Trilobite Photogrammetry

 I chose to scan the trilobite fossil next because it is quite different to my other fossils and was hoping to capture some of the unique forms and details that remain part of the fossil.

The first few attempts failed to recreate many of the photo's positions. Usually the surrounding environment was enough to recreate the model accurately but since it was failing this time, I found a new more consistent, reliable method. I placed two guitar picks, as you can see in the the photo above, in opposite corners of the book I use as my scanning backdrop. This allowed each frame to show at least one of these anchor points at all times and ensured that Zephyr had as much spatial reference as possible.

Since the bottom of the object is relatively flat I was confident that I would be able to cut away the artefacts from the object it was sat on and replace it with a flat face, replacing the texture later. This saved me from having to join a seam around the whole object.

Using similar methods to my other models, I was able to remesh and reduce the object to a suitable level. After getting some peers to review the difference between 48K, 24K and 12K poly versions of the object there was not a noticeable difference between the 3 so the 12K version was perfect to use.

I cut the bottom of the object using a plane and the fill hole tool. After retopologising the new face, I UVed the bottom faces and took a UV snapshot into Photoshop, where I placed an image of the bottom side over where the new UVs were. This took some time to get right but eventually lined up quite well.

Once I applied the new material in Maya, I noticed that the asset was missing some information on the cutting plane that was made obvious by the texture. A darker patch of texture was highlighting a small cut-out made into another piece of rock. I used the multi-cut tool to make edges around the border of this section and extruded it inwards, giving the texture a more natural place to sit. It also added more detail to the bottom edge making it look less flat where I made the cut.

Once I took the asset into UE4 it revealed another problem to me with my workflow that I had not encountered before. I will make a separate blog post detailing my error but the problem left fragments of green texture from the book the fossil was on top of. It was an easy fix when I discovered the issue and I was able to remove it easily and simultaneously improve the quality of the texture.

The last adjustment to make was colour-correcting the seam. After the success in Substance Painter with the mosasaur tooth, I chose to use this method again. Once again this method proved to be really effective. The only downside to this attempt was that since there was only one texture for this model, unlike the two halves of the mosasaur tooth, I couldn't avoid adjusting the main colour slightly too.

After these corrections, I used the same normal mapping website that I have used in the past to complete the object. I am happy with the speed I created this model and the quality of it's final result. It remained relatively low-poly at just under 10K in the end which is better for a large scale production and I also continued to learn about the pitfalls of using certain settings when creating photogrammetry scans.