In this global emergency time, many designer-makers are working hard to help the community. What has emerged is that everyone must do their own part.
At this stage, valve prototypes (for compassionate care) and the design of the protective mask templates and related workflow for prototyping are being tested from home and in the Fablab.
My experiment picks up and improves some proposals available online, before showing you some pictures and the result achieved, however, I want to make some specific practical and technical observations:
1) The quality of the images for the photogrammetric reproduction procedure of the face has imperfections (quality of the point cloud computed, laying of the model not perfectly symmetrical);
2) The use of the parametric approach for the mask design from a practical point of view is not clearly immediate as direct modelling could be (as is often suggested in the tutorials proposed online;
On the basis of these two observations, I have experimented (observation n.1) at least two passages that I consider quite valid because, unlike the proposed tutorials, they make immediate use of two free software, one of which is opensource.
The first one is Meshroom – through this software and a series of selfies (therefore self-shot photos without the help of a second person as proposed by most online) I was able to reconstruct my face through the photogrammetric process as a mesh object. Through the application just mentioned, it was possible to reduce the complexity of the mesh (decimation) and the subsequent export with its texturization (the latter is not essential for the objective to be achieved).
The second software (an old and current “glory” that is part of the reverse engineering toolsets) is Meshmixer – through this mesh object manipulation software it was possible to proceed quickly to clean the mesh generated from the previous phase and refine it further to give it a smoother shape.
At this point, after importing the OBJ file into the Rhino environment, I fixed the virtual object (face) to make a tailormade protective mask (so more comfortable). I do not go into the details of the parametric modelling but is clear enough by importing a new “face”, the algorithm detects a series of lead-lines obtained from the 3D face automatically – the fundamental one consists of the nose and cheek height as well as the shape of the chin.
Fundamental for the purposes of protection (real objective) is to resort to the use of a front spot to position the TNT ffp2 / ffp3 (three plies are recommended – at least compared to the fabric that I bought online). In addition to the solutions that optimize this insertion, there is that one to provide a sort of cap, which in my experimentation has an inner position and minimum geometric dimensions obtained from the biometric characteristics of the face. These guarantees, for the same length of the Tnt cloth, to obtain a greater number of the Tnt filters (a sort of patch).
Fixed these “sine qua non” conditions, the next consideration was to improve aesthetics, design. Unfortunately online, except for a few instances, the final products are a bit like those weird clothes that the fashion stylists show on the catwalks while others look like the respirators that we saw in the Star Wars (or the Bane’s mask in the Dark Knight).
From this moment on (observation n.2) it is pure experimentation, design and digital fabrication …
Aside from 3D printer and filaments costs, this workflow compared to those proposed online can potentially be free of charge and carried out by a single person. Obviously, for parametric and direct modelling I prefer the Rhino+Grasshopper environment, due to professional reasons, but you could easily proceed with open software such as Blender.
To be clear enough again, this was a personal homemade experiment (or rather made “from home”) is not the solution but it is certainly one of the solutions – Omnes viae Romam ducunt!
Conclusion, a tip to reduce errors and printing issues – the front filter spot on the mask, in which the Tnt ffp1 / ffp2 cloth-filter will be inserted, should have an orientation parallel to the tip nose-mouth line so as to generate a flat basis which facilitates the positioning of the object on the print bed. This tip will reduce (unless further fixing actions) the generation of the supports and the improved adhesion on the print bed. I used a 0.2 mm printed layer height and the classic PLA material (white colour).