After the exhibition and taking it down last week, I wanted to go back to working on trying to create the facial sculpture which I included the print version of in the exhibition. I purchased a 3D printer and began to calibrate the settings in order to speed up the print time while maintaining quality and reducing the effects that warping could have on the design. Here you can see a video of the printer functioning with very little warping occuring. The more I can minimise this the better as I continue testing before I finally go ahead with the full scale prints which I have planned. I will most likely segment the head into several sections in order to reduce the risk of the build completely failing on me.
Here you can see me beginning to attempt to print the head sculpture generated from the machine learning training in 3D at full scale. I split the model into three sections which are shown above in order to make the prints less time intensive, and also to reduce the chance of an error ruining huge parts of the print. I also realised that it would reduce the amount of support structure required for the entire build because less high stem supports would need to be generated for the build.
Using Cura, a slicer for 3D printers, and having created a suitable build print setting module, I started by printing the head. This print ended up taking over 20 hours and the results are shown bellow, and along with the progress video. This was the largest and most complex segment so if this worked, I was highly confident the rest of the sections would also print really accurately.
This second video shows the base section of the design being printed in the same way. The whole process is going to take around 1300grams of PLA and therefore splitting it up also allowed the structure to be spread across 2 one kilogram rolls rather than having to time a filament change and risking a z axis shift which was happening before due to temperature changes.