Photogrammetry

My introduction to the practice of photogrammetry came from the R3T club where we built 3D printed prosthetic hands. In order to get a good fit, we 3D scanned the subject using an iPad that stitched the photos together to form a 3D object. While practice worked great, it took a long time to collect data and often failed altogether. We thought that it was not the best solution so we looked to create one of our own.

In order to create a better experience for taking data, we have created a concept that would allow us to collect as much data in a short amount of time using a rig of cameras around the subject. We hope this will help us in the future and cut down the initial measuring process.

Concept

Using a system of cameras, we can provide a 360 view of the subject. This will create a 3D model that can be used to make a perfect prosthetic fit. We hope this will help us in the future and cut down time on the initial measuring process.

Range of Motion

The two camera rings sit on a system of rails that allow for easy adjustment. Sliding up and down, the scanner can be centered around an object despite a variation of height. Sliding forward and back provides the scanning motion for creating a series of images around the object.

Cameras

By creating a ring of cameras, we can collect a series of images that can be stitched together and create a 3D model. Each ring contains 16 cameras that will provide a 360 view of the object being scanned.

Assembly

The scanner will be made up of a system of 3D printed rings that can be adjusted to create the easiest scanning scenario. The two rings both provide a 360 view containing 16 cameras on each ring. Each ring is then connected to a series of Raspberry Pi zeros along with a Raspberry Pi 3.

While this project turned out to be just a proof of concept, our research and work done on the project led us to new opportunities in a variety of fields.

VxMed Studio

Based on our work on photogrammetry for the prosthetic hands, we gained interest from a company out of the Texas Tech Innovation Hub. In collaboration between Raider for 3D Technology and VxMed, we built a photogrammetry studio used to create digital scans of patients.

My role in the process involved lots of individual research on the topic of photogrammetry. Essentially, we had to figure out the best way to turn a large number of photos of a subject and stitch them together to create a detailed 3D model. It is a process mostly used to both save time and money when creating unique digital models. Over a few months, I researched different photography techniques, photogrammetry software, and studios to get a good understanding of which direction was the best way to go.

We decided that the best way to create a studio would be to use modular cameras that would be synced together. Arranged in a circle around the subject, the studio would capture 360 degrees of images to provide as much data in the shortest amount of time. By using Raspberry Pi zeros, we could send all the photos into the same place while actively taking them at the same time. We set up a server that linked them together so that there is no issue in getting the results.

Using inspiration from a variety of camera mounts on the market today, we designed a number of mounts to best fit our needs. They are designed to snap fit a raspberry pi and to be easily fastened to t slot extruded aluminum. After figuring out the generic shape that satisfied our needs, we went through a few more steps to get the most out of the materials used.

Through topography optimization, we could see the strengths and weaknesses of the mounts and see how we could improve them. By doing so, we reduced both the material used and the weight of each mount in the process. After several iterations, we settled on a mount that was modular, low cost, and well built.

Here is an example of different mounts that we tested to see how they compared to one another when mounted on the main structure.

After months of researching the different photogrammetry softwares on the market, we decided to compare two: Meshroom and Reality Capture. Meshroom is a new open source software that is free to use. While it did offer some decent results, there is still not much known about its consistency and reliability in constructing 3D models. Reality Capture is one of the industry standards and can produce solid results with only a few photos. Not only does it perform at a much faster speed, it allows for the selection of data reducing the amount of memory stored per model.

We went through many iterations of the overall design and how everything would eventually link up. After months of testing to see which was best we decided that collapsible columns would work best. Not only would it provide the 360 degree range of images we would need but also allow it to be very mobile. The cameras would be easy to move and arrange as well as the lighting in terms of brightness and diffusivity.

Here is an example of a column with cameras used for testing the quality of the software.

Portable Scanner Concept

Based on our work on photogrammetry for the prosthetic hands, we also gained interest from the Texas Tech Archaeology Department. Their interest was in creating a low cost, modular and portable 3D scanner that could work in extreme weather and climate conditions.

When we first met up with the department they wanted to know how they could incorporate 3D printing into their workflow. Since many of the expeditions are off grid, they often face issues in getting replacement parts for things that break. After discussing solutions and possibilities using 3D printing in the field, that got us discussing the need for a portable 3D scanner. It would primarily be used to scan artifacts so that their current state could be preserved forever. By doing so a digital archive could be created allowing people all over the world to see their discoveries. In addition to this, they wanted to be able to scan sites and artifacts and have them 3D printed for physical touching purposes. Not only would this allow a more hands on experience for learning, it would also provide a solution to those who are visually impaired. With a few ideas in the works we started to figure out a solution to their problem.

Given that the equipment will be used off grid in the most extreme climates, we designed the scanner to be as modular as possible. The scanner is made up of four section that attach to a common trigger hand grip. Each section has three raspberry pi zeros along with three cameras attached on top. Secured by an assortment of rubber gaskets and bolts on top of a translucent poly-carbonate plate, the sections are able to be taken apart making it easy to transport.

The scanner will operate by the process of photogrammetry where the pictures will be stitched together to create a 3D model. To gather the photos, all that has to be done is to press the trigger and all four sections will take pictures of the subject. By creating a 360 view of the object at hand, the scanner will be able to provide enough information to get an accurate and reliable model. We are also looking at new attachments such as lasers for more difficult materials to scan.

One of the main issues with scanners is that they break. By making every part able to be replaced, we hope to be able to fix that problem. We hope to provide replacement parts or even parts that can be 3D printed so that it would no longer be an issue. It is also capable of being broken down and place into a travel case. This way there would be less opportunity for it to break during transportation. While we believe that the design should work based on the specifications of the Archaeology Department, we hope to continue to discover how we can improve our design. Whether it is making it more weather proof or improve its durability, there are countless ways we could really make a product that the users will enjoy.

Here is a render of the scanner along with the carrying case showcasing its ability to be broken down.