I’ve been playing with Autodesk’s 123D “Catch”
It’s a free app that transforms sequences of still images into a 3d model of an object.
Here’s a sample video I did:
It does the processing “in the cloud” – in conjunction with a client-side application.
It is -very- beta (really alpha) – has some probs – but is still quite cool.
Take a lot of shots in a continuous sweep around the object
Don’t jump around when shooting
Try to get lots of overlap between photos (move around the object maybe 15 degrees per shot)
After you’ve rotated around the object once – you can move a bit higher / closer on a second pass around for more detail
Somewhere between 25-60 shots total seems optimal
Try to keep lighting constant
Don’t use a flash or HDR
I’ve generally found that Catch either comes up with a good model the first time around or it doesn’t. I have had very bad luck trying to manually adjust things if Catch didn’t figure it all out the first time. My advice is if you didn’t get good results – redo the photo shoot.
Couldn’t get direct youtube upload to work
Video export seems somewhat broken (most codecs resulted in unplayable videos)
Was able to get a useable export by choosing “Raw” as the video codec – then knocking the resolution down to 640×360.