CompuServe Thread

Photomodeler

5 messages in this thread
#149995From: EDWARD HARVEYJan 29, 1995 11:23 PM
Has anyone tried Eos Photomodeler? If so: Does it work? Is it worth it? Any particular problems or concerns? Thanks, Ed
#150963From: Cal HainesFeb 4, 1995 12:25 AM
Ed: I have EOS Photomodeler. I bought it about two months ago and haven't had time to try to use it for anything serious. I ran through their $25 demo/tutorial version and was impressed enough to sink about $3k into the software and a negative scanner. The only concerns are as follows: 1) Highest accuracy is achieved with a calibrated camera, so if you are working with someone else's photos — as is often the case in our business — you may not get optimum results. You calibrate the camera by taking 8 photos of a special target, marking four point on each, and then letting the computer crunch for a couple hours (not kidding). 2) Photos need to be taken using a camera with a known focal length — zoom lenses are out. 3) Getting the photos into the computer is a problem. Using a flatbed scanner with prints introduces another lens (the enlarger's lens) and reduces accuracy. The best method is a film scanner, like the Microtek, Nikon or Polaroid. Next best is Photo-CD. The problem with the latter is that you don't get accurate positioning of the scan relative to the geometric center of the negative. The work around is a piece of plastic that you super glue (not kidding this time either) into your camera. I didn't want to do that to nice cameras, so we went the film scanner route. With the right film scanner you just scan the entire negative and then digitize the left side corners of the negative mask, this allows the software to locate the exact center of the negative. On the plus end, the people at EOS are very knowledgeable and helpful. We've been using an in house app to do 3D-photogrammetry, but can't claim near the results that Photomodeler offers, so I can say with conviction that photogrammetry IS THE WAY TO GO. If you like I'll update you when I've had a chance to really put it through it paces. In the mean time, you might want to get the demo version. Hope this helps! Cal 4) -Cal Haines
#151044From: EDWARD HARVEYFeb 4, 1995 2:43 PM
Hello Cal Thanks for your reply. I have access to a film scanner so thats not a problem. From their ad I understand that reference points must be made on the images. Is this a matter of a few points or are many points required? I guess Im trying to find out if extensive work is required after the images are imput into the program in order to get a detailed mesh. Also is there any way to control the complexity of the mesh. Thanks so much Ed
#151422From: Cal HainesFeb 7, 1995 12:11 AM
Ed: >> trying to find out if extensive work is required after the images are input into the program in order to get a detailed mesh << probably… Photomodeler doesn't generate a mesh. Output is 3D points, that can be connected by polylines or 3D faces. Lines automatically connect points in the sequence that you mark them. If I recall correctly, 3D faces can be then be built between points and additional lines added. Would tend to work best with objects that have a lot of straight lines. You need something like AutoCAD that can use points marked on profiles to generate surfaces (via EDGESURF or REVSURF, etc.). I haven't used it enough to come up with lots of clever techniques. One thing is clear — Photomodeler has more promise to get 3D point data into the computer than anything else that I'm aware of. Particularly if your time gathering data is limited for some reason. I think that they're working on extensions that will automatically recognize and mark targets in the photo — they already do that with the camera calibration utility. Anyway, its a product that bears watching. Cal -Cal Haines
#151425From: EDWARD HARVEYFeb 7, 1995 12:47 AM
Thanks Ed