A similar stabilisation was used with theZapruder JFK footage.
Is there some crazy ass ninja excavator Olympics somewhere? 'Cuz I really want to go after seeing all these videos.
2 Likes
The Zapruder stabilisation is much better, though. The only clues about the position of the active frame are the ephemeral pieces (the bits what move over time like the car and motorcycles) and the left edge of the film which held picture information but also the spoke-holes used to feed the film through the camera and subsequent projectors (so the picture information was deformed).
(see a still to see what I mean by spoke holes)
This excavator one suffers from the projection not being properly stitched. (puts on professor hat)
All cameras have lenses which act to project some portion of the scene onto a flat sensor. This works about as well as map projections of the globe: you canât get perfect fidelity of a sphere by projecting onto a flat surface. (If you watch the trees at the top of the frame at the beginning of the video you can see that theyâre rendered in different positions and sizes based on what part of the frame theyâre captured in.)
What ought to have been done would be to read the lens spec out of the EXIF data (if not available, perhaps read against a database of device->lens records, since this was almost certainly shot on a phone which only ever has one lens) and then use that to project the frames against a world-sphere of a guessed distance based on per-frame stitching (which the algorithm already has to do for the static frame). This would be a much better approximation since most of the scene is roughly the same distance away. (If there was too much variance in distance, an approach like Microsoftâs HyperLapse could be used to generate scene geometry from the individual frames, thus taking most of the guesswork out of the projection)
hugin is a wonderful (and FOSS) collection of libraries and UI to make these projection and stitch-point calculations available on a per-frame basis. I use it to stitch panorama that I capture using my still cameras.
4 Likes
Are those Ewoks you can hear in the background�
statistically? OSHA.
2 Likes
But collecting statistics need not involve knowing anything, if itâs data they canât themselves use. Thatâs the paradox.
They add to the atmosphere.
Looks like quite an inspiration for the world of robotics. Make a bomb-disposal like robot, with one or two long arms. Use the arms for assistance when the tracks wonât do on their own.
Also, length of the video - six minutes. Time to wait for arrival of bureaucrat-approved loading mechanism, in the range of many hours. Money saved by the exercise, lots.
As long as it works. When things go pear-shaped, though, cost will likely be the last concern.
Just judging by the placement of the tiered platforms next to the end of the rail tracks seems to indicate that this is established protocol for getting it in to position.
That could certainly be the case, but I was speaking to @shaddackâs contention regarding govât oversight/regulation (aka OSHA in the US). That said, would it be so hard to build a freakin ramp?
cost will likely be the last concern.
perhaps for the workers.
of course, the cost of having to deal with a splatted employee and the subsequent freakout will also have been statistically estimated.
1 Like
Now⌠if a trampoline had been involvedâŚ
Thatâs why you get dead peasant insurance.
2 Likes
The coal industry has dealt with millions of human pancakes, crushed under collapsed mines, rolled over by loaders, broken by faulty elevators. Whatâs one more sticky piece of profitable fossil fuel?
This topic was automatically closed after 5 days. New replies are no longer allowed.
