How can I correct barrel/pincushion distortion?
PTools can for you - using a polynomial function, which will transform each pixel of your image in such a way that it is 'corrected' and barrel/pincushion distortion disappears. But the polynomial function needs to know four parameters a,b,c and d which are specific to your particular lens. You need to find them out in a calibration process, which PTools supports. But since you are not yet familiar with the scripting syntax it would be too complicated to explain now. I reccomend that you take a picture of a tiled wall and decide if you have barrel- or pincushion distortion. When photographing the camera should be positioned exactly horizontal and exactly parallel to the wall. Open the image in your image manipulation program and select Filter - Panorama Tools - Correct and find the parameters (they are the same for all three colors in most cases) of the Radial Shift dialog with trial and error. Note that negative parameters act against barrel- and positive against pincussion distortion. First try out with the b-parameter and leave a and c zero. d should be chosen so that the sum of a, b, c and d is 1 (a+b+c+d=1). Your picture is sufficently corrected, if all lines appear straight.
If you are more familiar with the scripting syntax you can calibrate your lens as described by Peter Murphy in one of his articles.
What is pitch/roll/yaw?
For creating panoramic pictures you need to tell Panorama Tools some information about your lens and its orientation while taking the individual images. Panorama Tools will then rotate and warp and tilt the images so that they fit together. Therefore you need to know what the following terms mean:
- yaw: is the angle you turned your camera between each image. It determines the horizontal position (along the x-axis) for each picture in the whole panoramic picture.
- roll: If you held your camera tilted left or right while taking one picture (try to avoid this), Panorama Tools can rotate the image back to vertical for you.
- pitch: If you held your camera tilted up or down, your images have to be tilted back to vertical if they are to fit into your panoramic picture. The pitch determines the vertical position of one image (along the y-axis) in the finished panorama. This ability to handle tilted shots even lets you create spherical panos though you do not own a fisheye lens. Great!
What is the HFOV?
HFOV: means "horizontal field of view". In this context we should distinguish between the HFOV of your lens and the HFOV of the final panoramic picture.
- HFOV of your lens:
It indicates how many degrees of the whole panoramic picture is contained within one image, which depends on your lens. Of course you need to take fewer pictures for the same panoramic picture with a wide angle lens than you would with a teephoto lens. For nearly any stitching program you need to know the HFOV. If you use a 35 mm camera you won't have any problems in calculating this by your own using the JavaScript calculator below. If you own a digital camera, the thing is not as simple, because the dimensions of the CCD imaging device in the camera are probably unknown. But often the 35 mm equivalent focal length is given in the user guide for the camera. So use this equivalent focal length for the JavaScript calculator below. Note that in this case this is just an estimate which will be more or less accurate.
If you are in doubt, send an email to the manufacturer of your digital camera since they should know (if you are unlucky enough to own a casio camera, don't even try to ask the customer service, they just don't know about technical details of their cameras and probably won't even answer you).
- HFOV of the final panoramic picture: To understand what is ment here see the illustration below:
Imagine you took three pictures for creating a pano. Between each image you turned your camera for 45 degrees and the HFOV of your lens is 80 degrees. Then the entire pano covers 170 degrees of the entire 360 view ( 2 times 45 degrees yaw and 2 times one half of your cameras HFOV gives you the effective view). Since PTools needs to know the approximate HFOV of the final pano you should note the yaw between the single source images.
What about image quality?
Keep in mind that always when an image is scaled or distorted, pixels need to be interpolated. This means that given pixel's properties are approximated to some extent by the software when the image is interpolated. Since PTools author Helmut Dersch is a mathematicican, the interpolator is the highlight of this software, exceeding most commercial products. PTools lets you select between four interpolation options providing good to excellent image quality. (Better options mean the interpolation process is somewhat slower).
Image quality can be lost further due to some compression algorithms. For example each time you save your image as JPEG file quality becomes poorer and poorer. Hence you should never work with JPEG files when you are saving your intermediate working steps. Use the PSD or TIFF file format instead.
