This project dealt with picking an object to be digitized into a 3D model. To begin, numerous tests of different objects were attempted to understand the complexity of photographing a model in a specific lighting situation that would produce optimal results. This in turn ended with working with Andrew Frueh in the department of art and his photogrammetry setup. Part of the emphasis of this project was to also lower the amount of polygons in the resulting model, lower the image resolution, and create a stable and easily loaded model. Andrew Frueh and his setup in the department of Art @ OSU (Images from OSU) The object that I decided/attempted to digitize was an antique statue from an African-based tribe. The statue itself had many crevices and holes to attempt to capture, as well as specular highlights and minute textures. The process of generating a stable amount of photos includes having a proper background and foreground, allowing the software to denote 'markers' within the space to generate a solid 3D model for later manipulation. Images of the statue at different angles -- captured from the setup. *Click to enlarge* Above you can see the setup of where the object was stationed. We would place them down and take photos (in multiples of 6), for a single 6-10 degree rotation. The different heights allowed all of the cameras to capture different angles of the object itself. The total amount of photos taken in about 2-3 minutes was 300. Things to note:
Zephyr software reconstructing the photographs in 3D. *Click to enlarge.* Above you can see the images being shown in Zephyr. In the first photo is a representation of the different cameras with the 360 degree coverage. Here you can see the circular representation of the false camera position based on the position of the object. The second is a closer image of the actual object in the space; the last is a live photo of the object that was digitized. Notice how the white background was completely exempt from compilation of the model. 1 - Base statue || 2 - Base geometry of statue || 3 - WIP of statue retopology || 4 - Retopology of state After exporting the base geometry model from Zephyr, the model has a good amount of fragmenting from the dense cloud-points. I used the quad-draw feature in Maya to create the new topology from scratch, because at the moment of using Mudbox 2018 there were some retopology issues that couldn't be fixed. This was partially because of the small holes in the statue itself, along with the complex model of the statue in the first place. I unfortunately made a mistake after retopologizing my model by sending it over to Mudbox and accidentally overwriting my work with a higher resolution in the poly count. I'm sure it's noticable to see the difference in the amount of polys in the quad-draw vs. the final output. But, something to remember for next time! (always have back-up saves) Process video of retopology breakdown. After retopologizing in Maya, I sent the model over to Mudbox to adjust and add in detail to the Normal, bump, and diffuse maps. This included using different tools like the clone stamp, a subtractive wax brush, and switching between the layers as necessary to make sure everything was lined up correctly. Using the option of the 'tone-mapper' view in Mudbox, I was able to see all of the darker areas of the model--making it easier to fix/fill those areas as needed. Some render shots of the final output. *Click image to enlarge* While the render itself is much lighter in hue than the actual statue, I believe this is part of the ability of being able to digitize the textures. Having the freedom to change how the statue looks and perhaps *improve* upon what is initially visible is in my opinion, a viable way to capture an object. It's still unfortunate that I lost the low-res base that I had created, otherwise it would have been 5k in the poly count--rather than 20k. I can always go back into Mudbox and retopo it as necessary; there's a function to set out specific curves for the program to recognize to at least make it clean. Regardless, below is a turn-around of the statue and a comparison of the wire-frame. Enjoy! Comments are closed.
|
All PostsArchives
May 2020
Blog ContentsInterests: |