We recently captured thermal images of a building using the H20T camera, and I’m looking to use the zoom camera images to create a 3D model. While I understand the H20T isn’t ideal for photogrammetry, I’d like to experiment and see what can be achieved. If we can develop a solid workflow with decent results, we may consider using it for photogrammetry on some projects when the P1 camera is unavailable.
I have two sets of images, one with 2x zoom and another with 5x zoom. Could you advise on the optimal settings, such as distortion parameters or any other recommendations, for this test?
Hi @Rango986,
we don’t recommend to use zoomed images for modelling. So, I suppose there won’t be some optimal settings for the process.
I would keep the pre-defined settings to see the firs result of your inputs. Even before that I would use grouping of the images, as you are using two different types of them.
Hi, thanks for your quick response.
I understand that it’s not recommended, and I’ll try to minimize its use. However, I did get decent results with smaller datasets and wanted to explore its potential further.
Could you clarify what grouping does? I see it involves sorting by EXIF data, but how does that affect the processing?
Yes, as you wrote. It will sort the images to the groups according the EXIF data. The images from these groups share the same calibration and lens distortions settings. Keeping so different images in one group can bring some mesh issues, so using this settings you are trying to avoid them.