I’ve been playing around with photogrammetry a lot, lately, and use the opportunity in Brussels to take a ton of pictures of the many mounts. Of the cast of Stan the Tyrannosaurus rex I took some 350 photos! Some of these I turned into the HDR image I showed before, but the vast majority were intended to be used for photogrammetry. Day before yesterday I ran the first step: loading the images into the program and making it compute a sparse point cloud and the camera positions. Yesterday, I ran 3D mesh generations (which includes the creating of a dense point cloud) on three settings: Very Low, Low, and Medium. Here’s screenshots of the results (I cleaned up the low and very low ones, because the “noise” was hiding Stand too much):
This is the lowest setting, and I had to remove over half the polygons because they were noise until I got to this stage, where the general proportions of the skeleton are visible. This is good enough to build a volumetric model, though! Next higher accuracy level:
less cropping here…. and the medium quality, no cropping at all!
rather cool, hu? Check out how well the gastral basket looks! The haemals are all connected, but the resolution is good enough that I can cut them out of the noise and close the holes. Looking forward what ultra-high resolution will look like, once it is computed in – oh, about two weeks LOL!
Here’s a close-up: the right hind limb, cleaned up and holes closed.
The hip looks dislocated, but isn’t: the femur head simply didn’t turn out very well and is missing.
Been doing a fair bit of this myself recently, and loving every minute of it, although mine is centred around ichnology and a lot less complicated than Stan.
What software are you using?
Those are from Agisoft’s PhotoScan 🙂
And yeah, ichnology…. there is a lot in the works here with that 🙂
… two weeks …
gulp. Will be interesting to see if it was worth it.
That rex leg is sexy! You’d probably have to have a photo series from a higher vantage point to get the femur head to register, right?
indeed – all I could do was walk around at ground level and hold the camera up as high as I was able to. I’d have needed a pretty high ladder to get the missing parts.
I’ll play a bit with meshing only parts and see if that improves the recessed areas, too.
An amazing image. How many photos for the leg reconstruction?
that’s just a cut-out from the MEDIUM reconstruction, so I can’t tell you how many of the photos were used for this. All 350 photos contributed to the entire model.
Pingback: Photogrammetry is perfect – if everything work fine | dinosaurpalaeo
Pingback: RBINS: the rest | dinosaurpalaeo
Pingback: More RBINS Photogrammetry | dinosaurpalaeo
Pingback: Theropod Thursday 39: Freeze! | dinosaurpalaeo
Pingback: Theropod Thursday 40: Freeze! | dinosaurpalaeo
Pingback: Stegosaur photogrammetry – more WOW! | dinosaurpalaeo
Pingback: Stegosaurier-Photogrammetrie – mehr WOW! | dinosaurpalaeo auf Deutsch