The software I use is called Photomodeler. (Check it out.). It generates very nice 3D models that you can import into Maya, Blender, CAD, or GIS software. It costs about $1000 USD, and requires a few days to learn to use. Almost any camera can be used, even historic photos.
However, a few things will help generate better models for you.
- First, use the highest megapixel digital camera you have access to, and use it only at one zoom setting (if possible). All cameras require some calibration, and each zoom setting requires its own calibration. For my software, usually all this requires is a single photograph of a piece of Letter sized paper from a forty-five degree angle. If more than one zoom setting is used in your project, you must re-photograph it at each setting. Ideally this should be done with each individual camera used, but generally one picture for each specific model of camera will be okay.
- Next, when photographing your subject, try to shoot in overlapping rings. Shoot inside and outside of all surfaces. Be creative. Small cameras can fit inside tiny spaces.
- One way to get good outside shots on a hot dog budget is to use a camera monopod and the self-timer. From each camera location, shoot once from a natural height, and then set the timer and hold the camera high over your head on the monopod. Sometimes you will have to re-shoot, since it is impossible to frame the photo this way. It looks silly, but it works great.
- Make visible targets that can be seen from multiple photos. If you are thinking about doing a project with me, contact me and I will send you some stick-on targets that will make the process smoother and more accurate. If you are doing the processing yourself, or do not have software that uses coded targets, use some visible anchors to make sure you cover the whole surface. I like golf balls for ground surfaces, and stick-on dots for vertical surfaces. You can also use pin flags, driveway markers, thumb tacks, road cones, etc… choose whatever seems appropriate for your working environment. Some software can automatically identify targets like round dots.
- If possible, get GPS or Total station locations for a couple of these markers (and make them distinct). This is not necessary to produce a model, but it is necessary if you want to integrate the modeled object into a larger reality.
- At least one object in the photo array needs to be of a known length. I use a 2 meter stadia rod, aligned N-S magnetic in most of my projects (example). If this is a problem, I can often use some known object in the photograph for scale. For example, the diameter of a tire of a known vehicle can be used for scale. In archaeological excavations, most of the time a scale is not needed because of the visible grid.
- Last, shoot a couple pictures without any markers visible for use as surface textures in the final model. Try to get these with good lighting and without weird shadows. Some model types (wireframe) will not use surface textures, but in others quality textures make for a high wow-factor.
- Video is also an option, but generally yields a lower quailty result. Video frames have a very small pixel count. However, it is easy to film lots of angles and cover a scene well using a video camera. Try to avoid using the zoom, however, as this is a problem for calibration. You can also use historic photographs, or photographs taken with a film camera. Once again, calibrating the camera is a problem. It is possible to use these sources, but usually takes considerably longer to process.
Good luck. If you have any questions about photogrammetry for archaeologists or anthropologists, feel free to contact me.
