Originally Posted by climbdenali
How many do you need? What angles work best? Can water level vary? Etc. I have access to TONS of photos of Sunshine (tens of thousands), but if you have specific parameters that are ideal, it may be more appropriate for me to just go on a photo mission and shoot a few hundred from specific spots and angles. Only problem I see is that even now, at 500 cfs, and even as low as 250 cfs, there's a lot underwater.
I think more is better, I'm not exactly sure with geometry as complex as this. The more angles the better because it lets me fit everything together and try to capture everything more accurately. I think it would be better for the pictures to come from the same day, but since I'm asking others for help I am happy to take anything I can get. I think the lower the water the better, but I completely understand that at 250 there is still a lot of water in there; I know its possible to get a raft through in the 250-350 cfs range. If nothing else, having water will let me put in the deepest channels more accurately.
I'm trying to use AutoDesk ImageModeler. It can take a bunch of pictures and by linking different points in the images build a 3D model from them. Its supposed to be able to capture more complex shapes, ie the rocks, from the images, but I'm not sure how well it will work. This is why I think all the photos from 1 session would be better. I've only ever used the software on buildings. If ImageModeler doesn't work I'm planning on putting the pictures into AutoCAD, and tracing the structure of the different features from different angles. This will be more difficult cause I'll have to place the pictures for each feature, and then move them for the next feature. Once I have enough of them I'll be able to create the shapes that I'm looking for. For tracing in AutoCAD I don't think the water level matters, just the more I can see the better. I don't think highwater shots will be of too much use for me, other than trying to verify the accuracy of the model once I have it running in the computational model.
Thanks again for any help.