Creating a 3D model from different resolution images

Hi, almost complete newbie here, ODM looks awesome! I have a large research project using aerial imagery, of varying resolutions. I’d like to turn it into 3D and was wondering what’s the lowest res that would be possible to do this? My range is 0.75 cm to 7.5 m (originals), which I’ve resampled to 1m x 1m to standardise. What would be the drop off in quality across this range? Thanks!

Welcome to the ODM Community Newkid.

With 1000X range in image resolution across your images it would be difficult to get a good result.
7.5m resolution sounds like satellite imagery, resampling to 1m is going to make them look quite blurry, and even if you resampled the 0.75cm GSD images up to the same 1m, finding common features to tie them together might be very difficult.

How much area does the imagery cover, how many images of each resolution, and also what overlap is there in each set?

1 Like

Hi - just to be clear, I need to make one 3D model for each site, and each site has different resolution images. So I’m not trying to create a 3D model from mixed resolution images, as each site is independent.

In that case, the linear resolution of the 0.75cm GSD images will be 133X worse.

I doubt there is anything to be gained by resampling 7.5m GSD images to 1m GSD. The result will be so blurred that you’ll want to zoom out a lot.

Are the images suitable to produce 3D models? You need lots of overlap, and preferably some non-nadir imagery, more if there are complex shapes in the area of interest, such as trees.

What area are you trying to model with each dataset, and how many images in each?