Do you have an idea for an awesome feature we should add… or hate the way we’re currently doing something? Share your ideas and suggestions here.
Videos give much more detail than a series of photos usually gives, unless the person taking the photos is exceptionally dedicated. A person could just slowly pan around with a video camera, and easily capture a complete scene. Maybe if photosynth could take video files, and convert the individual frames into separate pictures for a synth, it could be much more coherent?
Not every frame would probably need to be considered, perhaps one in 10,20, or 30 frames could be used, for purposes of efficiency?
There have been a few attempts at doing this with the product. There is freely available software that will create JPG stills from the frames of a video. Here's one example:
A common problem with this technique is panning too quickly causing motion blur on the frames.
Everyone talks about synthing a video... which I like quite a lot to be honest. To me, though, the beauty of Photosynth is that it honors photos so highly. When I hear people talk about synthing video it almost always seems that people are thinking of generating a very detailed point cloud.
What I have always anticipated and hoped for is that when we are able to synth videos, we will be able to combine them with photos in a single photosynth and that Photosynth will honor videos as first class citizens just as it currently values photos as first class citizens. That is to say I hope it will continue to value photos individually for their own merit and it will also value video as actual video and not only as a series of frames.
What excites me is thinking of watching a video move correctly through a synth of high res photos, providing Seadragon can keep up with all the loading.
I agree Nathanael, video synthing would be very cool and should also be a lot easier to compute than a random selection of images.
For anyone who wants to try the frame method: download Virtualdub from virtualdub.org and drag/drop your video in to the program. You will need the correct Video For Windows codecs before it can open the video, CCCP contains most of the ones you'll need so it's worth also installing that.
Go to the video menu, choose frame rate and put a number in the "decimate by" box (to skip frames). Use the Home and End keys on your keyboard and the scrollbar at the bottom to select a range, either choose the range you want, or use the DEL key to delete ranges from your video.
Finally go to the file menu and select "export image sequence", select JPEG as the format and choose a location to dump the frames.
Would me nice if they optimize something specifically for videos. Video could help where thereisno texture. Even combine some video to synth with few photos. You film mostofthe place, than take pictures just ofthe important details. The system uses the video to put the photos correctly in the space even withoul to much overlaps.