The bandage has already kinda three quarters been pulled off. The vast majority of the data we need has been recreated in the new form that it needs to be in, but because a lot of it had to be done by hand there's plenty of room for issues. Now it's a case of just tracking down where things went a bit off the rails and fixing the them.
For example, the KQ twins busts. We never actually had seperate busts for Model 17 and 18, they were just crops of an image with both of them in. Whoever did the bulk conversion of the bust data didn't setup the cropping data for Model 17 (so it displays the whole twins image), and skipped setting up Model 18 entirely (because it functionally used the same image but with different cropping data).
Regardless, I've cleaned up a bunch of the busts throughout the KQ segment and I'll chip at whatever I notice with some additional logging shoved into the development code plus whatever gets reported. Trying to exhaustively ensure they're all fine would be a monumental task no matter which way we slice it:
Image processing manifest path: C:\Users\gedan\AppData\Local\Temp\image-manifest-proc.json
progress [========================================] 100% | ETA: 0s | 5286/5286