Body scannings

Body scannings, testing and comparing different 3d scanning processes and softwares (Fabricademy, FabTextiles, Fab Lab Barcelona, 2018)

Why digital masks ? Because instead of making a tangible mask I decided to explore the different 3d scan softwares. After all, we are becoming more and more digitaly represented no? The scanning booth are becoming a reality in our clothing stores. How will we be digitaly represented in the futur ? By body scanning, reflexion of ourselves ? or in a fictive personlized way ? How do I want to represent myself ? In this exercice I wanted to play with 3d scan softwares and see how it reveal different facets of my digital identity. I started this project by testing two scanning techniques usually used in our lab : Scanect and Recap. Then, I wanted to test Agisoft Photoscan famous in the video games professional field. But during my research I found Regard3D, a free open source software that have a nice settings and definition. In this assignment I compared the price, the axessibility and the complexity of each 3d scan softwares I used.

Project : Testing the different scanning softwares, designing my own digital masks.

Process :
(1) 3d scan your body with a kinect & Scanect 3D software
(2) Take pictures of your face and 3d model with Recap Autodesk
(3) Experiment with Agisoft Photoscan
(4) Try the different settings on Regard3D
(5) Compare the four 3d scanning softwares used
(6) Set up the Gcode of the best 3d face model on RhinoCam
(7) Mill the 3d face model on the ShopBot CNC

Material :
- a computer
- a kinect + Scanect software (Mac & PC)
- a camera + Recap software (PC)
- Rhinoceros software
- Rhinocam software + a CNC milling machine
- a high density foam of 120 mm of thickness
- sand paper
- Agisoft Photoscan software (Mac & PC)
- Regard3D software (open source, Mac & PC)

(1) 3d scan your body with a kinect & Scanect 3D software
Skanect is a free software for Mac and PC but you need to have a kinect that cost approximately 100€. How it works ? Open Scanect 3D software and connect te kinect to your computer.
*Main menu : Prepare , Record, Reconstruct, Process, Share
Go to Prepare *New file -> Basic settings :
- Scene : (Body), Object, Room, Half Room
- Bounding box : (1x1x1 m)
- Aspect ratio : (Normal), Height x2
- Path : (backup path)
- Configuration file : (None)
*Click on 'Start'
*Record : You can see on the top right 3 small screens : the kinect connecting view, the webcam infrared view and the scanning monitor view. You can control the time of recording, for exemple 1 min. Start recording, by moving the kinect all around the person you want to scan or putting the kinect in a fix place and turning the object. For having the best mesh definition it's better to not rotate too quickly. You can see in the scanning monitor view the model becoming more and more green. It's better to have the less red parts as possible. When the recording is finish you're going to have a ruth grey 3d model, not so precise, like the first mock-up of a sculpture.
For having a better render go to *Reconstruct : Fusion ('Perform an offine reconstruction on required data.') :
- Processor : CPU ,( GPU)
- Fidelity : Low, Medium, (High) Clic on 'Run'
For having more details and more texture you can reload the data recording. This option can help sometime for having a better definition.
Then go to *Process :
- Mesh : -> Watertight, option medium for exemple (for filling the holes if you have any) it's going to close the mesh and colorize the 3d model with the scanned colors You can also find more options in :
- Geometry : Simplify, File Holes, Move & Crop, Remove Parts
- Color : Colorize, Remove colors, Remove colorless ?
After you go to *Share
You can Save the file localy on the software or Export the 3d Model in different Format like : PLY, OBJ, STL, VRML
- Choose Colors : Per-vertex Texture UV
- Faces of the mesh
- Scale : in meters, mm, inches… Then clic on 'Export' to save the file.

(2) Take pictures of your face and 3d model with Recap Autodesk
Unfortunately this program only works on Windows, because I don't have a PC, Laura made my 3d model on her computer. Apparently, ReCap is installed on Autocad software by default. You can directly open it from the point cloud panel by clicking on the ReCap Autodesk tool. Then you can choose to open the free ReCap version or try for 30 days the ReCap pro version. ReCap pro is more precise because the option 'survey points' allows you to enter the exact dimensions between two points on the object. How ReCap works ? It’s an easy program to use, very intuitive.
On *Home screen : new project, open other projects, adjust program preferences, open the autodesk online help center click on "new project"
*New Project :
On the left side of the screen you can see the *Project Menu :
- Home : Save, Open, Publish in ReCap 360, Importing additional scans, exporting the current state of the project, viewing registration results, adjusting program preferences, online help and exit the program.
- Display settings : colors, lights, point cloud, views
- Limit boxes : 2 options
- Navigation tools : window, pan, orbit…
- Provide feedback to Autodesk product
*Add the pictures that you took to turn them into a 3d model. Before import all the pictures, there is a validation process to check the quality of the images. It will not select the picture if they are out of focus or blurry. For taking good pictures you need a nice light and good camera settings. Take pictures by moving in a circle all around the object. Do it few times on different angles.
*Start the reconstruction process : it will probably take awhile to do it, around 30 minutes to 1 hour depending the number of pictures you have. For this exercice we didn’t took a lot of pictures because it was only for making a face scan.

(3) Experiment with Agisoft Photoscan
Photoscan is a professional software develop by Agisoft. It’s use for developping video games, geographical platforms, documentate cultural heritage... Photoscan is an expensive software, the professional version cost 2940 € and the standard one 150 €. You can use it on every operating systems : Mac, Windosws and Linux. Photoscan have advanced parameters and allows you to have a really nice 3d model rendering. Like Regard3D, it’s possible to edit point cloud and mesh. But Photoscan is a bit more complex to manipulate. I tried by curiousity without knowing how to use it and I didn’t succed to get a nice 3d model by using the default settings like the other softwares. At the end, my 3d face model looked more like an Art work than my digital face. But I must say, I kind of like it. Why are faces should be digitaly identical to our real face ? Maybe a little bit of freestyle and software error is a good way to get a personal digital mask.
How Photoscan works ? Open the software, then go to *Workflow and *Add a new Chunk, Select the pictures you want to use for your 3d face model. Then go to *Workflow *Align photos (Medium and generic settings). After the alignment process you can see your first point cloud appear in the main window. The blue panels are the positions of the camera. You will probably need to continue to align manually your pictures by marking them one by one. As you can see in the images bellow the software didn’t recognize the order of my pictures and assembled them inside out. It’s probably because I didn’t import enough pictures for making a propre 3d model and this professional software is lost. I decided to keep my digital mask like this and continue to play with it. I went to *Edit *Build Dense Cloud (Medium quality and default settings). Then I create a mesh by going to *Tools *Mesh. It created a crazy shape that I will call my Experimental Digital Mask.

(4) Try the different settings on Regard3D
I liked working with Regard3D, especially because it’s an Open Source software, free to use for commercial purposes. This software is simple to use and have advanced options if needed. This program autosave the files and own a classification system that save parameters used for each steps of the creation. Really practical to make comparison between differente settings tried. It allows to export pointclouds and mesh models. You can also export textures with OBJ.
How Regard3D works ? First open Regard3D. Start by create a *New Regard3D project , set the project path and gave it a name.
On the left you can *Add a picture set and select the pictures you want to import in your project. Give a name to the picture set.
Next you will click on *Compute matches for finding correlations between the images, you can set the Keypoint sensitivity and the Keypoint matching ratio. I used the default parameters and it worked well. If you go to *Matches results you can see in the preview window the Keypoints of the images.
Click on *Triangulations for creating your first point cloud. Choose *Global Structure from Motion and leave the other default settings. Now you can see small cloud of dots appear on your main project page. You can also see green dots representing the camera locations.
Then, click on *Create dense pointcloud , options of Densifications : CMVS/PMVS or MVE. I choose the first option and I played with the Level, Cell size and Threshold settings. I tried several densifications settings and I had really different results.
After finding a good densification setting your can *Create a Surface.
When you get a clean Surface you can now export and *Save your file as an OBJ. or PLY.