Testing Tokina AT-X 107 DX f/3.5-4.5 fisheye-zoom
Intro | Panorama | Color Aberration | Depth of Field | Coating/Flare | Links |

Here is my first quick test with Tokina fisheye-zoom lens on a Nikon DSLR with 1.5 crop factor:

The panorama was done with Nikon D70 with my standard setup for my Peleng 8mm (not optimezed for the Tokina): Nodal Ninja III set to same pivot-point as the Peleng and with the 8 stop ring . I took 8 shots around. Glad to finally have a wide-angle lens communicating with the camera (the Peleng gives no read-out so all have to be set and evaluated manually before final shooting!!) I decided to give Aperture Priority a try resulting in shutter speed from 1.6 to 1/13 secs @ F/11

Here are the 8 photos: 10 mm - F/11 - ISO 200 - Exposure Comp: +1.0 EV - WB: Direct Sunlight - mouse-over to see the different shutter-speed

 

The setup gave almost no problem in PTGUI and the photos stitched and blended exceptionel fine given the fact that the setup wasn't optimized for this lense. The fov came out at @ 166° with the most narrow part @ around 87.3° - se remapped image sample here
View fullscreen Panorama: QTVR - DevalVR (rather low-res 3900x1950 pixel made from 8 shot 1500 x 997 scaled down tiff. OBS: the other panos in the pull-down menu are done with Peleng 8mm!)

8 shots 87.3° wide is obviously an overkill (or perhaps not using Aperture Priority). Just didn't want to mess with my Nodal Ninja as I had a job to do the following day. 5 or 6 around + zenith and nadir should be enough. I'll be back with some more test with this later on when I got the time. I'll be very convinient if the pivot-point will stay the same for less shots. So that I can exchange the 8 stop ring with a 20 (4 and 5) or 12 (4 and 6) stop - and only have to worry about dust on the CCD shifting between Tokina and Peleng.

I thought that 5 shots tilted 10° + zenith would be fine.

5 shots tilted 10° + zenith => aprox. 16° ~ 18% min. overlap (mouse-over to see thes size and overlap)

But as indicated by the arrows the 10° tilt gives rather narrow overlap in some places. This could be dealt with by taking 2 zenith shots 90° apart. Or you could do it with 5 in level around + 1 zenith and 1-2 nadir.

This image indicates the overlap with 5 shots + zenit & nadir.


And for scenes with moving object 6 shots => aprox. 27° ~ 30% overlap would give more space for masking

 

Another option might be to rotate the lens/camera 60° so you have the full 180° diagonal going from zenith to nadir, similar to the Agnos MrotatorC# setup . Haven't tried it myself but theoretically a full 360° x 180° could be done with 6 shots, though it might be too tight and not worth the trouble.

----------

Flemming V. Larsen

1/2.5 sec 1 sec 1.6 sec 1 sec 1/4 sec 1/13 sec 1/10 sec 1/4 sec