Jman13
Online
• • • • • •
Upload & Sell: On
|
 p.2 #12 · Optical Limits reviewed Canon RF 16mm f/2.8 STM | |
I took a shot with my 16mm of roughly an object that size, and focus distance is approximately 0.7m, so fairly close. Certainly not closer to infinity than not. With that said, I haven't noticed it being a huge difference in corrected images based on focus distance, as most of the resolution loss is due to the distortion correction, with some minor softening due to the optics.
However, there is a notable difference as to what converter is used, and therefore what lens profile is applied. The profile for Lightroom is not great...it overcorrects the image, such that there can be a bit of moustache at the extreme corner. Your distortion tests don't show this, so I'm not sure whether you are using the JPEG to measure corrected distortion or, or something like DPP.. I'm assuming your resolution tests are based on RAW files, so I'm curious what program you use for conversion.
In any case, there is a notable difference in apparent resolution when using Lightroom vs something like DxO Photolab, which does a much more complete correction. (in the crops below it looks angled still as if there is uncorrected barrel, but this is due to tilt of the camera.). While there is still the inevitable resolution loss due to stretching those corners so severely, I think it is much better handled in DxO, and while not 'good' is certainly more than usable for the vast majority of photographic situations.
100% crops of extreme upper right corner, LR conversion with distortion profile correction on left, DxO PhotoLab corrected imageon the right. Focus distance here is about 3-4m.
http://www.jordansteele.com/2022/lr_dxo_test.jpg
Interestingly enough, both Lightroom and (especially) DxO have SIGNIFICANTLY wider fields of view than what Canon expects you to use for this lens. The DPP / JPEG output, which is at a 16mm FOV, is significantly cropped in from what the corrected image on LR is, and even moreso on DxO, which actually provides a corrected near 16:10 output, giving you cropping ability on the long end beyond the traditional 3:2 frame. Here is the same file as above with the DPP profile from Canon, and then the DxO converted profile cropped to the same FOV (this results in a 7482 pixel wide image, or 37MP image, which is still notably sharper than the DPP corrected output. At this crop, the corners actually look pretty good, though you lose ultimate resolution.
http://www.jordansteele.com/2022/dpp_dxo_test.jpg
If you want to see just how much wider a DxO corrected image is vs. what Canon considers the final output, see below.
http://www.jordansteele.com/2022/dxo_dpp_size.jpg
Uncorrected, the image is 13.5-14mm in width. The DxO correction gives an image horizontally at about that same 14mm width (it's very slightly wider than my Samyang 14mm f/2.8, which is a bit wider than my Sigma 14-24mm itself). If you crop to 3:2 on that image, and assuming the Samyang 14mm is exactly 14mm, the 16mm corrected with DxO yields an image with a focal length of 14.95mm. However, based on comparisons with other 14mm lenses, it is likely the Samyang 14mm is actually closer to 13.5mm, which would make the DxO corrected 16mm lens final output the same FOV as 14.4mm. In any case, it's a wider lens than Canon marks.
Looking at a landscape shot with the wider FOV of LR and DxO, you can see the resolution difference between the two a little more clearly in the corner. Cropped to the DPP output, you'd still see even better corner resolution due to cropping out the worst parts of the image.
Full shot:
http://www.jordansteele.com/2022/cedar_16.jpg
100% crops of the lower left corner, LR and DxO (respectively):
http://www.jordansteele.com/2022/lr_dxo_test2.jpg
Of course this shot doesn't really need distortion correction, and you can see if you turn it off, the corners become quite decent:
http://www.jordansteele.com/2022/lr_dxo_uncorrected.jpg
In any case, I'm quite pleased with the lens overall, as the poor resolution corrected is only really in the extreme corner, and even then it's OK for most situations. It's pretty good across most of the frame, and has excellent color and contrast. For the price and size, I think it's a winner.
|
OM lenses can vary. I know some of the 21mm OM popular to adapt, probably later mc (rightly or wrongly ) and had a commensurate high price. So I didn't stick with the video long enough to see the version but I doubt they came to be high priced if they had bad corners. Mine isn't particularly good on digital, either (possibly in part due to a canoeing incident, another story I suppose)
And our methodology completely eliminates field curvature. What you see in our MTF charts is the maximum resolution a lens achieves at that spot... in our camera based lab tests. Yes, lensrentals uses a completely different approach with an optical bench. Which, you could argue, doesn't tell you anything how a lens behaves on different cameras. The camera is always part of the equation... both in testing, as in the field.