I am not sure, but I think VRML should do the job. This file format describes a scene (for instance it can contain cameras), so I see no reason why you could not describe a link as an object and then instantiate this object several times with different position matrices (a matrix is more complex than just giving a angle and position offset, but still more efficient than describing the link again).
yep. sth like this but last time I Uploaded a file with this, I got a strange error mail. And it takes a day to get the mail the "your model is ready to print."
OK, let's assume you can pack more chain links into your file using such a technique. At some point the memory limitations are transferred to the printer, if not a physical limitation on how much chain you can pack into a single print run, or the ability to pay for it!
I am unsure the limitation is the size of the file, it is rather the number of polygons (and the price, of course, you are correct). With longer links (at is as been suggest before) the limitation could also be the maximum bounding volume of the printer. And finally, the link must be strong enough to support the weight of the whole chain. This leaves a lot of degrees of freedom, I guess.
I was thinking that a "compact" instance file format could get expanded to a much larger polygonal based file after it gets to Shapeways, although there might be other bottlenecks like MeshMedic in their process flow. At some point all of the files to built in a tray must be combined into one huge master build file. I would think the memory or polygon limitations for an individual model are somewhat arbitrary for practicality. But the Shapeways people should be able to create very long interlinked chain files in house if they so desired, up to a maximum number of polygons that a particular printer could handle. Is there a recognized world record for the length of a 3D printed chain, or length/build volume ratio? I found this for a 50 foot chain built in a 5x5x6 inch volume. http://www.3ders.org/articles/20130917-mit-researchers-creat es-the-world-largest-print-with-form-1-3d-printer.html
Yes, you are probably right: the polygon limit per object is perhaps arbitrary but a polygon limit by print tray is very likely. Thank you for the link: I had seen the video of the 50-foot chain, but the interesting thing is that their link they used - the one on the last pictures at least - is very similar to mine (a double loop) although there are some cuts I don't really understand (can this chain take another shape? or do they just explain that the shape it has when printing is different than its shape when unfolded?). 50-foot length seems easy to achieve, it is only three times what I have done yet. And the price would be something like $150.
Using VRML's DEF / USE function will not avoid the maximum triangle count limit. As suggested once it gets to the 3D printer it can't really use instancing. At some point in our backend pipeline its get's deinstanced. We've been moving our software to allow 2 million triangles. Officially the max is 1 million but feel free to try a model bigger then that. There are still pieces that might fail above 1M but our tests have shown success with some 2M models. Instancing will help you overcome the 64M zipped file limit on upload and it's generally a bit faster for us to process.
Not that I'm suggesting this as a must have capability or anything, but for the people interested in making bigger/longer things like chains... If there were an option to assemble 2 or more models together in the build tray after upload you would be able to increase the size of something like a chain by a factor of 2 or more, circumventing the upload limits. This would require some set of registration or alignment markers in the uploaded model such that they could be properly aligned to each other and assembled into a larger instance when the build tray is assembled by a SW tech or algorithm. Just a thought for people or Shapeways resident artists with extra clout who enjoy pushing the envelope. Here's a simple 2D rendition of what I'm talking about. The orange blobs are supposed to indicate alignment markers of some type (that may be printed and discarded unless there is something like a no-print alignment material definition available). Alternately you could use numerical offset parameters to determine the positions of the assembled base units. I'm showing the same base unit being replicated here in one dimension, although with a bit more thought and more rules you could come up with a way to combine the same or different uploads in three dimensions.
Would you accept this as being longer if I ever get it printed in elasto plastic? It's about 15 meters of wire, 750000 faces, just one shell...