The 8mm PELENG FISHEYE for Astronomical applications

2009 Update : this lens is now very expensive (used to be at 150€, now close to 400€ !!), too expensive regarding its performance compared to new Sigma and Fujinon, so be aware when buying this lense that its price should not be above 150€. Mechanicaly speaking the lens performance is very poor, the rings are moving a bit, giving a very bad finishing impression.

This page is reporting some tests with the Russian 8mm focal length f3.5 PELENG FISHEYE that can be used for Astronomical applications. Regarding standard photographic applications, informations can be found in this excellent review carried out here. Also C.Buil's site here, using an EOS5 D with this lens.

The PELENG Fish-eye is a 8 mm focal lengh lens (Yes, 8 mm !) that provides a 24 mm circular 180° field of view. The image is not a rectangular image, this is really a 24 mm circle, outside this circle, no photon can be found !

A CHUNGARA KAF-6303 camera has been used to carry out the following test. A Chungara camera equiped with a 6.3 millions 9µm square pixel detector has been used first, as the detector to test this lens. Then a CANON CMOS detector has also being used.

Fish-eye Peleng lens mounted with a Chungara camera (Image 1).

Fish-eye Peleng lens mounted with a Chungara camera (Image 2).

Fish-eye Peleng lens mounted with a Chungara camera (Image 3).

The optical layout "looks like" to the following layout (from US patent 3734600) : Most of the fisheyes share the same optical design

Image 4

The huge front lens plays a peculiar role : it gathers all the light from all angles from +90° downto -90°, leading to a 180° field of view.

The next chart shows the spot diagram of such a lens : this would be the image of a star located at different fields from the center (IMA :0.000 mm).
As the field increases from the center (optical axis), the star image gets degraded, at 85° the star shape is pretty extended !

Image 5

Also vignetting is an issue, as mentionned in this excellent paper, this kind of lens exhibit vignetting that is decreasing when increasing the f-number. The next table compiles the vignetting performance of the Peleng lens, 100% is the light level at the image center.

f 3.5
f 5.6

This kind of vignetting can be corrected by the means of algorithms that knows the vignetting value for each pixel. This can be be calibrated once for all (not an easy task anyway).

The lens and the camera has been installed overnight, and the full moon is closeby to the zenith. Since the detector measures 27mm x 18.5 mm, and since the fisheye produce a 24 mm circle field, the circle is cropped on top and bottom. Nevertheless, on the horizontal direction, the image circle fits more than well due to the 27 mm detector width.

The next image is a 5 sec at f5.6. The full moon shines soo strongly, that it blooms over the whole image. Nevertheless, most of the constellations can be recognized ! It this case 1 pixel is 3.8 arcmin.

Image 6

The field deep is very large (8 mm focal length, don't forget !)and a black piece of paper can be dropped directly to the front lens surface, and be used to hide the moon (as moon mask in the next image !).

By clicking HERE, you will get a FULL resolution bin 1x1 image, this is amazing !! This a F/8 image, 20s exposure.


Image 7

A piece of the image has been selected (red rectangle) and according to the F number, a close examination of star sharpness can achieved. This is centered to the "Big Dipper" constellation wich fits more than well inside the rectangle, that extend to the zenith (Theses images have been recorded in January, arround mid-night).

Image 8 : At F 3.5 (full open) (binning 1x1)

Image 9 : At F 5.6 (binning 1x1)

Image 10 : At F 8 (binning 1x1)

By closing the diaphragme (F number increasing), the image quality on the edge (bottom of the images #8 to #10) gets improved. Nevetheless, at F3.5, the image is more than acceptable. On top of that, the center of the image circle (top of the images #8 to #10) is extremly sharp ! The FWHM measurment was impossible on star located closeby to the optical center ! All the star energy is confined within a single pixel !

Considering the price of this lens (may be found for 200$), compared with one from CANON, this is not so bad !

This king of lense can be used for monitoring clouds, contrails and locate any kind of disturbance that may affect observation quality.

Once the moon starts to dissapear behing the house, a large fog halo can be detected, whereas it was difficult to notice it by naked eye ! Some fog (including ligh pollution) can be seen closeby to the horizon.

This kind of tool is been used on regular basis at Paranal Observatory (MASCOT has been developped by S.Rondi), this page describe same kind of system using a Kaf400L CCD, the image diameter is about 512 pixels, whereas in this page, it has been extended up to 2600 pixels !!! I agree, this may be an overkill, nevertheless, for detecting all sky event, this tool could be "a must".

The images can be "de-fished" by a special algorithm (PRISM software has some, polar to rectangular transformation) to recover an image that is more "suitable" for watching .

PELENG 8 mm with CANON EOS 20D

For the next section, a CANON 20D has being used to evaluate this fisheye. For that purpose a 42mm to CANON mount adaptator has been purchased, because this lens is set for a 42 mm mount.

Once installed both lens and adaptator, images could be recorded and no major issues has been noticed. The only comment I could write is the following : the adaptator is slightly to thin, and requires the lens to be focused to shorter distance (with respect to the lens focus ring annotations) to get infinite focused. This is better than having the opposite situation, where the infinite cannot be reached because of a too thick mount adaptator. I won't complain !

The CMOS detector embedded to the CANON EOS camera is a 6.42 µm pixel square detector featuring 3504 x 2336 pixels (8.2 MegaPixels).
The detector is a 22.5 x 15.0 mm CMOS sensor. Since the fisheye provides a circular field of 24 mm, it is obvious that all the fisheye optical field could not fit into the CANON EOS CMOS detector.
It bearely fits only in the X direction (1 mm missing right and bottom). The Y direction misses much lot : 9 mm (4.5 mm top and bottom) to cover the whole field of view provided by the fisheye.

Nevertheless, it can be used as it, another test using the CANON 5D would have been nice !

The ISO speed has been set to ISO800 which I think personnaly is the best trade off to get a good detectivity versus signal to noise ratio.

So as to avoid star trailling due to earth rotation, "short exposures" has been used. Keep in mind that such a short focal length (8 mm) can lead up to expose as long as 12s because each pixel field of view is 166 arcsec The camera has been attached to a fixed tripod, that cannot compensate for earth rotation, and all images here been recorded with 20s exposure.

I was lucky because outdoor temperature was about -1°C, which helps a lot to get rid of detector dark current and hots pixels, so the image presented hereafter are all pure raws, no preprocessing applied.

Optical resolution

When the lens is used at F3.5, the images from bright stars have a kind of halo, that was not noticed when standard camera CCD has been used. This is somehow unexpected despite focusing optimisation !

F3.5 image

At F5.6 the images gets improved by far, despite a 2.6x light loss factor, compared to F3.5 !

F5.6 Image

The elevation circles have been drawn (manualy), 20° by 20° : the scale is regular over the field, except nearby the horizon, where the focal length seems to increase a lot, as well as distorsions, increasing distances also.

A black and white negative image is better to evaluate sharpness and limiting magnitude, in that case magnitude 7.3 stars can be detected, imagine that the light went thru an aperture of 8/5.6 mm = 1.4 mm (that's the diameter of the "telescope" used here !). Trial has been made at F4 and the image quality is really in between F3.5 and F5.6. Moreover, F8 test also achieved, image quality improvement has been noticed towards the image edge, but the exposure time needs to be increased !!

The next image is the same image as previous, except the constrast has been highly enhanced. This is convenient to watch light pollution extension effects, weak clouds and so on.

As a conclusion, despite his rough appearance and some mechanical weaknesses, this fish eye can be used for astronomical applications, provided F stop is set to 5.6, such as :

Have a look here about all the application that can be achieved !

Interesting link here also

Hereafter, the peleng have been used with a U9000 Apogee camera that has a KaF-9000 CCD inside (36x36mm 4096x4096 12µm sensor), the Peleng full field is used. Moon is just rising.



by C.Cavadore, Jan 2006, The author wish to thanks Mecastronic for lending the Fish-eye.