Only vertically and horizontally adjacent images should be required to match. Those are especially cross-matches of horizon and sub-horizon rows of images. Always try to remove foreground match points which have large error and are irrelevant for the task to match the horizon.
In the process, start with Optimizing Positions incremental from anchor, then go for view and barrel optimization, and so on. It should now be possible, by changing the values only for the anchor image and re-optimizing, to come to a panorama with only minimal error. Now you see “Yaw”, “Pitch” and “Roll” values of camera-to-world orientation listed in the photos list.
In Hugin’s Photos tab, select the Positions view on the right side. Now it’s time to estimate the quality of the panorama. Select the double-arrow tool to move the panorama via mouse drag and cursor keys over the grid, and align the outline of the photo horizon’s southern point with the measured line. , load your photo panorama, and then set layer transparency in the Layers dialog to about 50%. The process depends on processor speed, image size, your training and – most of all – your requirements in accuracy! In the GIMP, load your grid image with horizon line. We now have to iteratively bring deviations to a minimum. The optimal solution consists of a photo panorama which aligns perfectly with the measured line and features. Try to align the center of the image (where the geometric anchor has been defined remember: this should be the image pointing south!) with the measured horizon line or the distinct features. If captured on a tripod, they should still match.
Most important are good matches along the horizon, the images in the lower rows are clearly less important. 7.3 Panorama Postprocessing 85 Figure 7.6: Hugin’s Fast Panorama Preview can be used to check which images are connected to its neighbors.
0 kommentar(er)
