile to one of the two cards and moves to that first.
That's a really cool idea! Assuming both cards are scanned moving the piles based on proximity rather than original order. Also take a peek at some crude math I did out in the video description, a cheap stepper motor should be able to move the (full) piles around pretty quick!
@@jackbaumgartel My initial thought of this is simply adding another servo at the exit of the merger, a simple gate (or even "ensure card ends up in pile cleanly" sweeper style?) such that you can have a one-card deep queue waiting for the bin/pile mechanism to be in the final placement for receive. If you are extra clever, you could make it a N-length queue that just progresses one card at a time once pile/bin is in place but that depends more on mechanism and code complexities.
@@admalledd Good idea! Thanks for sharing
Hey, I did a Design project on designing one of these.
That's awesome! I'd love to hear more about it if you're willing to share!
Love this project. Definitely need a pile for multicolor cards. I would love a cost analysis for this project.
Thank you! And I'll likely share a cost breakdown for the project later on, that's a great idea!
@@jackbaumgartel I think you may be starting something that could become like the old maker-bot challenge.
I would love to see competitive open-source card sorting machines. Especially after you showed us the cost of similar proprietary machines?! The companies can only charge such high rates because they don't have healthy competition. You are singlehandedly the most competition these companies have right now.
As you showed, each machine on the market is astonishingly expensive, and they seem to try to be cordial in their designs. They try to be application specific via each machine to hopefully hit different levels of consumers.
You're making a machine that will be a fraction of the cost of any competitor, and be adjustable to each application. Meaning it would become economically advantageous to build or buy, 4-20+ of your machines instead of ONE competitor machine.
You're going to make alot of consumers very happy, and some companies very angry. I love it all!!!
Thanks again! Keep up the good work! I love this project so much.
@@nickybeingnicky Thank you so much for the kind words, I certainly hope that what I'm doing will prove to be such a disruption!
I have recently bought EPS32 camera modules, I paid 25€ for 2. It should have enough pins left to control all motors. At the highest resolution for the cheapest module it can take a little over 3 pictures per second which should be plenty. 1600x1200 should also be plenty to just find the color of 2 cards.
Wifi control also means you could offload image processing to another computer of some cloud service.
Thanks for the input! That's definitely a good option, I appreciate it
@@jackbaumgartel Further clarification:
The modules should be much cheaper.
I basically bought sets on Amazon that include an USB programmer.
So just the boards with wifi and cam should be available around 4-7€.
Just did a quick check: probably the exact same product is around 8€ on Aliexpress.
I think the way of sorting really depends on further usage: deck building, collecting, trading and even on the format played. In my case (EDH) I have worked this way of sorting: rarity, colour, A-Z where rarity mixes rare&mythic, then uncommon, then common. Non-basic lands are mixed together without rarity dependence, basic lands are separate (treated as bulk). Also, A-Z sorting is just by 1st letter only, because it would be overkill for me to go with full A-Z every time I have to add cards (usualy dozens).
Yep, like I said, everybody sorts differently! The machine will let you choose your own adventure so to speak
@@jackbaumgartel just wanted to share. Happy to see progress of the project.
@@sebastiannowak1981 Glad to hear it, and thank you for sharing!
I'm eagerly waiting to purchase a functional model.
Next project: a small battery operated automatic shuffler for Commander decks across a wide range of sleeve types.
Glad to hear it, and fun idea!
This idea and engineering process walkthrough are both right up my alley, excited to see where this goes.
Out of curiosity, how do you intend to protect the card corners on the chute? or protect the cards when they fall into the piles? I would hate to have all my corners knicked up from the machine or for my cards to fall at an angle and fold from the weigh of more card being put on top of them should I leave it unattended.
Great to hear! And I'm planning on redesigning the chute after some solid feedback in the comments here, the new design should be more gentle on card corners. For the falling into the piles properly, I'm going to see how testing fares in the real world and go from there!
I would say what ever gets rejected can go into colorless tray....would be easy to find or maybe 1 more "others" tray
Yeah agreed! In my head the colorless was a catch all for multicolored cards as well as any rejects
I see a problem with the speed. You have the acces to scan fast and move to next scan fast by double slider finger. But your sorting pile area can not move as fast as you can scan.
I had once a idea of a sorter, where not the storage area move but the slide ramp where the card slide on. The ramp can be mived quickly because it doesnt depend on the weight of the cards already sorted
Did you look at the math I laid out in the video description? Let me know if that changes your mind!
highly interested in the project as it's something I've thought about making aswell
Awesome, glad to hear it!
@@jackbaumgartel do you have a 3D printer? would help a lot with rapid proto typing
@@PresleyHudson919 Yep! And yes it will haha
@@jackbaumgartel awesome, I do aswell, could be awesome if you release the stl files, would definitely drive your price figure down. Maybe even open source it, could really help make this thing competitive with the big brands. Of course I understand if that's not something you'd want to do. I'd definitely be very interested in trying to help out, I'm sure other could be aswell.
@@PresleyHudson919 Thank you for the support, and I have been considering open sourcing the project!
Just a thought, I know there is a lot of the work done for the coding based on the other videos, but what about seeing if somehow you could integrate it with the mtgstand site. It already has a card recognizing script, and uses a webcame or any camera that can stay open.
I don't want the project to rely heavily on any third party processing like you mention, even though it definitely would make things easier for me! However, MTGstand accepts CSV imports, so once the machine scans cards, you'll be able to easily import the collection!
Really great starting point. Can't wait to see where this project goes. I'm a software dev with experience in computer vision. If you need help with the programing for the camera I would love to collaborate!
Thank you, I appreciate the support and I'll certainly reach out if I find myself needing any help!
that arm with the rubber arm can quickly go from soft to scour texture if you cheap out on the rubber.
imagine those pencils with erasers that become stones or charging cables that become stiff. imagine that rubbing across the face of every card.
the orientation shouldn't matter. most AI card scanners I've used can also recognize the image being upside down
Thanks for the input. Yes all rubbers eventually wear out so material selection there will be important. For the orientation, in the video I more meant is the card face up or face down, I agree that software can flip a card rightside up if need be
@@jackbaumgartelWould not be posting about something as insipid as "wear" unless we are talking about what happens to the card when it meets a dry eraser/rubber band.
One can imagine that it would be rough waking up and finding small vertical scratches on an entire stack of cards. Easily bump a card from near-mint to played because of a scratch.
@@evilgibson For sure, I certainly understand the importance of protecting the cards, especially if the machine is to be used by any resellers.
The chute should go to a trap door with 6 gradients, so the piles stay stationary
That's pretty similar to how the Phyzbatch works! A good solution as well.
if the flick method of pushing cards does not work i can show you a cheap way to make suction work. the only downside is the cost in motors more then anything. lastly i think you can easily make it 10 piles with little extra room. you can do a 2x5 config of piles and make the first set go under the shoot to save space. where both of these ideas might be an expansion they might be worth thinking about. o ya are you making this a open project or one you plan to profit from? i would like to build or buy one depending on how it turns out and what your plans are.
One of the other big problems with suction cups is that they are so slow but it might be the only way that actually works
Thanks for the input and ideas! Definitely worth considering and I think you're right they sound like expansions from the current design. I'm not sure if the design will be open source or not yet, that's something I've been considering though, and great to hear!
@@jackbaumgartel if you wanted, you should look at something like the Voron style of open source. basically the code and files and BOM are open and free but people can sell kits. so you could do that to get the power of crowed sourcing and refining but still make some money for your time and effort.
@@GamerWannaB Someone else here mentioned Voron as well! I appreciate the idea and it's definitely something I'm considering
instead of servos and the long arms i think small wheels at the corner closest to the ramp would be easier and break less easily
I’d worry slightly about the cards going down the slide getting caught and causing a jam, though I feel like this can be accounted for
Yeah, I’ll be redesigning the chute system as a few others have expressed similar concern and I agree!
Jack, could you elaborate on why you don't feel comfortable using a friction feeder mechanism? It's technically how the sheets were fed in the manufacturing process in the first place and properly tuned wouldn't cause any damage to cards.
Additionally, I don't think the wipers are a bad idea I just think you might run into issues without some sort of gate or bottom side wiper preventing double feeds.
Edit: also I just thought of this, do you have any idea how you're going to handle the increasing weight of the sorted piles when trying to move them at speed? I guess if you go slow enough?
I agree a properly engineered friction feeder would work great here, but it would necessitate a taller machine if the piles are still to be gravity-fed. That's the main reason I prefer an upward facing stack honestly. You're right about the wipers though, I haven't added it to the model yet but I'll be including a lower bumper/rubber pad to help prevent double feeds. And yes, I did out some crude math in the video's description showing how quickly the motor would theoretically be able to move full piles around! I appreciate the feedback and good questions!
I also like to sort a bulk rare/mythic pile.
Nice, thanks for sharing!
Seeking knowledge of the active machines, and perhaps future suggestions.
Which of these can alphabetize stacks of cards? Which of these is best at sorting by set? When a mixed box of bulk comes in, is there a better machine for that, or do we still need excessive manual sorting first?
Thanks.
While I don't have access to any machines other than my own, I think it's safe to say they are each capable of what you're asking. I went through each machine's capabilities in detail here if you are interested: th-cam.com/video/sy7dtW8CvQ4/w-d-xo.html
Couldn't you also as a second shoot on the back of the machine to another gantry and 6 more slots. you could make it where the gantry is easy to detach ignorer to load cards. Then you could be sorting two cards at once cause lets say you have a red and a blue at the top it could push the red backwards to go in that slot and then the blue forward then you would not have to wait for the gantry to move back and fourth.
Unfortunately you need a good program that can set up customized organization which means lots of slot and piles. If I could I would organize everything down from besides what the card such as control, blocker, whip, fog, etc with sub cats for the abilities. Ex a hydra that a mana dork but also makes tokens
Oh boy you sort by specific card use? That's the first I've heard of such a method. How do you handle cards that belong in multiple categories? I'd love to hear more about how this organizational strategy!
I can see how this would do grouping easily (UWBRG), but what about actual sorting (A-Z)?
Thanks for the question, it would just take a few passes through the machine! 1st pass perhaps sorts all cards A-E into pile 1, all cards E-I into pile 2 etc. Then you can put those piles back in afterwards to fully alphabetize them!
Personally, I believe the number of piles isn't much of a concern as long as the device is easy to load. Through software, the machine can sub-divide piles with multiple run-throughs. For instance: Let's say you're trying to sort by Converted Mana Cost but you only have 5 piles. The machine would be able to sort these cards with 3 run-throughs. The first run would separate the cards into two piles. [0-4] cmc and [5-9+] cmc. then, put the [0-4] cmc pile back in, then reassign the sorting parameters to 0,1,2,3,4 cmc (One for each pile). Then you do the same thing with the other pile.
I guess this is a long-winded way of saying if you change your mind about the amount of piles late into the design process, then it's no problem, just subdivide!
This design looks really good, but I have doubts about the reliability of a chute to funnel the 2 card piles into one. This seems very prone to possible jams due to inconsistent sliding of the cards. Perhaps you could have each card go straight down a path, and simply reposition the output piles on the rails to correspond to whether it is receiving from the left or right input pile.
Thanks for the response! You’re certainly right about subdivision and the number of piles, the more piles you have the less you’ll have to subdivide. And yes a few others have expressed concern with the funneling chute design as well, I’ll likely be doing exactly as you said and instead having two chutes, where the cards don’t have to turn or rotate at all
With cards really liking to stick together, I'd be worried that the "swiping arm" might occasionally move two cards at once, which could cause obvious problems.
Agreed, I haven't added it to the model yet but I plan on having a rubber bumper at the chute edge, which should hopefully help prevent doubles. I'll definitely be doing lots of testing to ensure it's reliable!
Awesome project!
Thank you, glad you like it!
I fear you'll have a lot of flipped over and side clinging cards if you're just relying on it falling into the stack neatly. Maybe light spring pressure plates unter the card piles would help reduce the fall distance and prevent flipping. Don't know how accurate those weights would be at that height.
Thanks for the input! A few others have expressed similar concerns and I agree that could be a failure point. I'll be doing some testing on the pile & chute design to ensure it's reliable. If that testing goes south then your solution is a great one!
The dual-actuator arm setup is amazing.
Once the machine is complete, and can successfully sort cards by color, would the software driving this thing be customizable? With 6 piles, you essentially have a very easy time coding wise, as you have (and I am dumbing this down, not to be obtuse, but just conversationally) 6 if-then statements. Could say, the colorless pile be configured to catch cards outside of bulk price (>=$3)?
Obviously it complicates things in the API department, since color can be retrieved from the Gatherer API, but not market price.
Thank you! Yes I apologize for not making it too clear in the video but I plan on making the piles programmable so that you can sort by a wide variety of metrics. Eg. sort by market price, card type, set...
Not an engineer or even adjacent, but this looks awesome! My only concern is mitigating damage. I don’t imagine cards would take too kindly to being slid on a plastic surface or potentially impacting the sides of the chute. But it may just be a non-issue since if memory serves me right, none of the designs in your last video had these issues.
Would some sort of soft fabric lining the chute and gantry be a solution if it’s even a problem? Or even a tighter chute, if that makes any sense.
A few others have expressed similar concern so I'll be redesigning the chute to avoid too much bumping/rotating of the cards. And yes, a soft fabric lining is a great idea there!
Without a way to set sort and alphabet sort, LGS and card sellers won’t buy it. I’m staying tuned for this project though. Always exciting
I love this concept! However wouldn't the moving tray of sorted cards make for a large area on both sides of the machine that needs to be kept clear of obstructions?
Yes definitely, the last thing I'd want is the machine to knock over a cup or smash into another desktop device.
I like the design so far but I'd be a little bit concerned about dinged corners after the cards are removed from the pile and slide into the wall of the chute.
Thank you, and yes a few people have expressed similar concern and I agree! I'll probably be switching to two individual chutes instead.
This was also my initial thought. If you go to a 2 chute design will the machine wait to fire the second off ? Or will the gantry be moving quick with all the weight of the sorted cards? Seems like a potential part failure or loss of sorting speed. Following for updates, looks super cool.
@@245ti thanks! and yeah I’ll make sure on the software side that the gantry has enough time to get in position before a card is swiped into the chute!
Can't wait to see the progression! Super cool project.
@@245ti Thank you! I appreciate it :)
Bet the cards wont fall into their respective slots cleanly. Might be even trickier than getting the sweeper to consistently pull cards without cards sticking together.
Very much looking forward to seeing your finished product. If this works well enough, are you thinking about producing more than one and selling them?
Yeah I'm eager to get testing and work out those potential issue areas in particular! And yes, assuming I can get it working well I will likely launch a kickstarter campaign! I appreciate your input & support!
5 for colors, 1 for colorless/artifacts, 1 for gold cards, 1 and lands and 1 for tokens. So 9, cant think of a lower number than that.
Haha 6 is a number lower than that!
Id be interested to see if this project could be adapted to smaller card game sorting (yugioh perhaps?) And if that would pose a significant redesign. I'd also be interested if your design only sorts, or if it could scan your collection for cataloging purposes, perhaps in a later software update as a setting
Yes, a few others have asked both of those as well and cataloging will be an easy feature to add, while, like you mention, adding support for card games with different size cards will likely require some redesign work! Definitely on the road map though assuming all goes well!
I would suggest a rotating pile sorter :-)
Haha you want all the piles! Thanks for the input
@@jackbaumgartel nah keep your piles but adding a turntable with 6 stashes would be easier to build, is way more compact and might be faster 😊
@@corestyler Did you see where I drew out the circular design in the video? It did have 12 piles, not 6, but it was drawn to scale and it was not very compact. Also I had trouble sourcing cheap turntable components of appropriate size online. Where could I find such parts at a reasonable cost?
Any ideas on how to sort by both alphabetically and by color?
I noticed that's one of the things that was mentioned at the 0:17 mark, and I think that would be a really great feature to have.
Sorted by color isn't all that helpful on its own, as you still have to look through massive piles to see if you have a card or not. And even if you have the software side log each card, so you can search through your collection quickly, you still have to manually look through massive unorganized color piles to _find_ the card.
You could make 2 machines, one to sort by color, one to sort alphabetically. Or you could run your cards through the same machine multiple times with a different algorithm to sort by first letter of the card name.
So:
First Pass: Color
Second Pass: First letter of Card name into 5 piles (with a 6th pile for anything that isn't the first 5 letters of the alphabet)
Third Pass: Sort the 6th pile into the next 5 letters of the alphabet.
Fourth Pass: Sort the new 6th pile into _the next_ 5 letters of the alphabet.
Continue making passes until all cards of a given color are sorted by the first letter of their name. Of course, that means you're making 6 passes for each color, totaling 40-ish passes. (5 colors + Multicolored + Colorless + Lands) So, that's not ideal, but it's the best I can come up with.
Yeah! It wasn't super clear in the video so my apologies but you'll be able to sort however you want! My preference is color but you don't have to do it that way!
Will there be any concern about the cards falling into the piles maintaining their orientation? From the simple chute/pile design it seems like there might be a risk of the card flipping over as it falls into the pile, especially with the first portion when they have a longer distance to fall. Since the end looks to be open for ease of accessing the card, there could be a possibility of cards actually falling out of the sorting piles if something like this were to occur.
Thanks for the input! Yeah there's some concern there and it's something I'll be testing out and adjusting as needed with the prototype
Nice project, and good ideas - here are my 2 cents:
Edit: I just read a few of the comments under the video - most of the things I wrote are already said in them, sorry for that, I didn't read them before I wrote my comment...
You said you like to sort by color, thats why you use 6 piles. That's white, red, blue, black, green, and colorless. What pile do you want to use for multi-colored cards? Or for cards that are for some reason not identified/scan-able? I know you said it is easy to extend the amount of piles to sort into, but you can only go so far till you need to reconstruct the lower part of your machine to accommodate more sorting piles and the mechanism that moves it.
Regarding the moving arm: part of me thinks: that's a good idea. Sadly, another part of me knows, that magic cards like to stick together - that's the reason most sorters pick them up and bend them slightly to drop off cards sticking to it.
The aluminum extrusions you use: you might want to look at 30x30 instead of 20x20, they are far more stable - it stands on a table, so there is a high chance that people bump against the table, or even the sorter itself, with a small Profil like 20x20 that might become a problem, and 30x30 is not that much more expensive...
Regarding the speed you want: in your other video you said you want to be able to sleep in the same room - while you are sleeping, you can't refill the machine, so either you can reduce the speed (maybe around 400-500 cards per hour) or you drastically increase the card feed size to hold maybe 3000 cards each, that would allow it to run for the entire night. To hold a pile that size, you might have to add guide rails (or linear rails) on the side of the plate the cards sit on.
Software. If you only plan to use it yourself, hey, no need to go overboard. But if you might want to extend your collection of cards, or plan to maybe sell the machine to others, you have to look into cards with extended or altered arts, cards in other languages, promotion cards, and so on... - you will need an entire database (that's also permanently up-to-date) just to identify the cards correctly. (you might be able to do something like that with a picture-identification tool like the one used in Googles search engine, but I don't know how fast that is...)
Enough criticism from my side - hope you don't take it bad...
As I said, it is an interesting project, and I would really like to see it work someday - good luck!
(P.s: if you plan to sell those machines some day, count me as definitely interested...)
Wow thank you for such a well thought out response! A few other people have mentioned multicolor cards as well, I didn't do a great job conveying this in the video but my plan was to lump those all into my colorless pile, then just run that pile through the machine again to differentiate from there.
For the swiper arm, I agree sticky cards are of concern, but I'll be adding a lower rubber bumper to hopefully combat the issue.
As for the stability, good idea, if the machine is too flimsy or unstable I'll go ahead and upgrade to wider stock.
And good memory! I do plan on having a sleep mode on the software side, as the motors running full speed will likely be noisy. That mode would just slow things down that way you don't have to refill it and it will stay quiet.
Speaking on the software side, I did put out a few videos earlier on some initial card recognition code if you'd like to check those out!
Again I appreciate your interest and feedback!
Just wondering if you are looking for a partial backer on this project? Very interested in building my own, and yours is the most affordable/reliable idea I have been able to find.
I appreciate the offer! I don't have an immediate need sadly but I'll be sure to keep your name in mind later down the road!
The flicker and hitting the side of the chute are potentially damaging. Also, like inkjet printers, how do you deal with 2 being pulled at once?
Thanks for the input! Agreed there's potential for card damage there and that's an area I'll be reworking. And I didn't model it for the video but I plan to have a lower rubber bumper to help prevent any double cards!
I think it's possible to just add another bucket for known DFCs to just sort manually. DFCs are relatively rare and all non-token DFCs will be clearly labeled as such. With the suggestion to add a pile for multicolor and lands, that's a total of ten different boxes, which seems like a lot to me. Probably easier to just sort into the six piles in the original spec and then re-sort the sixth pile into
1) Multicolor cards containing white
2) Multicolor cards not containing white
3) Non-artifact colorless cards
4) Colorless artifacts
5) Tokens, tip cards, etc.
6) DFCs
(I may be missing a few piles.)
This requires more human input once the machine is built, but saves a lot of time and money engineering for a large number of buckets that won't necessarily be similar weights.
If a poker card somehow wound up in your box, it might be a good idea to figure out what to do with it. I guess it's most similar to token, so maybe you sort it into pile 6, then into tokens and remove it from the collection manually when finished.
Thanks for the input and thoughts! You're right that DFCs make up a pretty small portion of printed cards, and you nailed the process for more fine-tuned sorting. I'll try to go through what I'm envisioning for sorting more clearly in my next video, again thanks for the comment!
@@jackbaumgartel No problem, I'm looking forward to your next video!!
I really enjoy this series, do you have a patreon to help with possible costs?
it would really be cool to release the blueprints / mock up after crowdfunding.
I'm glad to hear you're enjoying the content and I really appreciate the generosity. I don't have a patreon or anything like that but I do plan on launching the project on Kickstarter if my prototyping and testing goes well!
What about card friction from the bottom of the card, couldnt you be sliding more than one card of the pile into the slide?
Yep! Definitely something I'll be testing once I get a prototype up and running
Hello! This is really interesting, but one thing has been bugging me. From what I've seen in the video, this seems like more of a classifier than a sorter. How does it sort the cards into a specified order, to where the same cards are next to one another in the stack? For example, a Llanowar Elves from Eternal Masters being separate from one from Dominaria (based on their IDs), or if you're sorting alphabetically, every Llanowar Elves being next to one another (maybe with a secondary sort of their IDs). Or, is that outside the scope of sorting machines? Either way, I think the engineering problems you're trying to solve are really great, and you have amazing solutions! I really hope that everything comes together for you!
Thank you! And true this machine is more of a classifier, but you can achieve true sorting by just running the each pile through the machine again. Let's say the first pass sorts your cards into pile A-E, pile F-K ... Just pass the A-E pile through again to get piles of cards starting with just A, just B, just C and so on. It would take a while but you could truly alphabetize this way if you wanted to!
@@jackbaumgartel Hello Jack, thank you for the response! Do you think there may be any way, such as an optional extension to the machine, to accomplish this automatically? I could very well see it storing the values on the initial run, automatically reloading the piles into the machine via some mechanism, then using the empty piles to sort? I think there are some algorithms which do such a thing. Bucket sort could be a starting point. However, this would also introduce several issues, such as severely increased cost and mechanical complexity. I feel though that this could be an interesting future extension for the machine, and may be worth some thought
@@FoxLivestreams Perhaps! I know the Cardbot, instead of focussing on physically sorting the cards, just digitally notes the location of each card within the piles. That may be a good alternative as you can still easily find cards by querying the digital database. But yeah assuming the project goes well an extension like you mention here could be a good add-on.
I understand foil detection, but what's the advantage of dual-sided card scanning? I thought all double-sided cards had both sides unique, so you can still identify them with only one.
There are just a few cards that have identical front faces and different rear faces. Cat-dog vs weird-dog tokens being a good example!
@@jackbaumgartel Maybe they were introduced after I stopped playing. I guess I shouldn't have assumed there weren't other card types.
Thanks for the response. I'm looking forward to your next videos!
@@LolaliciousSmiley They're pretty few an far between, I didn't know they existed until starting this project! And of course, I appreciate your support!
this would be perfect to scan and sort by price. like bulk, 50ct ,1$, 5$ and so on
Yep! You'll be able to choose your own sorting metrics!
How do you prevent the sorted cards from not twisting as they fall down the slot?
They might! That's something I'll be putting to the test as I get building one. If need be I'll come up with a better solution