During my recent stay in the Bay Area, Stephen suggested I join the San Rafael-based DevTech team to record another ADN DevCast. Our hope was to cut down the length somewhat by focusing on a single topic, but those plans pretty much went out the window as soon as I started talking. :-S :-)
Thanks to Stephen, Gopi and Fenton for their warm welcome in San Rafael, and for keeping the session interesting with all their questions!
[We experienced a few technical glitches due to my system getting very close to its end-of-life (it only has a 90 GB hard-drive, so I’m forever uninstalling/reinstalling software and often run with just a few GB of free space available, which didn’t prove enough for Camtasia to record all we wanted it to). Stephen has done a great job of editing the issues out, but you may see the odd continuity quirk.]
You can also download the recording to view locally (37.4 MB), should you so choose.
These sessions are (very clearly) unscripted and I can see there may be a few areas that prove confusing to people. If that’s the case, please post a comment and I’ll do my best to clarify my intended meaning.
Brian Mathews, our VP of Autodesk Labs, kindly watched the video and clarified a few points regarding Photo Scene Editor that I felt were worth sharing. (In case it wasn’t obvious to viewers of the DevCast, I’m far from being an expert in the use of the tool, so I would recommend that people especially interested in the use of Photo Scene Editor check out the videos linked from its Labs page.)
Brian raised two main points:
- Use of the push-pin tool to create reference points does something more complex than snapping to existing points. In effect it densifies the point cloud at the location you have selected by performing further analysis on the images in the scene. Just the act of manually adding points creates a more dense and accurate point cloud at the selected locations.
- When adding points to my glasses, during this demo, it may have given the impression that the pin didn’t quite get the right spot in 3D. In fact the positions were accurate, but the “splats” that were draped onto the model gave the impression they were not. Shrinking the size of the spats would have made this more clear, as the centre of each splat would have been accurately positioned, just not the edges.