Identification of photo number effect for 3D modeling in Agisoft software

Abstract

At present, photogrammetry projects, including Close-Range Photogrammetry, has been supporting by software that works automatically. Agisoft Metashape is one of the Close-Range Photogrammetry data processing software that can work automatically. In accordance with ISPRS recommendations, the success of Close-Range Photogrammetry processing is determined by the overlap in each photo. Based on ISPRS, Close-Range Photogrammetry will succeed if each photo has an overlap of 90-95%. In line with ISPRS, Agisoft always recommends users to input more than enough photos in each processing, where more photos will produce better results. These conditions will make the number of photos used in Close-Range Photogrammetry processing will be very large, very long time processing and the processors required must be highly certified. This paper will analyze the photo number effect for 3D modeling in Agisoft software, as well as prove whether large overlapping values and a large number of photos will increase the accuracy of the resulted model. There are three simulations of photo number will be compared in terms of geometric quality, while the suitability of the resulting 3D model will also be discussed. Furthermore, the detection of causes failure in one simulation of photo number will also be explained in this paper. Results show that a large number of photos did not produce a better 3D model, so to create a 3D model does not require a large number of photos but the optimal number.