Taking the respective street price of the camera, the total cost of the project is not very attractive...
While I think the IQ shall always be rather poorer than with other fisheye lens providing us with larger out pout panorama resolution, I also believe that "Double-Shot" panorama is interesting for a niche market. There is still unfortunately no adequate and affordable lens for "double-shot" panorama. Having in view this only purpose, my tests in particular of the much touted Sunex 5.6 mm and Sigma 4.5 mm were not really convincing principally due to the "limited" effective Angle of view which happens only to be slightly above 180 degrees.
Knowing this and wanting to keep this project within a reasonable cost envelop, I have decided to go another and different way than DSLR.
This is an illustration of the concept:
2 x Cameras : Nikon CoolPix P5100 (latest flagship 12 MPx P&S camera)
2 x Fisheye lenses : Nikon FC-E8 (fisheye converter x 0.21 with 190° effective angle of view: see below) can still be bought (April 2008) new from some vendors.
Note that this solution is not supported by Nikon that stopped selling this converter a long while ago;-(
2 x Adapters : AGNOS AFCE8NP5000 (Shaved: see below). Could be also Nikon's UR-E20.
2 x Remote mechanical shutter release system : home made
Support plate and mast : home madeParallax errors cannot be of course totally avoided. They are quite big but they are minimized with the selected accommodation and configuration. The horizontal planar arrangement of the cameras make the fixation on the mast really simple with an aluminum plate and two 1/4" screws.
The pivot point on the top of the mast is about 40 mm from both fisheye converter NPP for 90 degrees of incidence angle. I could not find a better camera overlapping configuration.
The CoolPix P5100 has only one electrical port connection. It's only intended for downloading images from the camera (internal flash memory and/or SD card) to a computer.
The CoolPix 5100 SD card compartment door is located on the bottom part of the camera near the threaded 1/4" hole (as is the LI-ion) battery: the user needs to access to the SD card compartment door and open it by sliding/rotation on hinges and extract the SD card to get the images in a computer. Therefore the corresponding part of the bottom side of the camera has to be free from obstacle and accessible to avoid necessary dismounting.
Taken altogether these constraints leaves us with only two possibilities where the two cameras are overlapping and with both of the bases in the same mounting plane. One of the two possible configurations (the one where the NPP offset is minimized amongst the two) is photographed in the image above.
The FC-E8 is specified by Nikon to give 183 degrees only. But this was true when it was associated with the CP 9XX series. I had noticed that the images from most recent CoolPix (e.g. CP P5000) that were posted on the Web were obviously wider in FOV than this specified value but still not wide enough to fit with my requirements. I have thus decided to verify this by myself and was procured with a CP P 5100 that is the current successor of the CP P5000. The dimensions of the CP P5100 CCD sensor are slightly larger than those of the CP P5000, but nowhere I could find information if the onboard zoom lens had been modified from one model to the next:-(
When the AGNOS adapter is used as is on the CoolPix P 5100, the resulting minimum angle of view (i.e. @ f/7.8) is about 185 degrees. And there is a blue fringe on the edge of the circle of the image.
I then built a test setup that allowed me to position the rear of the fisheye converter at any distance from zero (the rear of the converter thus touches the front part of the camera lens when it is set in Wide Mode) up to 12.5 mm (i.e. 0.5 inch). The less the gap is, the larger is the Angle of view: When the distance is nought, effective FOV is about 192 degrees!
I have made extensive experiment and found that the optimal position w.r.t. the CP P5100 lens is when the gap is about 1.5 mm. Then the transverse chromatic aberration is minimized, the blue fringe is very small and the sharpness seems to be the best while the circle has still a very large diameter. One should note that the JPEG image (no RAW available) is obviously heavily onboard camera corrected and especially for TCA and chromatic noise.
From the results of this test I have then deduced the modification that has to be machined on the AGNOS adapter in order to be as closed as possible to the best (optimized) solution for fitting the FC-E8 on the CP P5100:
Click on image to enlarge full scale
The resulting angle of view is then about 190 degree and the adapter is still useable for the other lens converter, including the Nikon TC-E2 (x 2) tele converter and the WC-24 wide angle converter..
The lens of the CP P5100 is therefore largely modified and improved compared to its predecessor and I am wondering why Nikon is carefully hiding this fact as the latest design is obviously better.
I have however noted slight discrepancies between the two specifications of the lenses in the Nikon's products brochures and initially thought of typing errors but I now wonder:
|CP P5000 Lens||3.5x Zoom-Nikkor; 7.5-26.3mm (35mm  format picture angle: 36-126mm); f/2.7-5.3; 7 elements in 6 groups; Digital zoom: up to 4x (35mm  format picture angle: 504mm)|
|CP P5100 Lens||3.5x Zoom-Nikkor; 7.5-26.3mm (35mm  format picture angle: 35-123mm); f/2.7-5.3; 7 elements in 6 groups; Digital zoom: up to 4x (35mm  format picture angle: 492mm)|
Note: I have ordered the genuine UR-E20 adapter (functionally identical to Agnos's) from Nikon France end of February 2008 and I am still waiting for a delivery (end April 2008) :-((
The project script template that was used to stitch the pair of images is also provided in the zipped folder. It was obtained from optimization by PTGui Pro 7.8 of two orthogonal pairs of DuoPix like images (4 images 90° apart).
5072 x 2536 pixels QTVR panorama made with the above sample images and template. IMHO this is much better than the Google Street View panoramas that I have seen up to now...
A long list of undesirable features (e.g. lack of electrical port for camera external control) can be overcome or marginally accepted. But the unability to synchronize sufficiently the shutter release on both cameras makes IMO this project a failure:-((
By contrast, using the CP P5100 + FC-E8 as a classical "Double-shot" system with a single camera on top of a precise 180 degree (i.e. two opposite direction) rotator and fixed tripod can be recommanded to the usual clientele for this double-shot technique (IPIX like) in particular because of the surprizing image quality and the amazing effective angle of view (190°) that can be obtained by an adequate shaving of the adapter. The NPP should be then set for 90 degrees of (beta) half angle of view. The resulting total absence of visible parallax error at the seam of the two images gives a nice, refreshing and reassuring impression to the panographer that uses a simple stiching template script.
Revised 26 April 2008