Kuijper, ArjanTausch, ReimarDomajnko, MatevzWempe, Leon JuliaLeon JuliaWempe2023-02-022023-02-022022https://publica.fraunhofer.de/handle/publica/435486Focus stacking is a frequently used technique in photography [6]. Also, this technique can be applied in product photography to extend the depth of field in the images. In product photography, the object is captured from multiple views. In my approach to focus stacking, I want to use this additional information from multiple views to improve the focus stacking of every individual view. The method consists of three steps. First, the image is stacked with an initial stacking that only uses pixel information. Second, this initial information is used with Graph Cut [2] to improve the sharpness selection. In the third step, all images are combined with NeuralRadiance Fields (NeRF) [10]. The evaluation shows that Graph Cut improves the sharpness detection compared to the initial detection. And I show that NeRF is not a suitable model for focus stacking because it fails when the images have artefacts generated by the focus stacking.enLead Topic: Digitized WorkLead Topic: Visual Computing as a ServiceResearch Line: Computer graphics (CG)Research Line: Computer vision (CV)Camera calibrationMulti image stacksDigital imagesmaster thesis