Alex Zhang's presentation on results of modifying the baffle on the HD DA 20-40mm Ltd zoom (
Hacking HD DA 20-40 for K1 - PentaxForums.com) inspired me to consider modifying the baffle on my own 20-40mm Ltd zoom.
I have another wide-angle zoom, the full frame film-era SMC Pentax-A 24-50mm F4 zoom lens (
SMC Pentax-A 24-50mm F4 Reviews - A Zoom Lenses - Pentax Lens Reviews & Lens Database), that I got not too long after getting my K-1 camera. Before committing to modifying the baffle on the 20-40mm Ltd lens, I removed the baffle from the 20-40mm Ltd (as Alex says, this is relatively straightforward to do - I used a #00 JIS screwdriver - see page 2 of his thread, post #25 for pictures) and then ran both lenses through some lens tests to see how they compare.
After testing both lenses I've decided to go ahead and modify the 20-40mm Ltd baffle, as soon as I figure out the safest way to remove the central part of the plastic baffle. I wish it were metal, I'm leery of filing or cutting or dremeling thin plastic. Advice on how to proceed is very welcome!
I already had the SMC A 24-50mm when I bought the 20-40mm Ltd lens, and I like it quite a lot. I bought the 20-40mm Ltd lens, covering pretty much the same zoom range as the lens I already had, for several reasons: Water Resistance, silent autofocus (my eyes aren't getting any younger, manual focus only works well for me when using a tripod and live view), light in weight, small, and covering a range I happen to like a lot (the last three reasons being the same reasons I got the SMC A 24-50mm lens in the first place). Not to mention the DFA 21mm lens was only a future plan when I bought the 20-40mm Ltd, and currently a bit out of my budget.
Here is a summary of my findings:
Lateral chromatic aberration: This is the sort of CA that makes color fringes along tree branches against a bright sky along the edges and in the corners of images. Previously I had checked both lenses at 40mm (right after I got the 20-40mm Ltd lens), using a home-made test chart, but checked again, this time with shots of a black metal lamp against a backlit white curtain, checking at 24, 28, 35, and 40mm. I kept the aperture constant at F8, as lateral CA is not something that improves when stopping down (
Lens Rentals | Blog). At all the focal lengths the HD DA 20-40mm Ltd zoom had less lateral chromatic aberration than the SMC A 24-50mm F4 zoom.
Out-of-focus (OOF) longitudinal chromatic aberration: OOF LoCA affects the center of the image as well as the edges: This aberration is most obvious when the lens is slightly defocused on high-contrast scenes with sharp edges, and produces green and magenta color fringing. Neither of the two zoom lenses shows OOF LoCA, whether aimed at a Seimens star on a test chart (where the two zooms showed absolutely minimal signs of OOF LoCA, not enough to ever show in a real-world scene) or at my standard real world scene: a close-up of a wall switch with black lines on white plastic. For comparison, my 50mm "M" F4 and HD 100mm F2.8 macro lenses also are free of OOF LoCA, but my SMC 77mm Ltd and SMC A 50mm F1.7 both color-fringe wildly.
Distortion: Based on various shots of tiles and perforated boards, both lenses display a bit of barrel distortion at 24mm, decreasing as the focal length increases. The 20-40mm Ltd seems to have somewhat less barrel distortion, but I think the two lenses are about the same. Once I set up a suitable test chart I'll check more carefully.
Monochromatic aberrations, specifically coma and astigmatism: Wide open (F4), the SMC A 24-50mm lens shows a rather spectacular coma/astigmatism combination in the corners of the frame at all focal lengths, displayed as angel/bird wings spreading out from point light sources such a string of Christmas lights in a darkened room (placed five feet from the camera, also place about 15 feet from the camera). The same thing happens with street lights across and down the street placed in the corners of the frame. By comparison, the 20-40mm Ltd shows essentially no coma/astigmatism for the Christmas lights. I haven't yet tried shooting street lights at night with the 20-40mm Ltd.
Star bursts: On the one hand, the 20-40mm Ltd with its 9-rounded-blades aperture is not especially prone to making star bursts (
HD 20-40 Ltd Starburst - PentaxForums.com), though it is possible to do. On the other hand, stopping down the SMC A 24-50mm zoom quickly elimates the coma/astigmatism and starts to produce very nicely defined 6-pointed stars. Sometimes I do, and sometimes I don't want star bursts, it's nice to have a choice of lenses. (As an aside, the Pentax review page for the SMC A 24-50mm zoom lens says this lens has 8 blades. Various other sources around the internet also say 8 blades. My own copy only has six blades, various copies I've seen on ebay have six blades, and various night scenes I found on flickr show six-sided star bursts. Maybe there are 8-bladed versions but I haven't found any.)
Decentering: Checking for decentering using Cicala's Seimen's star test (
Lens Rentals | Blog), neither lens is decentered.
Vignetting: For each combination of focal length + aperture that I tried, the HD DA 20-40mm Ltd zoom (without the baffle of course) has *less* vignetting than the SMC A 24-50mm F4 zoom. Here is how I checked for vignetting:
* The target was a backlit double layer (separated by an inch for diffusion) of white cloth (there being no evenly lit walls anywhere in my house).
* I shot raw, with the camera on a tripod and the lens within a couple inches of the cloth to render the "scene" maximally out of focus, using aperture priority, ISO 400, a constant manual white balance, and "spot" metering to avoid metering the vignetted portions of the "scene".
* I shot both lenses at 24mm, 28mm, 35mm, and 40mm, at F4, F5.6, F8, F11, F16, and F22 (the zoom range common to the two lenses) and from F4 through F22 (the aperture range common to the two lenses).
* The raw files were rendered at the command line using this command:
dcraw_emu -v -w -H 0 -p /usr/share/color/icc/pentax-k1-matrix-built-in.icc -o /usr/share/color/icc/sRGB-elle-V4-g10.icc -b 0.9 -h -6 -T -g 1 1 *.pef
which produced half-sized auto-white-balanced 16-bit tiffs for which the brightest pixels were uniformly set to 90% of solid white.
* I opened all the images in GIMP as layers, and reduced their size by half and by half again using linear scaling, so that any one pixel actually represented something like 32? 64? (math is not my strong suit) pixels in the original raw file.
* Then I put sample points in the middle, the upper right corner, and halfway between the middle and upper right corner, and compared results from the two lenses. The sample points were set to read out in xyY, where Y shows CIE XYZ luminance. In the center, all the layers read essentially the same Y values. But the other two sample points always had lower Y values for the SMC A 24-50mm zoom, compared to the 20-40mm Ltd zoom at the same focal length + aperture values.
Field of focus: I checked the field of focus for both lenses at 24, 28, 35, and 40mm, all at F4, using a "close-up" version of Cicala's "grassy slope" field of focus test (
Lens Rentals | Blog). I focused the lenses pointing down at an angle at a table top covered with a heavy cotton canvas cloth, with the lens about three feet from the center focus point on the table. Both lenses have more or less symmetric fields of focus at all tested focal lengths. Comparing the two lenses, the 20-40mm Ltd field of focus is more consistent over the various focal lengths, with a gentle curve backwards at each side. The SMC A 24-50 field of focus has a very slight mustache wave back and forth across the center line (all 3 of my macro lenses including the 55mm micro-nikkor have the same gentle mustache wavering), and almost seems to tilt up a bit (not much) on the right side at 35 and 40mm. I've tested a couple of my other lenses at close-up and also more normal distances, and the field of focus is about the same for these other lenses regardless of how near or far away the target area is, and with the same general shape over all apertures (same general shape but covering more area as the aperture gets smaller). Which doesn't mean the same will be true of my two zooms. Now that it has finally warmed up a bit and stopped raining outside, I'll find a nice grassy field and do some additional testing. But it's nice to see how uniform each of the two zooms is at the tested aperture over all the tested focal lengths.
Center and edge sharpness: I had previously compared the 20-40mm Ltd (before removing the baffle) to the SMC 24-50mm lens at 40mm using shots of a home-made test chart. Everywhere that vignetting was low enough to compare the two lenses - everywhere except the corners - the Ltd lens was sharper at 40mm than the SMC lens. I haven't yet done test chart shots comparing sharpness with the baffle removed (after several months of storage the paper wrinkled on my first home-made test chart). But I did take shots of a 60"x60" piece of cotton canvas suspended vertically on (and clipped more or less tightly to) a frame, with the lens aimed straight at the canvas, 3.5 feet away from center of the canvas, at 40, 35, 28, and 24mm; at F4, F5.6, F8, F16, and F22. For most focal length/aperture combinations and especially at wider and middle apertures, both in the center and along the (non-corner) edges, the 20-40mm Ltd was at least as sharp as the SMC A 24-50mm lens, and often noticeably sharper. However, when stopping down to the smaller apertures diffraction seems to take a toll on the 20-40mm Ltd sooner than it does on the SMC A 24-50mm.
Corner sharpness: At 35 and 40mm and at wider apertures, the 20-40mm Ltd is sharper in the corners than the 24-50mm SMC lens. At wider focal lengths and narrower apertures, the 24-50mm SMC is sharper in the corners than the 20-40mm Ltd. But for all focal length + aperture combinations *both* lenses produced shots of the canvas that were soft and smeared in the corners, to the point where some of the shots had essentially no detail at all in the corners.
Initially I attributed this lack of detail in the corners to the lens being a mere 3.5 feet from the center of the canvas, putting the captured corners considerably farther away than the centers, and especially so at the wider focal lengths. To test this theory I made similar shots (same distance from the canvas) with (1) a Sony 28-70mm lens on an A7 camera (at 28 and 40mm); (2) 21 and 43mm Voigtlander "M" lenses adapted to a Nikon Z6 camera; and (3) Pentax SMC 50mm F1.7, SMC 77mm Ltd, and HD DA 100mm prime lenses on my K-1 camera: *All* of these other lenses had respectably detailed corners when photographing the canvas at the same distance of 3.5 feet from the canvas, though of course softer in the corners than at the center. To double-check I repeated the shots for the 20-40mm Ltd and the SMC 24-50mm, with the same result: soft, smeared corners for both zoom lenses.
Fortunately, moving the two zoom lenses to a distance of about 5 feet from the hanging canvas and shooting photos of the canvas (as mentioned above, this time in a darkened room with Christmas lights strung across the canvas so I could check for coma and astigmatism) did restore respectably detailed corners, still softer than the center, of course, and still smeared (stretched out in the corners, pointing towards the center of the image), but not smeared or softened to oblivion at any f-stop/focal length combination.
Flare and glare: something I haven't checked yet, probably should.
Bokeh, overall image quality: Bokeh is important to me, not in the sense of wide aperture lenses producing razor thin depth of field (not my taste, not my style) but rather in the sense of smooth out-of-focus highlights and smooth, gradual changes from in-focus to out-of-focus areas when available light or aesthetic choices dictate opening the lens up a bit. I've seen nice bokeh and great images from both lenses. I haven't taken nearly enough images of my own with either lens to know which one I might prefer and for what specific reasons (other than autofocus and WR), so that's the next step.
Originally posted by eles 
Post #1 by eles
Similar Threads 
