Like any good story, I’ll start in the middle. As the project went chronologically we started modeling donors long before working on the fuselage, but this story probably makes more sense starting with our reconstruction of the major shapes of the iconic ship.
Jason had received a couple original ILM castings and this would be one of our primary starting points. We also had v5 casings from Mike Salzo as well as a mountain of vintage photographs of the original models. Getting digital representations of each of these assets was the first step to resurfacing the fuselage.
Unfortunately, the ancient castings had seen better days, and they were very warped and distorted. I was able to “undistort” them to certain extent, but doing this would probably result in some unintended stretching giving us a false reference.
After some digital tweaking, I had a composite of the three different reference files that could be used as surface reference. It wasn’t perfect, but it was a good start.
The scan data was great, but not stellar. So, Andre pulled another trick out of his hat. We had about 90 photos of Red Three that Jason had taken. They we’re roughly shot as a turntable sequence, so Andre ran the sequence through 3D tracking software to produce a series of points-in-space that became another point of reference.
I was able to bring the tracking points into my CAD packages (Alias, Fusion 360, and Maya) as additional reference to supplement the laser and photogrammetry data.
This whole process of gathering data and compositing it in 3D space took a good chunk of the first year. All along we were refining our workflow and coming up with new ways of using it to better affect.
As we collected the above reference data, I was concurrently surfacing the fuselage in Alias. By the end of the project we had three generations of surface CAD each with a dozen iterations as we gained a better understanding of the nuance of the fuselage. Honestly, I really underestimated the subtleties of the ship having seen blockier models in the past. The body of the X-Wing is quite shapely, and clearly hand-carved. A beautiful piece of sculpture – until you need to replicate it in CAD.
By the time I had gotten to the third generation fuselage model, we had introduced a new step in the modeling process. We had the scan data, and the 3D tracking points, but now we were round-tripping the model back into into the scene we used to track the 3D data points as a way to verify what I was modeling was lining-up to the original reference. This approach let me model, tweak, and test across a dataset of nearly 100 different images to determine how accurate I was being with the surfacing. We got pretty close – asymmetries in the original model not withstanding. Andre and I have started using this same technique to divine the exact curvature of the Falcon’s domes and has improved our solves greatly.
While there were other aspects of the X-Wing that were also being worked on simultaneously, the “final” master surfaces took us just over a year nail down; from August 2018 to about December 2019 working an hour or two a night here and there; a near constant back and forth between CAD, scans, and photo-matching to get it just right.
Of course, along the way, Jason was printing full-sized (i.e. 1/24th scale) models to verify the surface data as we went. First in low resolution FDM prints, then eventually in higher resolution SLA prints… but let’s not get ahead of ourselves.