Looking for a partner to develop AI print approval nd monitoring software

[blue moon]

As the title says, am looking into making software that would us AI to approve the print (based on the master file) rather than having the art department or somebody else manually approve. For us this would save several minutes of production (printers currently take the first print to the art dpt where they have to find the art and give a go ahead. Combined with walking it would save us several minutes per job which adds up when running multiple jobs on multiple presses each day).
The system would have a camera and would automatically check for missing parts of the print and confirm the colors are within specified tolerance.
It would then monitor for color drift and pin holes or other issues. Additionally it would count the garments to provide the confirmation it has all been printed.
Total cost for the software would probably be in the $10k range. Ideally we can split this three ways (find two partners). On top of software, each station would require a camera (about $300) and a computer that can handle 3 cameras/presses (probably pretty close to $2k).
There is a good chance this could be spun off into a product for sale after some testing.
Thoughts?
Pierre

[T Shirt Farmer]

Great idea. Imo by the time you hit 2 years development and making it sellable the deocrating industry will be blazing forward leaving tradtional screening behind. look back at all the changes in the last 2 years, probaly a good forcast of the next 2 years. Curious, could the camera filter out random dots in a heather shirt without thinking it is a pinhole?

[blue moon]

Great idea. Imo by the time you hit 2 years development and making it sellable the deocrating industry will be blazing forward leaving tradtional screening behind. look back at all the changes in the last 2 years, probaly a good forcast of the next 2 years. Curious, could the camera filter out random dots in a heather shirt without thinking it is a pinhole?

I think we can have alpha testing in 60 days, beta in 90. Shoukd be ready by spring with bugs worked out. It would be version 1.0, but we can keep developing.

[Admiral]

Great idea. Imo by the time you hit 2 years development and making it sellable the deocrating industry will be blazing forward leaving tradtional screening behind. look back at all the changes in the last 2 years, probaly a good forcast of the next 2 years. Curious, could the camera filter out random dots in a heather shirt without thinking it is a pinhole?

Camera settings should allow the reflection of ink vs shirt fibers to look very different.  I believe it would be easy actually.

We were thinking of doing something like this a few years ago but it's a very big set up and upkeep and it was going to be $50K per press at the time.  I'm definitely interested but couldn't do the upfront cost and wait for it to develop into what it needs to be, that's just too expensive still.  We don't have the time to R&D stuff right now but could be interested next year, after we see how this year goes.

[rusty]

Wasn't M&R testing an on press system that would catch mistakes or variations on press? Couldn't this be a variation of that?

[ebscreen]

Not to be a debbie downer but I don't see this ever actually working well, and I also think it's kind of important to have human brains in the process.
My guess would be if they aren't using machine vision in the offset world already (which would be 1000 times easier to implement) than we won't ever see it in ours.
Like a lot of things.

[GraphicDisorder]

Not to be a debbie downer but I don't see this ever actually working well, and I also think it's kind of important to have human brains in the process.
My guess would be if they aren't using machine vision in the offset world already (which would be 1000 times easier to implement) than we won't ever see it in ours.
Like a lot of things.

^^^^ This.

Also keep in mind in general this industry is ULTRA slow to adopt anything. Think about how many are still on manual presses, still have flo bulb exposure and still wash screens in a kitchen sink. Its wild. More of those by far than modern shops.

[Homer]

what about a tablet on press or end of the dryer? or even a webcam. have a stand alone platen with camera overhead, I know colors shift on screen but might close? We use webcams here for a few things, mostly inappropriate though.... Or a pneumatic conveyor like the bank teller, roll the shirt up, stuff it in the tube and send it!..... AI sounds neat and inspector gadgety and all, but sometimes multiple eyes on cannot be replaced.

[blue moon]

Not to be a debbie downer but I don't see this ever actually working well, and I also think it's kind of important to have human brains in the process.
My guess would be if they aren't using machine vision in the offset world already (which would be 1000 times easier to implement) than we won't ever see it in ours.
Like a lot of things.

^^^^ This.

Also keep in mind in general this industry is ULTRA slow to adopt anything. Think about how many are still on manual presses, still have flo bulb exposure and still wash screens in a kitchen sink. Its wild. More of those by far than modern shops.

This is not for everybody! There are two ppl already interested on some level. If we can get 3-4 to band together, we can make this work.
As far as making it happen, we dont pay until we have results. AI and machine learning has come a long way in the last couple of years and based on what I know, it SHOULD be possible.
Yes, the human brain will still be involved, they have to set up the press and will not be asking for approval unless its ready. This is a tool to back up a human rather than to take over. All our prints go to art for approval and almost all of them are good to go, but every now and then, something is not right. This tool would catch those issues with about 90% certainty. No tool is perfect, but it should help.
Pierre

[Admiral]

Not to be a debbie downer but I don't see this ever actually working well, and I also think it's kind of important to have human brains in the process.
My guess would be if they aren't using machine vision in the offset world already (which would be 1000 times easier to implement) than we won't ever see it in ours.
Like a lot of things.

I Googled this and there are a couple systems that popped up right away (Lithec Litho Flash - In Press Camera System, EyeC ProofRunner).  So, they definitely are in that industry.  I have done offset printing / plate making, etc.  Lot less errors than in screen printing, so possibly not as cost effective! I can't remember "runs" of messed up printing for offset ever, and I did bindery and more as well.  I agree it would be easier to implement!

[DonR]

Someone approached me about doing this to monitor DTF printers.  We have several machines running all day and the system would allow for fewer operators.  At the time, I was not interested.  I don't still have the contact information for the person but if you can find him, they may be the perfect partner.  Lots of overlap.

[blue moon]

Someone approached me about doing this to monitor DTF printers.  We have several machines running all day and the system would allow for fewer operators.  At the time, I was not interested.  I don't still have the contact information for the person but if you can find him, they may be the perfect partner.  Lots of overlap.

I think this system could be set up to work with dtg/dtf. There would have to be a bar code with every design and a place to retreive the masters from. Not very different from what i am thinking.
Pierre

[Ron Pierson]

I think M&R has it - look them up, I saw a vid about this from them this morning.........10/17/2025

[blue moon]

I think M&R has it - look them up, I saw a vid about this from them this morning.........10/17/2025

Yes, similar idea. Does not provide approval, just monitoring.
Interesting that the announcement would come just a day or two after we started talking about it here. On the other hand, it does say version 2.0, so was there an earlier version before?
Does anybody have any info on the pricing?
Also, it looks like it would only work with M&R equipment…
pj