RF 14-35 and cropping

Stein Rune Risa wrote:

...

So to my question:

I understand that the correction of mechanical vignetting involves some cropping to remove the dark corners. But does this mean that it actually is not 14 mm? Or is it really "wider than 14 mm" and the cropping takes it down to 14 mm? Will I get the picture that I frame in my viewfinder - or will it be somewhat cropped?

Not the lens you're talking about, but with the RF 16mm f/2.8 you can't switch off the distortion correction in the camera and what you get is the field of view a 16mm rectangular projection lens, whether you use in-camera jpeg or export from DPP4. It's the same, almost to the pixel, as the field of view given by the EF 16-35mm f/4 set to 16mm.

(You can see better how photo editors correct barrel distortion if you import a photo into the GIMP and expand the canvas before adjusting the distortion. The original rectangular image ends up with parabolic (or 3rd order or higher) concave curved sides, and the largest rectangle you can crop out of the shape gets longer than the original.)

With DxO PhotoLab 5, the angle of view the program gives with its standard distortion correction is a little bit wider than that given by DPP4. (Same number of pixels, so DPP4 must magnify and crop more.)  It's closer to a 15mm than a 14mm field of view by my calculation, assuming DPP4's 16mm field of view is correct. I can even get a bit wider, 13½mm equivalent field of view, by cropping to 16:9, which is pretty amazing for such an inexpensive lens.

As a practical example, I put the camera on a tripod and photographed a mill with a 12mm lens, the 16-35mm EF lens, the 16mm RF lens and (without moving the tripod, an M100 + 11-22mm), then exported jpegs from both DPP4 and PhotoLab, superimposed them as layers in the GIMP, sizing and moving them to match, then drawing round each image to show the coverage. Here's the result.

@Sittatunga - Thanks for your comprehensive answer!

From your post, I understand that:

Yes - I can expect the images from the 14-35 to be 14 mm after correction, and that any dark corners etc are actually in "a wider part" than the 14 mm view. And that other software than DPP might actually be able to provide even wider - but maybe not with this lens.

Correctly understood?

I see in Dustin Abbott's review of the lens here:

https://dustinabbott.net/2021/10/canon-rf-14-35mm-f4l-is-usm-review/

that he shows some of this vignetting and distortion. I can see that his manual correction effort actually is wider than the in camera ones. But he also writes "I would say the corrected image from the Canon probably behaves more like a 15mm lens."...

cocoanud wrote:

That is the RF 16mm.

As with the excellent test results produced by @Sittatunga above, DxO PL4 is able to extract and correct the RF 16 lens to near 14mm !

If you don't need the (practically useless focal lengths of 24mm and above as they are already covered elsewhere) then you have a 14-16mm lens in the RF 16mm. If you have the R5 then even more flexible.

For me, 14 mm or 16 mm is too wide in many situations. Sure, one can crop, especially with a high-resolution camera such as the R5. But for people who often shot around 20mm, a 14/15-35 zoom lens makes sense. Especially as there are no 20mm primes available (other than the rather old EF 20mm which has strong field curvature and therefore can be challenging to use).

Stein Rune Risa wrote:

@Sittatunga - Thanks for your comprehensive answer!

From your post, I understand that:

Yes - I can expect the images from the 14-35 to be 14 mm after correction, and that any dark corners etc are actually in "a wider part" than the 14 mm view. And that other software than DPP might actually be able to provide even wider - but maybe not with this lens.

Correctly understood?

Yes, though the corners are also brightened by the vignetting correction. And I have noticed with a variety of lenses that correcting barrel distortion with PhotoLab gives me the option of getting a bit more of a wide angle by cropping to a longer shape, because of what that correction does to the shape of the included rectangle. My 16-35mm is newer than my EOS R, so it's going to be a long time before I buy a wide-angle zoom in that class, but I would fully expect PhotoLab to extract a wider angle from the RF 14-35mm than DPP4 when the uncorrected image shows barrel distortion. The catch, of course, is the quality at the extreme corners.

It's early days but my zoom is cleaner in the corners than my 16mm, yet field curvature can sometimes make the bottom corners of the prime look sharper than those of the zoom. I need to investigate this more thoroughly.

I see in Dustin Abbott's review of the lens here:

https://dustinabbott.net/2021/10/canon-rf-14-35mm-f4l-is-usm-review/

that he shows some of this vignetting and distortion. I can see that his manual correction effort actually is wider than the in camera ones. But he also writes "I would say the corrected image from the Canon probably behaves more like a 15mm lens."...

Franz Kerschbaum wrote:

At comparable apertures the RF 14-35 is not sharper than the RF 15-35! Both vignetting and corner sharpness at f/4 are better with the RF 15-35. The RF14-35 gives you that extra mm and is lighter and cheaper. Thats it. The RF 15-35 gives you that extra stop of light that is good to have at times. Depends on your usage. Both have great IS and allow for incredible long handhold times. Would stray away from the sigma if you can affort the canons for compactness and full compatibility and make up your mind on the criterias I stated above. (I have the RF15-35)

The Dustin Abbott review seems to disagree?

"In fact, when I compared my results from the 15-35mm F2.8L IS that I reviewed last year, I found that the 14-35mm results were clearly better at equivalent apertures."

https://dustinabbott.net/2021/10/canon-rf-14-35mm-f4l-is-usm-review/

As someone who has the RF 15-35mm F2.8 and previously had the RF 14-35mm F4, I don't agree with Dustin Abbott's findings based upon the copies I was comparing. The RF 15-35mm f f2.8 seemed just a bit better at the edges at least.

The RF14-35mm f4 is a decent lens all the same, providing your software has lens correction. I swapped the lenses because I realised I need f2.8 (for astro) and was inconvenienced by the fact that my RAW convertor of choice (C1) does not have a lens profile for the f4.

I noticed something interesting regarding the field of view between the RF 16mm and RF 14-35mm on an R5 at a distance of only a few feet. As one would expect when using Canon DPP to process images, the FOV of the 14-35 (set at 14) is noticeably wider than the RF16. However, if Lightroom or DXO is used to process the images (with their respective lens corrections), there is very little difference in FOV between the two lenses at a distance of only a few feet.

Not sure how much those results might be impacted at longer focusing distances. Nor do I have a 15-35mm to test for comparison.

BBR5 wrote:

I noticed something interesting regarding the field of view between the RF 16mm and RF 14-35mm on an R5 at a distance of only a few feet. As one would expect when using Canon DPP to process images, the FOV of the 14-35 (set at 14) is noticeably wider than the RF16. However, if Lightroom or DXO is used to process the images (with their respective lens corrections), there is very little difference in FOV between the two lenses at a distance of only a few feet.

Not sure how much those results might be impacted at longer focusing distances. Nor do I have a 15-35mm to test for comparison.

I actually did get hold of some raw files from an R5 with 14-35 and used converted them using both DxO PureRAW and Lightroom (using the provided profiles).

My observations were:

* The original image from the camera is wider than 14 mm
* DxO provides a wider image than Lightroom - probably wider than DPP as well and wider than 14 mm?
* DxO does a much better job at processing the image - the sharpness (especially in the corners) are MUCH better

The images can be seen here:

https://www.fredmiranda.com/forum/topic/1716216/11#15832268

I have not actually tried to convert the raw file using DPP - but maybe I will test it.

Edit: I now converted the image in DPP and see that it is the most narrow of DxO/LR/DPP. Lightroom is wider, but DxO is even wider again. See link to fredmiranda for updated post there.

Franz Kerschbaum wrote:

Check here at different focal lengths and same f/4 setting. In most of the cases the RF 15-35 seems to have the egde or is similar and the residual CA is very well correted in post. If one needs f/2.8 take the RF15-35, if f/4 is enough save the money and take the RF14-35. Both are great choices. But nobody will buy the RF14-35 for only better image quality.

I consider RF 14-35 because of its wider field, but also based on the reviews that it also is sharper at the wider end (with profile corrections applied). I am "only" interested in the wide end (14-24). I have a 24-70 that handles the rest.

But there seems to be some disagreement about the sharpness at the wide end of the lens. I think everyone agrees that it is worse at the close (e.g. 35 mm) end.

Both DxO and Lightroom now has profiles in place for it, and seems to do a very good job. Also remember that the unprofiled image is wider than 14 mm, so even if there are heavy vignetting etc at that - the profiled image has cropped away a lot of the bad areas.

Here's a shot with the RF16, processed in Canon DPP.

Here's the same RF16 shot processed in Lightroom.

Here's the same basic shot with the RF14-35, processed in Canon DPP.

Lastly, here's the above RF14-35 shot processed in Lightroom.

With regard to FOV, there is little difference between the RF16 and RF14-35 (at 14mm) when processed in Lightroom (or DXO). Canon DPP crops the RF16 considerably more, and the RF14-35 slightly more, versus Lightroom.

These are low res shots through double pane glass, so they're inadequate for evaluating much besides FOV.