Review: Canon RF 16mm f/2.8 STM by Christopher Frost

LogicExtremist

Lux pictor
Sep 26, 2021
173
106
Wow, we probably have never seen such amazing analysis and page long essays regarding the auto-correction of a lens!

Thank goodness, as a photographer, all I care about is the final image. If the Digital Picture's Image comparison tool is reasonably accurate, and the corners are indeed considerably better than the EF 17-40L and not too much worse than the EF 16-35L, then that is good enough for me.

Just curious, but to all those that are despairing about the auto-correction, do you not do any distortion correction in post-processing? Keystone or perspective corrections? Do you ever use the new image enlarging programs (such as GigaPixel) or comands in Photoshop or any other software? Those enlargement programs or commands create a great deal of new pixels, does that mean you won't use them? Or does doing the same thing the camera does on your computer make you forget all about your concerns?
As far as I know all lenses use distortion correction, that's not the issue. What is catching peoples attention here is the crazy amount of distortion of this lens that is only possible because it's being used on a mirrorless system, which can hide it because it does on-the-fly image correction when looking through the EVF and makes the same corrections in camera when shooting JPEG images. This level of distortion would never be seen on a DSLR with an OVF, as it would produce a severely warped image viewing through the lens.

The are three main problems:
  1. The lens has heavy vignetting, the corners of the uncorrected image are pure black, so no amount of shadow recovery will bring those details back. As such, they need to be cropped away, which by most estimates is thought to remove around 30% of the pixels! Rather than create a smaller cropped image, the software then upscales the image digitally back to full size, so 1/3 of the pixels are just made up. This is like digital zoom on a smartphone and on point-and-shoot cameras.
  2. The uncorrected image is extremely warped, that it requires extreme correction, so in some places, pixels need to be squeezed closer together (are multiple pixels averages and then thrown out?), and in other places, like the corners, they need to be stretched further apart to create a rectilinear shape. Pixels don't stretch, so the spaces need to be filled in with extra pixels of fake computer-generated data, which the computer can't possibly predict, so it uses a formula to guess what they might be from surrounding pixels, so the detail in the corners will look soft and mushy.
  3. When the cropped image with fake pixels in the corners is upscaled, it is filled with computer generated pixels right through the image. The corners which are already degraded and soft from being filled with fake data pixels are then also digitally enlarged and filled a second time with more fake data, degrading them even further.

Simply put, the reason why photographers are advised not to go too heavy handed with post-processing is because image quality can start to fall apart, and detail can be lost. To produce usable images, two post-processing operations need to be performed at an extreme level with this lens just to get a usable image that we can start working on.

Earlier in this thread, I compared the Rf 16mm to the equivalent APSC lens, the EF-S 10-18mm, and this old crop sensor lens has very little distortion by comparison, as shown here:


115897_can10-18_dys10.jpg


Here's the lens distortion of the RF 16mm f/2.8 for the Christopher Frost video:

1635791418828.png


One important point that everyone has probably missed here is that test charts shot in RAW don't have any post processing applied to them other than lens corrections. Real images will have additional post-processing applied, sometimes a bit, sometimes quite a lot, and when they're applied over really extreme previous post-processing changes, the images simply don't hold up.

Comparing test charts and using image test tools like I've posted in this thread don't give an objective and accurate picture of how the software corrections affect real world images. It would be interesting to see a comparison of two identical landscape images, one taken on the RF 16mm f/2.8, and one on another 16mm (that doesn't use the radical software corrections), both taken through the complete workflow to produce the final photo, to see how the end result looks!
 
Last edited:
  • Like
Reactions: Frodo

RexxReviews

I'm New Here
Sep 3, 2021
23
14
Here are some samples I have been playing with. These have been shot on the R5

The first is from my 16-35 f4. This is with the LR profile.

JCM-9220.jpg


This is with the RF 16mm. This is a .jpg directly from camera using the built in camera corrections. The built in camera crop is Pretty much the same as the 16-35 @16mm

JCM-9221-2.jpg


This is the RAW of the same photo directly above with my manual corrections in LR. RF 24-240mm Profile . 130 Distortion. +100 Vignetting and 11 Midpoint in the manual tab.

JCM-9221.jpg
 
  • Like
Reactions: mpphoto

gruhl28

Canon 70D
Jul 26, 2013
175
63
Here are some samples I have been playing with. These have been shot on the R5

The first is from my 16-35 f4. This is with the LR profile.

View attachment 201070

This is with the RF 16mm. This is a .jpg directly from camera using the built in camera corrections. The built in camera crop is Pretty much the same as the 16-35 @16mm

View attachment 201071

This is the RAW of the same photo directly above with my manual corrections in LR. RF 24-240mm Profile . 130 Distortion. +100 Vignetting and 11 Midpoint in the manual tab.

View attachment 201072
Thanks for the samples! The 16-35 seems to show a slightly wider FOV than the 16 mm jpg (see the right side of the image; the left side lines up identically but there is more visible on the right), and the Raw image with the correction from the 24-240 shows a much wider FOV than either of the two previous images.
 

LogicExtremist

Lux pictor
Sep 26, 2021
173
106
One interesting thing about the 16mm, which I think someone earlier made reference to, is that stopping down seems to have virtually no effect on image quality other than vignetting. See https://www.the-digital-picture.com...eraComp=1508&SampleComp=0&FLIComp=0&APIComp=4
That was me! :) Yes, that's correct, corner sharpness doesn't improve when stopping down like regular lenses do because the corners are streched out and digitally filled with fake pixels to make up what the optics never captured in the first place! :(
 

RexxReviews

I'm New Here
Sep 3, 2021
23
14
Its really starting to feel like this is going to turn into more of a "YouTuber" "Vlogger" lens than anything else as the video seems decent on it for the price. I think those of us hoping to do more with it from a photog point of view are just going to be out of luck. We were REALLY hoping to be able to use this lens for real estate on our training camera due to the size and weight vs the 16-35's.
 
  • Like
Reactions: LogicExtremist

BBarn

EOS M6 Mark II
Nov 2, 2020
69
54
... corner sharpness doesn't improve when stopping down like regular lenses do because the corners are streched out and digitally filled with fake pixels to make up what the optics never captured in the first place!

The uncorrected geometric distortion of the RF 14-35 is similar to that of the RF 16. Yet, geometric distortion correction of the RF 14-35 yields very good corner sharpness. That would not be the case if geometric correction inherently introduced significant softness to an image. As I said before, if you really want to understand why the RF 16 is soft in the corners, look somewhere besides geometric distortion correction.

My own observations of pics I've taken with the two lenses show that geometric correction isn't highly destructive to image sharpness. And while the corrected corners of the RF 16 are softer than the RF 14-35, they are still quite acceptable for a majority of uses. It's not like corrected corners of the RF14-35 are merely adequate, and those of the RF 16 are hideous. Far from it.
 
Last edited:

koenkooi

EOS 5D Mark IV
CR Pro
Feb 25, 2015
1,919
1,859
The uncorrected geometric distortion of the RF 14-35 is similar to that of the RF 16. Yet, geometric distortion correction of the RF 14-35 yields very good corner sharpness. That would not be the case if geometric correction inherently introduced significant softness to an image. As I said before, if you really want to understand why the RF 16 is soft in the corners, look somewhere besides geometric distortion correction.

My own observations of pics I've taken with the two lenses show that geometric correction isn't highly destructive to image sharpness. And while the corrected corners of the RF 16 are a bit softer than the RF 14-35, they are still quite acceptable for a majority of uses. It's not like corrected corners of the RF14-35 are merely adequate, and those of the RF 16 are hideous. Far from it.
Agreed, the RF16 outperforms my EF17-40L in the corners, which is exactly what I bought it for.
 
  • Like
Reactions: gruhl28 and dcm

Frodo

EOS RP
Nov 3, 2012
428
109
I took the same raw files that I posted about just above:
- Applied the profile for the RF24-240: This is quite an improvement on the uncorrected image, but is not quite there. Interestingly, Lightroom shows the same file dimensions as the uncorrected file. Does the profile upscale the cropped image?
- Added +25 distortion correction in Lightroom. This now matches my earlier corrected image, but the image file is 6204x4136 pixels . This is 25.6 MP or 85% of original resolution. I could live with this.

I also applied the LR profile for my Samyang 14mm f/2.8. This also generates the full resolution, in spite of being cropped.
I didn't know that LR profiles upscaled cropped files.

eosr_rf16mm_f2.8-4.jpg
eosr_rf16mm_f2.8-5.jpg
 

gruhl28

Canon 70D
Jul 26, 2013
175
63
I took the same raw files that I posted about just above:
- Applied the profile for the RF24-240: This is quite an improvement on the uncorrected image, but is not quite there. Interestingly, Lightroom shows the same file dimensions as the uncorrected file. Does the profile upscale the cropped image?
- Added +25 distortion correction in Lightroom. This now matches my earlier corrected image, but the image file is 6204x4136 pixels . This is 25.6 MP or 85% of original resolution. I could live with this.

I also applied the LR profile for my Samyang 14mm f/2.8. This also generates the full resolution, in spite of being cropped.
I didn't know that LR profiles upscaled cropped files.

View attachment 201077 View attachment 201078
Thanks for the new examples. Since Lightroom is non-destructive, I wasn't surprised that the pixel count didn't change after applying a profile. I see the same behavior on other lenses. But then why did the count change after applying the +25? I tried this with a shot on another lens, and a manual distortion correction also didn't change the pixel count. Only after manually applying a crop did the Cropped dimensions change.
 

Frodo

EOS RP
Nov 3, 2012
428
109
Thanks for the new examples. Since Lightroom is non-destructive, I wasn't surprised that the pixel count didn't change after applying a profile. I see the same behavior on other lenses. But then why did the count change after applying the +25? I tried this with a shot on another lens, and a manual distortion correction also didn't change the pixel count. Only after manually applying a crop did the Cropped dimensions change.
Sorry, I should have clarified: I cropped the second image after applying the distortion correction.
 

LogicExtremist

Lux pictor
Sep 26, 2021
173
106
The uncorrected geometric distortion of the RF 14-35 is similar to that of the RF 16. Yet, geometric distortion correction of the RF 14-35 yields very good corner sharpness. That would not be the case if geometric correction inherently introduced significant softness to an image. As I said before, if you really want to understand why the RF 16 is soft in the corners, look somewhere besides geometric distortion correction.

My own observations of pics I've taken with the two lenses show that geometric correction isn't highly destructive to image sharpness. And while the corrected corners of the RF 16 are softer than the RF 14-35, they are still quite acceptable for a majority of uses. It's not like corrected corners of the RF14-35 are merely adequate, and those of the RF 16 are hideous. Far from it.
The explanation is quite simple, the better optics in the RF 14-35mm resolve much more detail, so you start off with a more detailed warped image than the RF 16mm. After crunching the pixels equally to correct them, the L-series zoom will naturally look better. It can't be anything other than software correction because the corners and periphery don't sharpen when the lens is stopped down, and that's because they're fake pixels, which are just best guess filler-pixels that don't respond to aperture. Real pixels do on the RF 50mm f/1.8, which becomes very sharp across the frame when stopped down to f/5.6 and higher.

You mention that from your own observations using the two lenses, the geometric correction isn't highly destructive to image sharpness and the corners aren't too bad. Is that just from a test image? In a real world application of a real estate photo or a landscape shot, additional post-processing is required to turn the raw image with lens corrections into a proper photo. It's not accurate to say the image corners look okay with either lens if the images aren't subjected to the regular post-processing workflow.

The test is easy, for a quick landscape shot just crank up the lens aperture to say f/8 to f/11 for a good FOV and to eliminate wide aperture blur in the corners, put the camera on a tripod and take an outdoor shot, then post-process to get as much details out of the shadows and highlights, make any other adjustments so it looks good to your eye, then analyze the final image. If real-estate style photography is preferred, do what needs to be done there to get the angles and perspective right in a room, and avoid blown out highlights from light through windows using HDR or accurate exposure comp and detail recovery, and see how it looks. We're just emulating real world use.

When it comes to raw images with only lens corrections applied, anyone can get decent images using an entry level camera with a kit lens. To produce anything beyond SOOC JPEG quality, the images need to be pushed a bit further in post processing, and that's when the shortcomings of the lens and body begin to show.

Simply put, the only way to properly evaluate these lenses is to put the raw images through a complete post-processing workflow, like what would happen in real-world use, and then check to see how the corner details hold up. Then we can say wheter thre final images look okay or not, and then we can make an informed decision as too whether the quality is good enough for purpose or not for each individual. If shooting JPEGs then it won't make a difference as there's not post processing carried out on what comes out of the camera. If I had either lens I'd happily test it out for everyone, but I don't, I can't... :(
 
Last edited:

gruhl28

Canon 70D
Jul 26, 2013
175
63
The explanation is quite simple, the better optics in the RF 14-35mm resolve much more detail, so you start off with a more detailed warped image than the RF 16mm. After crunching the pixels equally to correct them, the L-series zoom will naturally look better. It can't be anything other than software correction because the corners and periphery don't sharpen when the lens is stopped down, and that's because they're fake pixels, which are just best guess filler-pixels that don't respond to aperture. Real pixels do on the RF 50mm f/1.8, which becomes very sharp across the frame when stopped down to f/5.6 and higher.

You mention that from your own observations using the two lenses, the geometric correction isn't highly destructive to image sharpness and the corners aren't too bad. Is that just from a test image? In a real world application of a real estate photo or a landscape shot, additional post-processing is required to turn the raw image with lens corrections into a proper photo. It's not accurate to say the image corners look okay with either lens if the images aren't subjected to the regular post-processing workflow.

The test is easy, for a quick landscape shot just crank up the lens aperture to say f/8 to f/11 for a good FOV and to eliminate wide aperture blur in the corners, put the camera on a tripod and take an outdoor shot, then post-process to get as much details out of the shadows and highlights, make any other adjustments so it looks good to your eye, then analyze the final image. If real-estate style photography is preferred, do what needs to be done there to get the angles and perspective right in a room, and avoid blown out highlights from light through windows using HDR or accurate exposure comp and detail recovery, and see how it looks. We're just emulating real world use.

When it comes to raw images with only lens corrections applied, anyone can get decent images using an entry level camera with a kit lens. To produce anything beyond SOOC JPEG quality, the images need to be pushed a bit further in post processing, and that's when the shortcomings of the lens and body begin to show.

Simply put, the only way to properly evaluate these lenses is to put the raw images through a complete post-processing workflow, like what would happen in real-world use, and then check to see how the corner details hold up. Then we can say wheter thre final images look okay or not, and then we can make an informed decision as too whether the quality is good enough for purpose or not for each individual. If shooting JPEGs then it won't make a difference as there's not post processing carried out on what comes out of the camera. If I had either lens I'd happily test it out for everyone, but I don't, I can't... :(
I agree that the way to evaluate whether the lens meets someone's needs is to take an image similar to those usually taken and then post-process in the way that one usually does. However, I disagree that post-processing is necessarily going to degrade the image further and make it look worse. Things like noise often get worse, but I don't see why a lack of sharpness would necessarily get worse. Of course, the only way to know is to actually try it.

I'm also not sure about your explanation of why the image doesn't improve with stopping down. These aren't completely "fake" pixels constructed out of nothing. Even if new pixels are being created, they are being created based on the pixels that were there before corrections. So if the actual image gets sharper when stopping down, then the corrected image based on the stopped down results should be sharper also. You rightly pointed out that the 14-35 resolves much more detail to begin with, and so the corrected image also shows more detail. That same logic would apply to stopping down; if the actual image coming from the lens improves, then the corrected image should also be improved. I'm not saying that the large distortion corrections might not be contributing to the apparent lack of improvement upon stopping down, but I don't think this can be the entire reason. Not all lens aberrations improve with stopping down. Out of curiosity I looked at the 14-35 results when stopping down, and I admit that these also don't seem to really improve (apart from vignetting of course), although there isn't a lot of room for improvement in that lens to begin with - the results are pretty good even wide open.

It would be interesting to hear from someone from Canon (which probably isn't going to happen) or someone else with real expertise in lens design and processing, who could explain fully what is happening here. We are all taking educated guesses, but I'd love to hear a complete explanation from someone who really knows what is happening and why we are seeing the results we're seeing.
 
  • Like
Reactions: LogicExtremist

LogicExtremist

Lux pictor
Sep 26, 2021
173
106
I agree that the way to evaluate whether the lens meets someone's needs is to take an image similar to those usually taken and then post-process in the way that one usually does. However, I disagree that post-processing is necessarily going to degrade the image further and make it look worse. Things like noise often get worse, but I don't see why a lack of sharpness would necessarily get worse. Of course, the only way to know is to actually try it.

I'm also not sure about your explanation of why the image doesn't improve with stopping down. These aren't completely "fake" pixels constructed out of nothing. Even if new pixels are being created, they are being created based on the pixels that were there before corrections. So if the actual image gets sharper when stopping down, then the corrected image based on the stopped down results should be sharper also. You rightly pointed out that the 14-35 resolves much more detail to begin with, and so the corrected image also shows more detail. That same logic would apply to stopping down; if the actual image coming from the lens improves, then the corrected image should also be improved. I'm not saying that the large distortion corrections might not be contributing to the apparent lack of improvement upon stopping down, but I don't think this can be the entire reason. Not all lens aberrations improve with stopping down. Out of curiosity I looked at the 14-35 results when stopping down, and I admit that these also don't seem to really improve (apart from vignetting of course), although there isn't a lot of room for improvement in that lens to begin with - the results are pretty good even wide open.

It would be interesting to hear from someone from Canon (which probably isn't going to happen) or someone else with real expertise in lens design and processing, who could explain fully what is happening here. We are all taking educated guesses, but I'd love to hear a complete explanation from someone who really knows what is happening and why we are seeing the results we're seeing.
The way software lens correction and post-processing works is by performing mathematical operations on the pixel data. It's not possible to create accurate pixel data that was never captured optically, as the software can't tell what's being photographed, it uses and algorithm to try guess the pixels from those around it. If one software operation calculates pixels, then another operation applied over that also uses those calculated pixels to generate new calculated pixels, then there is the potential to generate garbage data, which appears as aberrations, which are commonly seen in over-processed images, where things appear in the image that never existed in real life. Depending on the image post-processing workflow operation, this might or might not be a problem, will vary on a case-by-case basis and is easy to test.

To quote the following review of the lens "Distortion correction is destructive at the pixel level as some portion of the image must be stretched or the overall dimensions reduced. "


"This is an upper-left quadrant of an f/2.8 image, the misshapen circles seem to correlate with the distortion corrected area of the frame. "

Cats-Eye-Bokeh.jpg


You can see the distortion correction is turning the sharp circles in the image into blurred ovals in the left hand top corner. Performing further post-processing on these might create something even more random.

My understanding of why the corners don't sharpen at higher apertures is as follows:
  • The RF 16mm is optically sharp in the centre, and that part isn't being software corrected as much as the optically soft corners, which are being blurred more by fake data pixel filling.
  • Stopping down the aperture to sharpen the image is optical in nature.
  • Software corrected pixels in the corner are computer generated and can't be improved by changing aperture.

If anyone has a better theory, I'm curious to hear it! :)
 

LogicExtremist

Lux pictor
Sep 26, 2021
173
106
Here are some samples I have been playing with. These have been shot on the R5

The first is from my 16-35 f4. This is with the LR profile.

View attachment 201070

This is with the RF 16mm. This is a .jpg directly from camera using the built in camera corrections. The built in camera crop is Pretty much the same as the 16-35 @16mm

View attachment 201071

This is the RAW of the same photo directly above with my manual corrections in LR. RF 24-240mm Profile . 130 Distortion. +100 Vignetting and 11 Midpoint in the manual tab.

View attachment 201072
The image isn't that great unfortunately considering we want to compare corner sharpness! If you could stop down both lenses to f/8 and place something with fine detail in each lower corner, such as a box or book with fine lettering, that would tell us heaps. Interesting that the EF 16-35mm f/4 L is wider than the RF 16mm f/2.8. Thanks :)
 

stevelee

FT-QL
CR Pro
Jul 6, 2017
2,176
882
Davidson, NC
I am pretty sure that if Canon labels a lens 16mm or 14mm, the FoV delivered in the final images will be that. Keep in mind that focal length is specified at infinity focus. Since people testing for lens distortion aren’t using charts or walls large enough to fill the frame with the lenses focused at infinity, focus breathing must be considered. For example, the EF-S 18-200mm at the long end frames at ~150mm equivalent with a close subject, the EF 100/2.8L Macro frames like ~67mm at 1:1. The 16/2.8 probably has a fair bit of breathing, the 14-35 less so.
It would be helpful if someone would send a giant test chart into orbit.
 

Frodo

EOS RP
Nov 3, 2012
428
109
Interesting video here where the guy looks at the RF 16mm for astrophotography. He concludes that it is usable, although not as sharp as the EF 16-35mm f/4. Coma and vignetting quite severe in the corners, but he cropped to the same FOV as the 16-35 and much of this disappeared.
I wonder what a raw processed in DPP exported as tif to Photoshop or Lightroom would look like.
 
  • Like
Reactions: LogicExtremist

Czardoom

EOS RP
Jan 27, 2020
346
733
As far as I know all lenses use distortion correction, that's not the issue. What is catching peoples attention here is the crazy amount of distortion of this lens that is only possible because it's being used on a mirrorless system, which can hide it because it does on-the-fly image correction when looking through the EVF and makes the same corrections in camera when shooting JPEG images. This level of distortion would never be seen on a DSLR with an OVF, as it would produce a severely warped image viewing through the lens.

The are three main problems:
  1. The lens has heavy vignetting, the corners of the uncorrected image are pure black, so no amount of shadow recovery will bring those details back. As such, they need to be cropped away, which by most estimates is thought to remove around 30% of the pixels! Rather than create a smaller cropped image, the software then upscales the image digitally back to full size, so 1/3 of the pixels are just made up. This is like digital zoom on a smartphone and on point-and-shoot cameras.
  2. The uncorrected image is extremely warped, that it requires extreme correction, so in some places, pixels need to be squeezed closer together (are multiple pixels averages and then thrown out?), and in other places, like the corners, they need to be stretched further apart to create a rectilinear shape. Pixels don't stretch, so the spaces need to be filled in with extra pixels of fake computer-generated data, which the computer can't possibly predict, so it uses a formula to guess what they might be from surrounding pixels, so the detail in the corners will look soft and mushy.
  3. When the cropped image with fake pixels in the corners is upscaled, it is filled with computer generated pixels right through the image. The corners which are already degraded and soft from being filled with fake data pixels are then also digitally enlarged and filled a second time with more fake data, degrading them even further.

Simply put, the reason why photographers are advised not to go too heavy handed with post-processing is because image quality can start to fall apart, and detail can be lost. To produce usable images, two post-processing operations need to be performed at an extreme level with this lens just to get a usable image that we can start working on.

Earlier in this thread, I compared the Rf 16mm to the equivalent APSC lens, the EF-S 10-18mm, and this old crop sensor lens has very little distortion by comparison, as shown here:


115897_can10-18_dys10.jpg


Here's the lens distortion of the RF 16mm f/2.8 for the Christopher Frost video:

View attachment 201061

One important point that everyone has probably missed here is that test charts shot in RAW don't have any post processing applied to them other than lens corrections. Real images will have additional post-processing applied, sometimes a bit, sometimes quite a lot, and when they're applied over really extreme previous post-processing changes, the images simply don't hold up.

Comparing test charts and using image test tools like I've posted in this thread don't give an objective and accurate picture of how the software corrections affect real world images. It would be interesting to see a comparison of two identical landscape images, one taken on the RF 16mm f/2.8, and one on another 16mm (that doesn't use the radical software corrections), both taken through the complete workflow to produce the final photo, to see how the end result looks!
All I know is that Olympus,. Sony and others have been doing this for a few years now. No one who uses those lenses is creating the kind of hysteria that is being created on this thread. I own the Olympus 12-100mm f/4 - a lens with very heavy distortion correction automatically applied. Feel free to read all the reviews of that lens you want - or I'll save you the trouble - it is very highly regarded as a Pro quality lens.

You talk of the "black pixels" in the corners that can't be recovered. Not supposed to be, after correction they are beyond the frame. So, yes, pixels are being stretched. Interpolations are being made. Exactly what happens when you or I or most photographers use distortion correction when post processing. Or use Ketstone or perspective correction.

You write, "Pixels don't stretch, so the spaces need to be filled in with extra pixels of fake computer-generated data, which the computer can't possibly predict..."

Guess what, yes the computer can. It does it all the time. How about Gigapixel or Photoshop's new enlargent application. Good heavens I just used Gigapixel that doubled my pixel count. All those made up pixels - it must be a disaster. But, no actually. Looks pretty good.

Yes, some interesting facts have been presented regarding how much distortion, and and an estimate of many MP's are lost in the process. But this thread has gone way beyond that into "fake pixels" and other hysteria that is actually somewhat amusing to those of use (or at least to me) who have used the Olympus lens and others like it.

Canon's mistake is not that use auto-correction, their mistake is that they didn't make sure that every photo processing software had the profiles in advance. Then we would be spared all the analysis of un-processed photos that we will never see once the profiles are in place.

It's a $300 lens and by most or all accounts I have read, it done a pretty good job - out performinig the 17-40 L lens, for example. That, it seems to me, is the most important "fact".
 
  • Like
Reactions: mpphoto and gruhl28

LogicExtremist

Lux pictor
Sep 26, 2021
173
106
All I know is that Olympus,. Sony and others have been doing this for a few years now. No one who uses those lenses is creating the kind of hysteria that is being created on this thread. I own the Olympus 12-100mm f/4 - a lens with very heavy distortion correction automatically applied. Feel free to read all the reviews of that lens you want - or I'll save you the trouble - it is very highly regarded as a Pro quality lens.

You talk of the "black pixels" in the corners that can't be recovered. Not supposed to be, after correction they are beyond the frame. So, yes, pixels are being stretched. Interpolations are being made. Exactly what happens when you or I or most photographers use distortion correction when post processing. Or use Ketstone or perspective correction.

You write, "Pixels don't stretch, so the spaces need to be filled in with extra pixels of fake computer-generated data, which the computer can't possibly predict..."

Guess what, yes the computer can. It does it all the time. How about Gigapixel or Photoshop's new enlargent application. Good heavens I just used Gigapixel that doubled my pixel count. All those made up pixels - it must be a disaster. But, no actually. Looks pretty good.

Yes, some interesting facts have been presented regarding how much distortion, and and an estimate of many MP's are lost in the process. But this thread has gone way beyond that into "fake pixels" and other hysteria that is actually somewhat amusing to those of use (or at least to me) who have used the Olympus lens and others like it.

Canon's mistake is not that use auto-correction, their mistake is that they didn't make sure that every photo processing software had the profiles in advance. Then we would be spared all the analysis of un-processed photos that we will never see once the profiles are in place.

It's a $300 lens and by most or all accounts I have read, it done a pretty good job - out performinig the 17-40 L lens, for example. That, it seems to me, is the most important "fact".

You do realise that "they're doing it, so that makes it okay to do it too" is not a sound argument, that's the bandwagon fallacy, it's a logical fallacy which just appeals to popularity. My response would simply be that if Canon, who are known for the quality of their lenses, take the same shortcuts that Sony does (I'm trusting your claim on this, not sure how true it is), then that's one less reason to choose Canon over Sony.

Software correction is not all the same as you probably already can figure. To use an extreme example just to illustrate a point, stretching a triangle of pixels into a square is much harder, will generate more fake data, and degrade an image more than than simply making a square of pixels twice as large. Similarly, stretching out corners requires more pixels generated, and is less precise than an interpolation of a uniformly sharp image to make it twice as large.

You've proven my point with your example, images enlarged with Topaz Gigapixel look amazing, while the software generated corners in the RF 16mm f/2.8 look soft, blurry and smeared. The proof is in the results, the final images, and that's what matters when producing photos.

Incidentally, there's no hysteria here, just an objective evaluation of what the lens can and can't do, without any attempted rationalisation to make excuses for its shortcomings. What we're discussing is why the corners look like crap, and we more or less know why from this discussion and others.

If we need a lens or a specific purpose, then it will either work or it wont, regardless of whether the brand is a Canon, Sony or Olympus. Knowing the shortcomings of a lens helps buyers make informed decisions in respect to their specific needs.

When we consider what 16mm primes are generally used for, we can draw reasonable conclusions from the reviews and discussions about how this lens performs in each of these areas:

Architecture/interiors - no
Landscape - no
Astro - no/maybe
General video - yes
Vlogging - yes
Webcam - yes
Travel/hiking - yes

Does this not sound reasonable? :)
 

LogicExtremist

Lux pictor
Sep 26, 2021
173
106
It's strange seeing people wishing that a lens is something more than it is!

What does Canon say the Rf 16mm f/2.8 is for?

Let's look at the advertising from Canon's websites in The US, UK and Australia, noting the difference in features emphasized.

-----------------------------

The Canon US website oversells this lens in terms of vague, non-specific sales hype as expected:

"Compact, versatile, speedy and affordable, the new RF16mm F2.8 STM lens is a terrific complement to any EOS R series digital camera. Offering an ultra-wide angle of view and a bright f/2.8 aperture, it's an excellent choice whether taking interior photos in tight spaces, seeking out the perfect landscape, or as the perfect webcam lens when used in combination with an EOS camera and EOS Webcam Utility on a video call. With a 16mm F2.8 lens, you can experiment with enhancing perspectives, capturing starscapes, or with the close focusing distance of 5.11 inches, get up close to your subject while still keeping the background more visible. Video users will find this ultra-wide lens a natural for vlogging, especially with its supremely light weight. With its remarkable combination of optical excellence and refined performance, all in a small package that's easy to bring most anywhere, the RF16mm F2.8 STM is a stellar companion to any EOS R series digital camera."

-----------------------------

The Canon UK website states:

"Put yourself in the picture with this fast, affordable, ultra-wide EOS R-series lens. At 16mm this full frame prime is great for vlogging, landscapes, architecture, astrophotography and more."

The very first feature headlined is this:

Go wide, go bright and put yourself in the picture

Meet the RF 16mm F2.8 STM. It’s the ultimate affordable EOS R-series lens for vlogging and more.

Designed for vloggers and creative content makers

For crisp, clear ultra-wide views this lightweight and super compact full frame 16mm prime will transform your images and videos, inspiring creativity at every opportunity – from landscapes to architecture and even astrophotography. Can you afford not to have one?

-----------------------------

The Canon Australia site has a more honest description of its product:

The perfect wide-angle addition to your kit

The RF 16mm f2.8 STM is the first ultra-wide fixed focal length lens in the RF line up. This compact and lightweight lens is perfect for getting bright, panoramic landscapes to fun portraits and videos.

Versatile wide angle lens, the perfect travel companion
Image of the beach with trees taken with the RF 16mm F2.8 STM wide angle lens

The ultra-wide 16mm lens is a perfect addition to your kit while travelling, allowing you to capture scenic landscapes to large group shots and selfies.

-----------------------------

As we can see, Canon UK puts the emphasis on its use as a vlogging lens, while Canon Australia highlights its use as a travel lens. In advertising, just like in ingredients list, the most important things are listed first. Quite telling really... ;)