This manufacturing technique sucks
July 28, 2016 12:02 PM Subscribe
So maybe you've seen hydrographic printing. Last year a team devised a technique for computational hydrographic printing which gives amazing results on 3D objects. Now a team has used similar techniques to do computational thermoforming which give amazing results for printing onto a vacuum formed 3D object.
That seems really valuable and I hope they actually give their product/process the name "Our Solution".
posted by Mei's lost sandal at 12:17 PM on July 28, 2016 [1 favorite]
posted by Mei's lost sandal at 12:17 PM on July 28, 2016 [1 favorite]
How soon before I can print me a hot tub?
posted by ZenMasterThis at 12:24 PM on July 28, 2016
posted by ZenMasterThis at 12:24 PM on July 28, 2016
Or. how soon can I use this and a hot tub to print on myself?
posted by Devonian at 12:54 PM on July 28, 2016 [3 favorites]
posted by Devonian at 12:54 PM on July 28, 2016 [3 favorites]
...The only hardware is common workshop equipment, this is great, thanks for posting.
posted by Dr Ew at 1:09 PM on July 28, 2016
posted by Dr Ew at 1:09 PM on July 28, 2016
...The only hardware is common workshop equipment, this is great, thanks for posting.
That's in indeed the cool part. The software seems pretty sophisticated though.
posted by GuyZero at 1:20 PM on July 28, 2016
That's in indeed the cool part. The software seems pretty sophisticated though.
posted by GuyZero at 1:20 PM on July 28, 2016
The software seems pretty sophisticated though.
That's the really cool part, because, not really...it looks really similar to the processes for wrapping a 2D mesh on a 3D video game character, which is a pretty common piece of software. In fact it looks simpler than the hydrographic printing where (I assume) you have to take the water flow into account (water being much more fluid than thermoplastic)
posted by sexyrobot at 2:16 PM on July 28, 2016
That's the really cool part, because, not really...it looks really similar to the processes for wrapping a 2D mesh on a 3D video game character, which is a pretty common piece of software. In fact it looks simpler than the hydrographic printing where (I assume) you have to take the water flow into account (water being much more fluid than thermoplastic)
posted by sexyrobot at 2:16 PM on July 28, 2016
That is amazing. And it is one of those techniques we could have been doing 10-20 years ago if anyone had just put the pieces together. Or at least it doesn't seem to require a new break through in materials science or process.
posted by Mitheral at 3:12 PM on July 28, 2016
posted by Mitheral at 3:12 PM on July 28, 2016
Here (.pdf) is the paper.
GuyZero: Now a team has used similar techniques...
Both the computational hydrographics and computational thermoforming projects are joint ETH Zurich/Disney Research projects, involving several of the same people.
sexyrobot: not really...it looks really similar to the processes for wrapping a 2D mesh on a 3D video game character, which is a pretty common piece of software
The difficulty here is mainly that in order to know the mapping between positions on the initial image and positions on the 3D surface, it's necessary to simulate the thermoforming process. And before you can do this accurately, you need a calibration process to estimate parameters.
This is pretty computationally intensive: the paper says the offline calibration process takes 36 hours of parameter grid search, and then the simulation for a single model takes 5 minutes (both on a fairly beefy machine - Xeon CPU E5-2650 v2, 64 GB RAM).
posted by James Scott-Brown at 4:00 PM on July 28, 2016 [1 favorite]
GuyZero: Now a team has used similar techniques...
Both the computational hydrographics and computational thermoforming projects are joint ETH Zurich/Disney Research projects, involving several of the same people.
sexyrobot: not really...it looks really similar to the processes for wrapping a 2D mesh on a 3D video game character, which is a pretty common piece of software
The difficulty here is mainly that in order to know the mapping between positions on the initial image and positions on the 3D surface, it's necessary to simulate the thermoforming process. And before you can do this accurately, you need a calibration process to estimate parameters.
This is pretty computationally intensive: the paper says the offline calibration process takes 36 hours of parameter grid search, and then the simulation for a single model takes 5 minutes (both on a fairly beefy machine - Xeon CPU E5-2650 v2, 64 GB RAM).
posted by James Scott-Brown at 4:00 PM on July 28, 2016 [1 favorite]
Man that type of printing looks slow. I mean it takes ages to put the object into the liquid, and then you have to reset the surface each time. I would have thought printing methods for mass production would be a lot faster then that.
This is pretty computationally intensive: the paper says the offline calibration process takes 36 hours of parameter grid search, and then the simulation for a single model takes 5 minutes (both on a fairly beefy machine - Xeon CPU E5-2650 v2, 64 GB RAM).
Sure, but if you are doing a print run of 50,000 objects for a board game that will take you weeks of production time, and you do the calibration once then can use it again and again, five minutes at the start of several weeks suddenly looks a lot better. Also if it is a capacity your factory has that others don't, I'd think that is a salable ability. For producing a single object, I agree, that looks less interesting.
(Why don't they just use a bunch of robot arms with airbrushes? Too expensive?)
posted by Canageek at 4:37 PM on July 28, 2016
This is pretty computationally intensive: the paper says the offline calibration process takes 36 hours of parameter grid search, and then the simulation for a single model takes 5 minutes (both on a fairly beefy machine - Xeon CPU E5-2650 v2, 64 GB RAM).
Sure, but if you are doing a print run of 50,000 objects for a board game that will take you weeks of production time, and you do the calibration once then can use it again and again, five minutes at the start of several weeks suddenly looks a lot better. Also if it is a capacity your factory has that others don't, I'd think that is a salable ability. For producing a single object, I agree, that looks less interesting.
(Why don't they just use a bunch of robot arms with airbrushes? Too expensive?)
posted by Canageek at 4:37 PM on July 28, 2016
Re: the first two videos. So this is the best the 21st century can do?
Here Honda kids with your oh so crucial (and very manly) coffee can mufflers is a way we can bilk you out of more of your money by having you pay us to get your bog standard factory wheels printed with customized graphics.
I kinda get it because I remember being really young and spending money on things I thought were important at that time, but when you get a bit older the WTF? really hits you (or at least me) extremely hard.
posted by InsertNiftyNameHere at 9:34 PM on July 28, 2016
Here Honda kids with your oh so crucial (and very manly) coffee can mufflers is a way we can bilk you out of more of your money by having you pay us to get your bog standard factory wheels printed with customized graphics.
I kinda get it because I remember being really young and spending money on things I thought were important at that time, but when you get a bit older the WTF? really hits you (or at least me) extremely hard.
posted by InsertNiftyNameHere at 9:34 PM on July 28, 2016
« Older Fusion power: a new hope | Not a sunny day Newer »
This thread has been archived and is closed to new comments
posted by griphus at 12:13 PM on July 28, 2016 [7 favorites]