CAGD 350

3D Scanning and Application

Lidar / FARO Scene Project

    For our final two projects, we were instructed how to properly operate a FARO 3D Lidar scanner. For our first scene, our class group got hands on experience setting up the scanner on CSU Chico State's campus and performing our first scan on the stone ladies. While gathering the data, we were further instructed in practical uses of the scanner in different environments, weather, and lighting. Through this exercise, I was able to get a better grasp on the parameters available within the scanner, such as quality and resolution, as well as when to appropiately modify them for the scene.

    We took 5 scans around the statues, and didn't intend on using these projects files, as we had other project files supplied to work with in order to create our first render within FARO Scene. However, I wanted to see how well our class scan could be cleaned up, so I used these scans to create my first scene.

While the level-of-detail and quality of the scan weren't as clean as I had hoped, it was still a very fun process to learn using data that I had helped scan. I developed a workflow within FARO Scene in order to quickly and effiently take scans and clean up the data in order to create visual walkthroughs of the scanned scene.

    Moving on to our next and final scanning project, we were tasked with splitting into a group of 2 and scanning a portion of campus with a LIDAR laser scanner. My partner, Preston Farris, and myself scanned a structure near the library. We were assigned to bring the scans into FARO Scene, eliminate stray points and moving data, and then create a video using a camera path. I believe that my scene turned out very well, and preserves the detail away from the structure to give the viewer a sense of space.

White Light Scanning - Head Scan and Object Scan

    This assignment tasked us with using a handheld white-light scanner to scan an object in class, and to texture it within an external program using images taken of our object using a turntable. Having a lack of items to fit in my backpack as I walk to class, I settled on my shoe (at Dan's suggestion). This turned out to be super fun and provided a lot of detail to work with on my model.

    The shoe in question is from a pair of red Converse. To help the scan from working with a deflated and drooping shoe, we stuffed it full of towels to give it shape. I then took this scan through EXScanPro and exported the mesh to ZBrush, where I cleaned up the model. I then took the images I had taken and worked to color correct them. Using these newly balanced images, I projection painted the material onto the model in Substance Painter, before completing my renders in Marmoset.

    For the Head Scan assignment, we used the scanner in class to scan a classmates head. Afterwards, we used a technique called cross-polarization in order to take images of the subject rotated every 15 degrees in order to get the full detail of their head to paint onto a texture for the model. Using the scan, we processed the data in EXScanPro and exported a mesh. We would go on to later clean up this mesh within ZBrush, and create our remeshed versions of our model.

    It is important to note that for this project, I did not use ZWrap, which was graded as part of the assignment. I also believe that my high-pass textures did not get baked properly after importing back into ZBrush.

    My partner color corrected the needed images before handing them over to me, so they were ready to go within Substance Painter, where I projection painted the images onto the model. I struggled a lot with this tool, and it took me a few hours before I felt that I had properly aligned my painted images onto the 3D model of my classmates head.

    After finishing up in ZBrush, I moved my models into Marmoset and set my lighting and scene with the required backdrop. I am very happy with how my scan turned out after the work put into painting the textures.

Photogrammetry

    For this series of assignments, we learned how to properly take photos that focus on capturing detail and color, and focus less on the dramatic elements of photography. Using these skills and a series of carefully positioned images, a 3D model can be formed by finding symmetry and overlapping points while being processed within a program (such as RealityCapture). Below is a project that we worked on with images that were taken prior. Using these supplied images, we used RC to align our data and export a model into ZBrush. 

    From ZBrush, we cleaned up our model and exported into Maya in order to create a very primitive version of our model with low geometry. We then moved back into ZBrush in order to use the Z-Remesher tool in order to provide us with a great deal of control over the geometry of our model. We used this to export both a low and high-poly model out to Marmoset, where we rendered out a scene with our models.