Canon RF 10-20mm f/4L IS STM Review

[Richard CR]

OpticalLimits has recently reviewed the Canon RF 10-20mm f/4L IS STM. This was an impressive optic when Canon released it in October 2023, and we had predicted it would be a strong performer based on its MTF. You can read all about that right here. The key features of the Canon RF 10-20mm f/4L IS

See full article...

 

[neuroanatomist]

It really is an excellent lens, and while I initially held off on buying it, it's as good as my EF 11-24/4L in a far more portable package.

 

[hunck]

tbh, I was shocked by the quality of the images. Coming from a Sigma EF12-24 this lens is better in every metric. I made thousands of pictures with it in a short few months and every picture is outstanding. The AF and IS are as great as the optics. And the weight, wow, I carry this everywhere! So happy they made it f/4. This saves me 1200 euros and half a Noct. Now it\'s my fav travel lens.

 

[roby17269]

No complaints here, I like mine for its IQ and portability.
Do not love the hood design...
My biggest gripe with it it\'s not a fault of the lens per se: as a photographer, I find that it\'s difficult for me to use it successfully at 10mm. Not a matter of quality, just a matter of finding a composition that works at 10mm. I find myself using it much more at 20mm.

 

[nhandumuc]

Is the front element good enough to resist dirt and salt spray, I live by the coast so these are constant annoyance, I can use UV filters with normal lenses, easily wipe these elements out, but not sure about the 10-20F4.

 

[josephandrews222]

The OpticalLimits review hits all the high points and is quite readable. I'm going to check out other reviews there. Thanks for the heads-up, RIchard.

 

[Gogo7]

Would love to see a gallery at 10 to 12mm only ! Can’t see myself using such a wide angle lens ! But that’s just me of course

 

[Canon Rumors Guy]

It'd be cool if they could make more of them.

It's finally available to rent here, and I'll have one next week. I'm terrible with ultra wide lenses, but it's an intriguing package.

 

[exige24]

It really is an excellent lens, and while I initially held off on buying it, it's as good as my EF 11-24/4L in a far more portable package.

Not only that, it trips me out that these reviewers take issue with the digital correction at 10 mm. It is not physically possible to make a lens that has a square image at 10 mm on a full frame sensor. 11 mm is the last digit where it is possible.

 

[Jack Douglas]

Would love to see a gallery at 10 to 12mm only ! Can’t see myself using such a wide angle lens ! But that’s just me of course

I just tried to upload an 11mm shot and no matter how small I make it, it says too big. Anyone??

Jack

 

[cycomachead]

I own an 11-24 and love the lens, but damn if it's bulky and the EF adapter makes it just enough more uncomfortable that I haven't brought it with me much...

Currently own the RF 14-35mm (and selling my RF 15-35mm) and strongly thinking about replacing the 14-35 with the 10-20. For walk-around, the 20-35mm range is useful to have, and worry I'd want to switch lenses more. Anyone else own both or switched from a more classic wide zoom to 'just' the 10-20?

OTOH, a RF 10-20, 24-105, and 70-200 f4L kit is a seriously impressive range for a fairly light amount of weight...

 

[ashmadux]

Not only that, it trips me out that these reviewers take issue with the digital correction at 10 mm. It is not physically possible to make a lens that has a square image at 10 mm on a full frame sensor. 11 mm is the last digit where it is possible.

Well this begs the question, how much does the image suffer in order to stretch the view...and how much is Canon using this tech with other lenses?

right now as I write this, I don't really believe in buying lenses that rely on fake image stitching to show a full view. my view might change later, but seeing that we use third party tools, lack of compatibility with this type of image rebuilding seems like a detriment. I was disappointed to hear that the 24 to 105 2.8 uses this feature if I'm not mistaken.

Canon has invested a lot in this kind of image manipulation at the endpoint, and I don't agree with that direction.

But who knows let's see where this road goes.

 

[EricN]

Not only that, it trips me out that these reviewers take issue with the digital correction at 10 mm. It is not physically possible to make a lens that has a square image at 10 mm on a full frame sensor. 11 mm is the last digit where it is possible.

I wasn't aware of this.

 

[Richard CR]

No complaints here, I like mine for its IQ and portability.
Do not love the hood design...
My biggest gripe with it it\'s not a fault of the lens per se: as a photographer, I find that it\'s difficult for me to use it successfully at 10mm. Not a matter of quality, just a matter of finding a composition that works at 10mm. I find myself using it much more at 20mm.

yes, composition is the hardest for the SUWA lenses (super ultra wide angle)

Not only that, it trips me out that these reviewers take issue with the digital correction at 10 mm. It is not physically possible to make a lens that has a square image at 10 mm on a full frame sensor. 11 mm is the last digit where it is possible.

why is this? Not that I'm doubting you, but I'm curious on the theory surrounding this.

 

[koenkooi]

Well this begs the question, how much does the image suffer in order to stretch the view...and how much is Canon using this tech with other lenses?[…]

You can easily check that by opening a sample CR3 files from one of the review sites in a non-Canon RAW converter. That will show you the uncorrected images.

Keep in mind that some lenses are wider than advertised, so the RAW from the 14-35 at 14mm is more like 13mm, after correction it is the advertised 14mm, but it cut off the sides during processing.

 

[neuroanatomist]

Well this begs the question, how much does the image suffer in order to stretch the view...and how much is Canon using this tech with other lenses?

The thing is, if you want a rectilinear view (vs fisheye) with an ultra wide angle, the image always has to be stretched. Always. In film and DSLR days, that stretching had to be accomplished with glass elements in the lens, and even with that ultrawide lenses were left with significant distortion. For example, the Canon EF 16-35/2.8 III has 3.5% barrel distortion at the wide end, the Nikon 14-24/2.8G has 4%, and the EF 11-24/4L has 4.5%. The 'optical' (in-lens) correction of the distortion to make those lenses mostly rectilinear is part of the reason why those lenses resolve far less (50-90% lower) in the corners compared to the center, even when stopped down (which fixes many of the other reasons for loss of corner sharpness).

The bottom line is that there's no free lunch. Distortion can be corrected in the lens or in software, neither one is inherently better but the latter allows lenses to be smaller and lighter.

The RF 14-35/4 at 14mm is just as sharp in the corners as the EF 11-24/4 at the same focal length, and at that focal length the 11-24 has essentially zero barrel distortion. That shows the image is not suffering, at all.

But who knows let's see where this road goes.

Good that you're keeping an open mind, at least.

 

[ashmadux]

The thing is, if you want a rectilinear view (vs fisheye) with an ultra wide angle, the image always has to be stretched. Always. In film and DSLR days, that stretching had to be accomplished with glass elements in the lens, and even with that ultrawide lenses were left with significant distortion. For example, the Canon EF 16-35/2.8 III has 3.5% barrel distortion at the wide end, the Nikon 14-24/2.8G has 4%, and the EF 11-24/4L has 4.5%. The 'optical' (in-lens) correction of the distortion to make those lenses mostly rectilinear is part of the reason why those lenses resolve far less (50-90% lower) in the corners compared to the center, even when stopped down (which fixes many of the other reasons for loss of corner sharpness).

The bottom line is that there's no free lunch. Distortion can be corrected in the lens or in software, neither one is inherently better but the latter allows lenses to be smaller and lighter.

The RF 14-35/4 at 14mm is just as sharp in the corners as the EF 11-24/4 at the same focal length, and at that focal length the 11-24 has essentially zero barrel distortion. That shows the image is not suffering, at all.


Good that you're keeping an open mind, at least.

Logic dictates that canon will be doing a lot of this moving forward (since they have very actively been using this feature), so it will likely be harder to avoid anyways. My first concern was that is this a feature that would work better in the amazingly awful DPP as opposed to LR. Hopefully it is/will be as seamless to 3rd party apps.

When the first info was being shared here about canon pursuing this tech, it was also around the time of their plastic lenses being introduced (generally speaking, maybe, dont quote me)...so a a whole it looked like ways to simply save some cash (consumers perspective).

 

[neuroanatomist]

My first concern was that is this a feature that would work better in the amazingly awful DPP as opposed to LR. Hopefully it is/will be as seamless to 3rd party apps.

I found that ACR/LR did a good job, as good or slightly better than DPP. DxO did even better, and gave the widest corrected FoV for the 14-35/4 (DPP conversion matched 14mm on the 11-24/4, ACR was slightly wider, DxO was closer to 13.5mm). Computationally, distortion is pretty easy to map and fix.

When the first info was being shared here about canon pursuing this tech, it was also around the time of their plastic lenses being introduced (generally speaking, maybe, dont quote me)...so a a whole it looked like ways to simply save some cash (consumers perspective).

Plastic elements are generally cheaper to produce (note that not all plastic elements are created equal), and they're also lighter than an equivalent glass element would be. For example, with aspheric lenses there are ground/polished glass, moulded glass, and moulded plastic, and that's the order of production cost.

'Forcing' digital correction of distortion is about more than saving cash, it enables lenses that would not be possible without it. There's no way the RF 10-20/4 would be as small and lights it is if it was more fully corrected with lens elements. It's optically as good as the 11-24, it's wider, and it's not only smaller and lighter, it's much cheaper. Similarly, the Rf 14-35/4 could not have kept its 77mm filter diameter without requiring digital distortion correction.

 

[koenkooi]

Logic dictates that canon will be doing a lot of this moving forward (since they have very actively been using this feature), so it will likely be harder to avoid anyways. My first concern was that is this a feature that would work better in the amazingly awful DPP as opposed to LR. Hopefully it is/will be as seamless to 3rd party apps.
[…]

There is software that maps the Canon distortion correction data to DNG commands, which would make it work on any compliant DNG viewer. The downside is that that software also does its own debayering, so you can’t use fancy noise reduction or highlight recovery anymore.

Personally, I can live with waiting a few months for dxo and adobe to catch up. Recent lenses have had profiles available when the first units started shipping to customers.

The 200mm macro I desire likely won’t need correction

 

[Berowne]

Does the 17mm TS-E need correction in software? I guess not so much.
Now imagine Canon would make 10mm and 14mm Prime lenses without AF and IS - they could perhaps be superior in IQ without being too large and expensive and most correction done optical. But people want to have the easy-to-use zoom-lenses.