|
 |
info@grayfoximages.com |
|
We currently have five RF lenses, two wide zooms and three telephoto zooms. The short end consists of a 10-20mm f4L and a R14-35mm f4L. On the long end, 70-200mm f4L, 70-200mm f2.8L and 100-500mm f4.5-7.1L zooms complete the tele trilogy. As with virtually all past lenses, we resolution tested each on acquisition, mounted on an R5 45MP camera, primarily to screen out bad samples. (For reference, the R5's 4.4 micron pixel pitch produces a theoretical maximum resolution of 113.6 lppm before aliasing sets in.) We have had two lemons over the years, one of which wasn't tested until blurry regions showed up, past the merchant return window, on photos taken during a one-time historical architecture visit. Lost images and lens repairs on me. Hence, always test. Test ChartFor these tests, the R5 was tripod mounted and electronic shutter was used on a 10 second timer delay to minimize shutter induced vibration. The test target used is an inexpensive USAF 1951 chart from Edmund Scientific. The USAF chart has long since been obsoleted by ISO 12233 and others. However, to fully realize the ISO chart's benefits one really should have specialized software such as Imatest, a high end (and expensive) program that can pull a lot from a test image, including MTF data. Given the cost of a suitably large ISO 12233 chart and Imatest, as well as the limited nature of our goal, we give the ISO chart a pass. For our purposes, the USAF chart provides quantifiable and repeatable resolution data, in line pairs per millimeter, across the full width of a 35mm sensor -- albeit in no way equal to the precision or accuracy of a permanently installed laser aligned and fully instrumented ISO 12233 lab test setup. Evaluation of which line sets are resolvable can be a bit subjective, especially in the corners, where things sometimes get a bit fuzzy -- quite literally so. Also, given the discrete nature of the chart's lines, granularity of calculated lppm is only about 7-10 lppm between each line set. Therefore all lppm values are considered very approximate and relative to each other only -- by no means absolutes. Wide zoom summaryWide zoom resolution tests more or less mirror Canon's MTF charts (here and here). Both are more than acceptably sharp in center of frame, and actually quite far out toward the periphery. However, resolution drops off quickly as one nears the wide edges and even more so in the corners -- not untypical of even the best ultrawides. Both appear to be just about as sharp wide open as they are stopped down to f8. The RF10-20mm lens resolution, center frame, is about 80 lppm across all focal lengths and apertures before readily visible diffraction compromises show up, starting at f11 and more pronunced at f16. Corners are a different story. Blur is substantial there at the widest apertures, likely owing to astigmatism (visible as a separation of the sagittal and meridional MTF lines). The astigmatism effect is especially visible in sagittally-oriented corner line blocks. There, sagittal lines resolve at 90 lppm and meridional lines only at a much fuzzier 65 lppm. Maxing out Capture One's Sharpness Falloff slider doesn't improve the 65 lppm corner value. At the longer end, corner resolution is closer to the frame center value, albeit still retaining a bit more fuzziness. Corner sharpness is consistent in all four corners, implying absence of significantly tilted optical elements. The RF14-35mm performs similarly to the RF10-20mm near center frame, with differences in detail depending on focal length and aperture. Blur in the corners is similarly substantial, albeit with considerable variation in different corners or as the lens is zoomed -- suggesting tilted elements in the optical path of this particular lens. But then again, the RF14-36mm is a significantly less expensive optic. Adding Sharpness Falloff and Light Falloff correction helps a bit with both optics. Test results below.
One additional note on image blur in corners. Processing via Topaz Photo removes much of the blur but doesn't increase the information available in the image. In other words, if the resolution limit for a corner is 65 lppm, that doesn't increase after sharpening with Topaz. However, the level of blur or fuzziness is substantially reduced -- which can only have a positive effect on perceived corner sharpness. But, this also requires judicious use of sharpening amount. Moving the slider too far to the right introduces artifacts in the form of halos. Experimentation is advisable. Then too, Topaz filtered DNG output increases file size five-fold. Use with caution. It may seem redundant to have both in the bag, but having purchased the RF14-35 first -- and having sold an excellent but twice as heavy EF11-24mm f4L to fund the project -- the siren call of the RF10-20mm was just too strong to overcome. Telephoto zoom summary As with the wide zooms, the two RF70-200mm lenses perform more or less equally, with frame center resolution up to 80 lppm depending on aperture, focal length and location within frame. Drop-off in the corners is consistent with Canon's published MTF curves. Adding a little Light Falloff and Sharpness Falloff correction crisps up detail in the corners a bit. The RF100-500mm, if anything, outdoes the two RF70-200mm zooms, delivering 90 to 100 lppm center of frame, focal length in the 100-300mm range. Given the presence of a sensor anti-aliasing filter, this is probably about as close to theoretical as it's possible to come. Given the slow maximum aperture of this lens at the longer focal lengths, diffraction effects necessitate shooting at or near wide open to achieve best results. Testing on an equal basis at 400mm and 500mm was not possible since my basement test range is too short to accommodate the full zoom range. Thus, the latter's results are subject to considerable uncertainty. Test results below, all values approximate.
As with the two ultrawide zooms, why two 70-200mm lenses? The reason is that they serve different purposes. The f4 is incredibly light and easily carried on walking outings. It also close focuses to a .28X magnification. (It also packs conveniently with two other similarly compact zooms in a ThinkTank Turnstyle 10 sling bag.) With extension tubes, it may well serve a dual purpose as a macro lens. The f2.8, on the other hand, is a classic portrait lens focal length and aperture. And, having timed both purchases to instant rebate sales the choice was easy. What's left out Noticeably absent from these purchases is a midrange zoom, in particular the RF24-105mm f4L. Having tested a loaner against my current EF24-105mm f4L (original version), I find that there is no advantage in resolution and a significant upgrade cost for a trade-out. I'd gladly purchase a better quality RF version, but oddly neither Canon nor any of other manufacturer seems interested in my willingness to pay more. Sigh. |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
© 2023 Michael W. Masters |