Sigma 8mm Fisheye Canon 350D MrotatorCP
From PanoTools.org Wiki
Revision as of 23:01, 28 March 2005 by 220.127.116.11
- Sigma 8mm f/4
- Canon 350D
- Agnos MrotatorCP
- With the above combination I can shoot a panorama in either six or three shots with yaws of 60 or 120 degrees respectively.
- For the three shots process, position the camera 4mm backward in relation to its calibrated position for the six shots process.
- If your aim is the three shots process (a time winner for cheap real-estate service), you will be better served with a Nikon D70 for which the MrotatorC was originally designed. It has a larger sensor, which gives you more overlap. The overlap of the images with the Canon 350D is extremely small.
The process that works for me (with PTgui):
- import source images
- set approximate lens settings
- set panorama settings
- the crop is automatically good
- set image parameters
- distribute images by yaw (thank you Joost for the Fill yaw function!)
- roll all images -60 (the Agnos MrotatorC holds the camera at 60 degrees to use as much possible of the sensor surface)
- set approximate pitch (thank you Luca for the great idea of orienting the camera slightly up, giving better coverage of the zenith)
- set control points (i add at least two vertical lines)
- first advanced optimizer run:
- pitch on all images, including anchor image, linked
- all lens parameter excluding horizontal shear (g)
- important: no optimization of yaw and roll. The MrotatorC is solid and precise. I shoot with a wired remote and found that the positions are very consistent
- second advanced optimizer run: if necessary, optimize a second time for all parameters including yaw and roll
- blend and stitch, serve hot with either a Java applet or QuickTime
Can be stored in PTgui's lens database
- lens type: circular
- HFOV 182.772
- a: -0.07611
- b: 0.003985
- c: -0.041765
- d: 0
- e: 0
- g: 0
- t: 0
- the above process works with my guinea pig, a little 5x5 (1.50mx1.50m) shower/wc room, as well as with some larger rooms.
- I did not have to run the second optimizing run
- Hypothesis: since the entry point of the lens changes with the angle, for a perfect, uncorrected stitch exact slices should be used, like an orange. To be verified.