After reading Roger Cicala’s excellent article How to Test a Lens, I began thinking of how to apply those ideas to testing a fisheye lens… and then what the target ought to look like… and then on to meet Norman Koren’s fascinating page on Lens Testing. Both articles are a wonderful resource and Koren’s chart was used extensively on the 24″ x 36″ target that I developed for yesterday’s tests. Please feel welcome to download this chart for your own tests if you like (panocea target.png – about 20MB). It was printed on coated roll stock using my 24″ HP Designjet 510 at 600 dpi. I had originally planned to print it our on photo glossy paper, but even under a magnifying glass, looked pretty good on the C60198B stock, so didn’t bother wasting the ink or the paper. The target was lit using (4) 500-watt halogen work lights with the WB set at 2900°.
Assuming that the sensor to target distance spelled out on Koren’s page still applies to fisheye lenses, I proceeded to carefully set up the tripod at 1.360′, 1.706′ and 2.559′ for the three focal lengths tested using the EF8-15mm Fisheye; 8mm, 10mm and 15mm respectively. In the photo above, you can see the camera and Markins Q-Ball head perched on the adapter I made for the surveying tribrach with optical plummet. Using a plumb bob and 4′ long drywall framing square I laid out the positions on the floor and then kept the height of the instrument equal to the height of the center of the target while using a level to try and control the camera’s pitch and roll… crude but close enough? I’m not so sure after studying the results.
In all instances and at all focal lengths tested, auto focus of the 8-15mm was used. Ultimately, the idea from these tests would allow – or so I imagined – some measure to compare similar results using the EF-S10-22mm, fresh from the Canon Service Center just a few days ago. The focal lengths in the tests using the 10-22mm were limited to 10mm and 15mm.
In this particular side-by-side comparison we see 100% crops (maximize your browser window and then click on thumbnail) of the most upper left corner of the 10-22mm frame and its approximate corresponding area in the frame from the 8-15mm, both mostly as they came out of the camera save for a clip and saving as a JPG. Admittedly, in some aspects, such a comparison is unfair; however, illustrative of both lenses when adjacent to each other. One thing to note in this side-by-side, the CA from the 10-22mm tends to generally be along the edges of the frame whereas the lessened CA from the 8-15mm tends to be more evenly present across the frame.
A couple of things…
Doing this test was a total blast, but I’m still unable to conclude that the quality of the image from this L series lens is so vastly superior to the 10-22mm (like the ideal image dreamt of for nearly a year since the 8-15mm’s first press release). But then again, it’s certainly quite good and one that I’ll happily keep using. And I absolutely LOVE the wideness of this thing – it’s so cool!!
In an email conversation with Roger, he mentioned “defish”, something else never thought of since getting my first fisheye two weeks ago. What an interesting idea! More stuff to try and learn … and another experiment for another day