Animated comparison viewing of flare and veiling glare

on four popular fisheyes lens

Subject:

Four fisheye lenses were subjected to a cylindrical beam of hard light directly impacting on its front element.

A LED lamp was used for this purpose. It was positioned in order to cause the most effect when viewed in the viewfinder of a Canon EOS 5D on which the fisheye lens was mounted.

While informative, this test has a subjective part in it as it depends on the feeling of the test-operator (i.e. the author) to detect when the flare and veiling glare are at the maximum while moving and directing the beam of light in front of the lens.

The four lenses are:

  1. Nikkor 10.5 mm f2.8 "Shaved"from its lens hood and adapted(*) to fit on the Canon FF camera.
  2. Tokina 10-17 mm f3.5 "Shaved"from its lens hood. The focal length was set at 10.5 mm for this test.
  3. Sigma 8 mm f4.0: this older lens is still widely used for 180 x 360 panoramas (mostly fitted on a smaller APS-C camera though).
  4. Sigma 8 mm f3.5: this lens is more recent and has replaced the older model.

Results

The result shall be illustrated by means of four separate QTVR object movies.

The flare is shown at all the aperture settings and a final reference view shows the scene when the LED light is switched off (f/8 only for all lenses).

Nikkor 10.5 mm
Tokina 10-17 @ 10.5 mm
Sigma 8 mm f4.0
Sigma 8 mm f 3.5

Click on the images to view the QTVR. Then you may use the arrow key.

(*) Important:

The very strong flare that obviously affects the Nikkor lens may not be due to the lens itself. It could be that the conjunction of hood shaving, adapter type and (possibly) adapter modification may be the cause:
The adapter was made of bare machined aluminum. Its structure and the mechanical feature ("B" on this photo) that was here during the test could have possibly reflected stray light as the latter is made of bare stainless steel and it is located in a specific angular position in roll.
I use now a different system on a different (Elephoto) adapter that probably isn't prone to create such a phenomenon.

Michel Thoby

21 May 2008