For the revision, I tried to construct the sole part(鞋底), taking TA Zi-Xin’s suggestions, in the process of photography, I additionally photoed two side-views as coherent features(連貫特徵) to make it more easily to align. Total umbers of the two-side view photos are 16+16, while the sole part is 32. Perhaps, 16+16 photos are not sufficient enough for the software to align the difficult part also technical problems in photography counts. Hence, it automatically anchored so many feature points at the background but the shoe itself. After deleting lots of photos(those which anchored incorrectly and most of them are about the sole part), the photo left is roughly the same as the original one I built. Therefore, I ended up using the 66 photos (version 1, the very first set) to construct the model, but the parameters are different from the original. It is worth noting that, in my case, when the align accuracy was setted as low worked the best for my model.
Comments