What if I shoot too many images? Will they help produce better models?

It turns out that the more images you have of the same surface, the better the reconstruction is for smaller details. The minimum number of images required by the math employed by photogrammetry in order to resolve a point in space is 3 images. Going beyond that 3 image minimum has positive effects.

These images were shot at ISO 50, 1/125, f/2. (Note I only used f2 since I’m capturing a flat object, and shooting totally perpendicular to the face, where the the demand for large depth of field is non existent). ISO 50 was used to ensure the amount of digital noise was kept to a minimum and didn’t play a factor in the reconstruction. Camera and lens used was a Sony A7RII with a Sony 28mm f/2 lens. Images were not processed, and RAW images were fed into the software.
I shot 25 images (3 rows of images with 8 images across, and moved parallel to the fence). Laterally I moved about 2-3 feet (the white board is 2 feet/24 inch wide) and probably vertically I moved the camera 1 foot for each row.
For the 5 and 10 images reconstruction I selected 5 images and 10 images respectively from the original 25 image set. I tried to chose images that spanned the area covered (ie, I didn’t just choose the first 5 and first 10).
All 3 meshes have the same number of polygons (500k) to maintain a level of consistency. Highest possible settings were used in PhotoScan (Highest for Alignment, Ultra High for Dense Cloud).
So far I have only tested this in Agisoft Photoscan, but plan on trying this in Reality Capture to see how that software treats this scenario.

There is a definite increase in mesh quality going from 5 images to 10 images, and depending on how much quality you’re looking for in the details level, by again doubling the number of images, you get another step in mesh quality.

For reference, this is the size of the object in the frame (it’s the white rectangular object in the middle of the frame, sitting on my fence). Relative to how small the object is in the frame, consider this to be a small detail in the object/scene you might be photographing for photogrammetry.