Sander van Vliet

Fabacademy2016
@Fablab Amsterdam


3D scanning and printing

Week 5 (24 February - 2 march)


Lecture

In this weeks lecture Neil explained the strengths and the weaknesses of 3D printing. Other machines and processes are often preferred to use over a 3D printer with reasons like it's stronger, faster, cheaper etc, but to understand the application of 3D printing the general principle is: what you can not make by subtractive processes (cutting/milling etc) you would make with an additive process like 3D printing (building in stages by adding layers). This is mostly the case for parts with geometry where you can not get in with a subtractive tool.

Assignment


Testing Design Rules
In Fablab Amsterdam we have the Printrbot Plus and the Ultimaker Original available and we ran some tests to see how they would perform. For sending the prints to the printer we used Cura, which is an open source 3D printing software application. I did not install the most recent version, because our instructor explained that previous versions have more configuration options in the advanced and expert settings menus which the most recent version do not have. I am using Cura version 14.12.1.
We printed test models from Thingiverse to see what the the possibilities and limitations of these 3D printers are. The first one was the below model which we printed on the Ultimaker. The result wasn't good as you can see below, the filament did not form the shapes correctly at all. This was not the Ultimakers fault though, because what happened was that the laptop running Cura went into sleep mode while it was sending the print to the printer. This was noticed a little while after the printer stopped printing and then the laptop was turned on again. You can notice the higher ring was being formed quite okay again so it seems the Ultimaker is able to print this model correctly.

We also printed another test model on the Ultimaker and this one went in one go and formed like it should.


Design, model and 3D print an object (small, few cm) that could not be made subtractively

Designing and Modelling it
I wanted to make something that I could use, so not only a model that is proof of the 3D print concept (additive). I thought of making a 3D printed dice, or a few of them maybe. The shape of a dice can be a simple square box, but because of better rolling behaviour most dice have curved sides.
A curved shape like that can not be made subtractively, at least not with a 2 axis machine like the milling machines most Fablabs have. The pips on the sides are also hidden inside the outer perimeter of the dice.
I measured an existing dice and in Rhino I drew a box of 16x16x16mm. Then I drew a sphere which originated from the center of the box and used Boolean Intersection to subtract both shapes from each other, leaving the smallest volume within the intersection of both shapes.

Then I made 9 small spheres and used rotate (copy=Yes) to place them on all 6 sides of the dice.

Then I removed the ones I didn't need and with the command Boolean Difference I extracted the partial sphered from the main model.
dice source file - Rhino

I exported the model as .STL file and scaled it to also have a bigger dice of 30x30x30mm and exported that as .STL as well.

'Printing' it
(still isn't really the right term, right?)
Because my daughter was sick I couldn't go to Fablab Amsterdam, but luckily Bart Bakker's miniFABLAB is very close to my house. While somebody took care of my daughter for a few hours I went there to print both the 16mm and the 30mm models, you can see the used Cura settings of the 30mm yellow dice and the Ultimaker process below.
The only settings which were different for the blue one were: No support, Print speed (mm/s) 50 and Fill Density (%) 20. I turned on support (type 'Touching buildplate') for the yellow dice to see if it would help form the one pip better. I changed the speed and fill density to resp. 90 and 10 because I guessed this would be enough fill and make a faster print, while still a safe enough print speed.


They came out quite well, although the bottom took on the texture of the bed surface. In the case of the blue dice it was the glass bed / build plate. I chose to print the yellow dice on a layer of painters tape trying to get a bit of grain in the bottom surface and it did, but still the surface is not the same as that of the other sides.
A rudimentary pip is visible on the blue dice, I guess the material deformed while it was still hot, closing the pip for the biggest part. On the yellow dice it also didn't form correctly although I had turned on the Support option. A tiny bit of support material has been printed there and we tried to break it out but because there is so little space the support and the model material melted together. However, it was still able to create a little pip!

3D scan an object
Because I had modelled the dice in Rhino I thought it would be nice stay with the subject and use an existing dice we have lying around to make a 3D model of by scanning it. I started with Autodesk 123D Catch which is a smartphone app. As their website states:"A free app that lets you create 3D scans of virtually any object". On the right you see part of the photos I took from within the app. One of the tips was to have a high contrast between the object and the While shooting there are indicators which help you to get enough shots and from the correct angles. The processing which was done locally on my iPhone5 took really long. About two hours I think it was!

And below you can see the result, which I had expected to be a bit better because I had taken the photos carefully following all instructions closely.
Trying to get a better result I did it all over, making new photos and more of them and trying to adhere to all indicators and tips etc, but it did not change anything. It resulted in a similar quality model with big parts of sides missing and holes in it.


I didn't feel like spending more time on 123D Catch and downloaded Autodesk Memento. "Autodesk Memento is an end-to-end solution for converting any captured reality input (photos or scans) into high definition 3D meshes that can be cleaned up, fixed, and optimized for the Web, mobile or 3D printing/fabbing"
I thought this pc application looked more capable then the smartphone app and it does the calculations in the cloud. I hoped this meant that it would go faster, but also with Memento it took a very long time. This was mainly because my model was in the queue for processing for many hours before it was finally processed.
And when it was completed... quite a disappointment! Although three sides look okay the model has no color. The sides that are there do not have holes this time though, but only three sides were recorded somehow, while I took the photos all around the object from different angles!


Because there are many programs which can do this I went on downloading an even more serious counterpart: Agisoft Photoscan "a stand-alone software product that performs photogrammetric processing of digital images and generates 3D spatial data.".
Reading about it, this could be a bit high end as the learning curve is supposed to be quite steep and it has a lot of advanced options, but I wanted to try it and see what it could do with my photos. Just by trying and guessing I got to creating a 3D model, but it needs more work to make it look better and I did not have much time left. You add the photos to a 'chunk' within the workspace. Then the first step of processing starts where it creates a point cloud by matching points it identifies. Quite a lot of the photos were 'not aligned' at this point, although I can't see how they differ from the 'Aligned' photos. I selected the not aligned ones and clicked 'Align photos' just to see if that would help. It didn't. I read about you can set markers on these photos and indicate their projections on at least two aligned photos, but I can't find how to. Anyway, from the point cloud you can build a mesh which is your 3D model. See mine below.


Maybe it would be possible to get more out of this, but after trying to understand and trying out different settings and options I came to the conclusion that I had to accept that this was my third and last unsuccessful attempt of creating a decent 3D model of the orange dice. In the future I would like to spend more time on this and be able to do it succesfully, but for now my week is over.