Canon officially launches the RF 28-70mm f/2.8 IS STM

[Frodo]

Incidentally, that’s not consistent with Canon’s MTF for the wide/long ends, though based on TDP the RF 28-70 is a bit sharper at 70mm.

Optically, it seems to be a wash between the two, each being slightly better at some focal lengths. The real differences aren’t IQ, but focal range and aperture, size and weight.

I had/have two RF 24-105/4 Ls. The 28-70 is sharper than my current 24-105 (which is sharper than the previous one) outside of the centre. There may be sharp RF 24-105s, but neither of mine were stellar. These are my subjective opinions and I'll do more in-depth testing next weekend, which I intend to share. Image quality was an important criterion when getting this lens and I'm not disappointed.

 

[neuroanatomist]

I had/have two RF 24-105/4 Ls. The 28-70 is sharper than my current 24-105 (which is sharper than the previous one) outside of the centre. There may be sharp RF 24-105s, but neither of mine were stellar. These are my subjective opinions and I'll do more in-depth testing next weekend, which I intend to share. Image quality was an important criterion when getting this lens and I'm not disappointed.

I do wonder about copy variation or a possible bad batch of RF 24-105/4L lenses. @SwissFrank posted several times that his RF 50/1.8 was sharper than his RF 24-105/4L, and that seems unreasonable (not questioning his observations or yours, but lenses do vary). My RF 24-105/4L is not quite as sharp as my RF 24-105/2.8L, but it's close. It's also close to the RF 70-200/2.8L in the overlapping region. That's consistent with the MTFs for the relevant lenses.

Personally, I test all lenses when I buy them (I have a set of the same ISO 12233-type charts that Bryan/TDP uses, the cost for the largest one is more than that of the RF 24-105/4L). Canon's MTFs are theoretical/ideal, which is why I test them to ensure they perform as expected based on Canon's designs.

 

[Walrus]

Today I got my RF 28-70 STM; first impression is good, size is nice and light, but in the hands still feels solid and "full", it doesn't feel a plastic empty barrel.
I wanted to test it side to side with my EF 24-70 f2.8 II L USM before selling it away, but an eBayer made a direct offer I couldn't refuse, so it went before the new one arrived.
I had done some test pictures indoor at my home library (I usually have an outdoor wall with lot of details I use, but the day I sold the EF it was pouring rain) with the EF, replicated today with the RF; I marked the floor with the tripod position, but of course it's not easy to exactly replicate everything perfect after 4 days, and extreme corners may have suffered from a slightly different left/right rotation of the camera on the tripod.
And again, it's home testing, nothing scientific.

Testing was done on R6, sturdy tripod, 2sec delay, remote control; light was normal led warm diffused house lights illumining the library.
First, shooting in raw, the RF seems to be slightly wider, but also slightly more tele, then the EF at the same marked focal lengths; I actually did not shoot a test picture at 24mm to compare, but 28mm vs 28mm, the RF look definitely wider then that. I would say (if we consider the EF's markings to be correct) that the RF looks like a 26-75 in raw; I'll try shooting jpegs, I suspect that way it's going to probably crop slighly more on wideangle to make it look better, matching closely the EF's 28mm FOV.

Center sharpness and contrast are always better on the RF at any focal length and aperture combination; at 70mm f2.8 the RF is sharper then the EF at 5.6 which is quite amazing.
Corners differ some more: at 28mm the EF is equal or marginally better then the RF, at 35mm they're pretty much equal, at 50mm the RF is enormously better then the EF, at 70mm the RF is equal or marginally better then the EF.
The RF always vignette less then the EF (when both use their std profile correction in Lightroom).

I would say that at 28mm the RF really suffers the electronic barrel distortion correction, then corners equalizes at 35mm, gets hugely better at 50mm where there should be no optical distortion and so no correction is applied, and gets back suffering a bit at 70mm (I remember from Chris Frost's test that at 70mm there was pincushion distortion stroger then expected), while still not being worse then the EF's performance.
At centre and near centre, there's no competition, the RF always comes out first, with one or even two stops (like at 70mm) of sharpness advantage.

Light transmission wide open seems comparable, I shot in manual with exact same settings, and the lights, being wall/ceiling home fixtures, were of course at the exact same distance to the millimetre from the library, and didn't find any real visibile brightness difference. If I really have to give the edge, I would give it to the RF, but it's just a hint more brightness (seen also on the histogram that imperceptibly moves), no more then 0.1 stops, so basically no difference at all.

STM motor seems as fast as the EF's USM was; focus pulling in video was not tested, and lens IS is yet to be tested in the field in the next weeks/months.

Globally I would say for an extra 300€ difference, moving from a (relatively) old used L lens to this brand new non-L lens is still super positive, and it's worth the money spent.
The RF is basically inferior to the old EF only in the 28mm borders and corners due to extreme distortion correction, and has not a real advantage in borders and corners at 70mm for the very same reason: for any other aspect I'll conservatively say that the RF is at least matching the EF (so it's not a step down), with some aspects like 70mm central sharpness or 50mm corner sharpness where the RF absolutely outperforms the old L lens.

 

[Walrus]

Some details and highlights; EF 24-70 f2.8 II L USM always on the left, RF 28-70 f2.8 IS STM always on the right

The RF in raw looks wider then the EF; the latter was framed at the edge of the books, but the RF shows a nice extra chunk of the library frame


28mm top right extreme corner, where the EF seems clearly better then the RF


But on the opposite side, bottom extreme left corner, the RF is way better then the EF; so I would say that camera orientation was not 100% matching between the two sessions, and that the RF may also be pretty sharp in the extreme corners and borders if properly focussed. So I will actually call it even, the EF has not a real field relevant advantage over the RF even in the extreme corners, if the plane of focus is correct


This is 50mm bottom right extreme corner and border, it's the focal length where the RF majorly outperforms the EF in the extremes, and that's an all 4 corners and borders; the "DZA" on green in the super extreme last millimetres of the corner seems like central sharpness compared to the EF performance


Lastly the 70mm central sharpness, both wide open; the RF clearly shows the print reticle on the pink and orange books, and also on the red library frame the micro detail is crisp and sharp, while the books on the EF look like solid colours, without print details. The EF if stopped down to f5.6 slightly surpass the RF at f2.8 in terms of contrast, but the micro detail of the RF wide open is still better and crisper then the EF at 5.6 so the RF (on a 20mpx sensor) seems to resolve much more finer lines then the EF.

 

[Philnick]

I didn't do any comparison testing of my own before I sold my EF 24-70 f/2.8L II after getting the RF 28-70 f/2.8 - the online test chart shots at The-Digital-Picture.com convinced me, and the halving of the carrying weight (and shortening of the lens) made the sale. The addition of optical IS in the RF was a cherry on top, since both my R7 and R6/II have IBIS. Ironically, after selling the EF and mount adapter, I actually made money on the deal, even taking into account having to buy an aftermarket lens hood for the RF from JJC. Couldn't be happier with the change.

 

[Walrus]

I didn't do any comparison testing of my own before I sold my EF 24-70 f/2.8L II after getting the RF 28-70 f/2.8 - the online test chart shots at The-Digital-Picture.com convinced me, and the halving of the carrying weight (and shortening of the lens) made the sale. The addition of optical IS in the RF was a cherry on top, since both my R7 and R6/II have IBIS. Ironically, after selling the EF and mount adapter, I actually made money on the deal, even taking into account having to buy an aftermarket lens hood for the RF from JJC. Couldn't be happier with the change.

Actually the test charts at TDP (I made the comparison there, too, before pulling the trigger) were not convincing me in full, especially on corners and borders; but as you said, halving size and weight, and getting the IS as an extra bonus, was too tempting, and as soon as I found a good rebate I bought it immediately, and as you can read in my posts, I'm super happy, too.

Also got a cheapo hood, too, but the cherry on the cake was Canon actually sending me a special copy with L optics

 

[Jethro]

Today I got my RF 28-70 STM; first impression is good, size is nice and light, but in the hands still feels solid and "full", it doesn't feel a plastic empty barrel.

Thanks for the detailed comparison. I've just ordered one, and waiting for it to arrive at the store. I also bought a cheapo hood! Like 1/3 of the Canon price ...

 

[Philnick]

I like the red ring you put on it in your photo. I did something slightly different: I celebrated the cost and weight savings by decorating it with a green rubber band with a few red twist ties on it, and said "Happy No eL" (i.e., "Merry Christmas!")

 

[mkamelg]

A few days ago, a review of this lens model appeared on the OpticalLimits website:

Canon RF 28-70mm f/2.8 IS STM Review - OpticalLimits

Review of this more "budget-oriented" fast standard zoom lens for Canon R cameras.

opticallimits.com

 

[Frodo]

Thanks for posting that link mkamelg. I appreciate Klaus' reviews and find them valuable and they normally reflect my experiences. He values the excellent quality at the wide to mid range, the quick AF and the compact design, and I agree.
However, he pans the edge and corner quality at 70mm, but I certainly don't see such a difference between 40 and 70 mm as he reports, indeed I would call them close to equal. I wonder what distance he tests his lenses and if there is some field curvature. Its a pity that his tests of the RF 24-70/2.8 and RF 24-105/4 are on the 30MP R and so not directly comparable as I have found the 28-70 to be at least similar, but usually noticeably better than the 24-105 at similar focal lengths, apertures and points in the frame.
He criticises the retractable lens design, while also appreciating the compact size of the lens. Either you love this or hate it. I like the small size and don't mind having to extend it to take photos, but I can't understand why functions such as changing modes or shutter speeds are not available when the lens is retracted. A similar comment can be made about his criticism of auto-correction. This enables the compact size that he praises and as far as I am concerned, its the quality of the image that matters.
Elsewhere he comments that "The build quality is otherwise fine, although more in line to what you’d expect from a consumer-grade lens". Well, the inner tube wobbles less than the RF 24-105, and given the decentering issue I had with that lens, I rate the build quality of the 28-70 at least as good, if not better. He notes "There’s even some degree of weather sealing.". The lens rings appear to have the same gaskets at the 28-70/2.0, but the switches might not sealed. I think that's pretty good.
Finally, I agree about Canon not including a lenshood!

​

 

[bn5fg]

Thanks for posting that link mkamelg. I appreciate Klaus' reviews and find them valuable and they normally reflect my experiences. He values the excellent quality at the wide to mid range, the quick AF and the compact design, and I agree.
However, he pans the edge and corner quality at 70mm, but I certainly don't see such a difference between 40 and 70 mm as he reports, indeed I would call them close to equal. I wonder what distance he tests his lenses and if there is some field curvature. Its a pity that his tests of the RF 24-70/2.8 and RF 24-105/4 are on the 30MP R and so not directly comparable as I have found the 28-70 to be at least similar, but usually noticeably better than the 24-105 at similar focal lengths, apertures and points in the frame.

I own the lens and completely agree with you. I reached out to Klaus about the 70mm data potentially being incorrect, but he responded with something like, "Christopher Frost also said this." Well, okay—but test data from Bryan tells a different story: https://www.the-digital-picture.com...eraComp=1697&SampleComp=0&FLIComp=0&APIComp=0

 

[neuroanatomist]

I own the lens and completely agree with you. I reached out to Klaus about the 70mm data potentially being incorrect, but he responded with something like, "Christopher Frost also said this." Well, okay—but test data from Bryan tells a different story: https://www.the-digital-picture.com...eraComp=1697&SampleComp=0&FLIComp=0&APIComp=0

Copy variation is real. It's problematic when relying on just 1-2 reviewers. Personally, I test every lens that I buy with a method similar to Bryan/TDP (down to having the same charts he uses, though the set of them costs more than some L-series lenses). Knowing how my copy performs is what is most useful to me.

 

[Frodo]

Neuro, I agree with copy variation. I've personally experienced it with the RF 24-105/4L, RF 16/2.8 and RF 50/1.8. James Reader did an extensive test of the RF 28-70/2.8 with similar Canon lenses and it performed very well. The RF 24-105/4L was surprisingly poor.
I cringe when reviewers (including Chris Frost) state "Thanks for Canon providing [selecting?] this lens for me, but this review of the [carefully selected] lens is independent ".

 

[mkamelg]

In my hands since yesterday. I bought it online in the official European Canon online store, using the internal Easter promotion (a few hundred PLN less than the official price in Poland). My copy was produced in mid-January 2025.

I think I got a very good copy and at both extreme focal lengths there is a slight difference in favor of this lens model compared to my RF 24-105mm F4L IS USM (also purchased online in the official European Canon online store a few years ago), which I also consider a very good copy in terms of image quality. The biggest difference between both lens models is of course in the nominal aperture values. The f/2.8 aperture allows for a more interesting bokeh effect at close distances, and in combination with the lens stabilization to go down from ISO when taking photos with limited access to natural light.

Taking advantage of the beautiful weather I was able to take some comparative photos (CRAW + JPG) with the RF 24-105mm F4L IS USM and RF 28mm F2.8 STM lenses. Focal length 28mm at f/2.8 (photo mode Av) and at the aperture selected by the camera (photo mode P) and focal length 70mm at f/2.8 (photo mode Av) and at the aperture selected by the camera (photo mode P). JPG files were recorded using Picture Style Fine Detail. All lenses were attached to the R5 Mark II. All photos were taken handheld from the same position.

Download or view comparison photos:

Microsoft OneDrive

onedrive.live.com

It's 2025 and if someone asked me now what to choose for their first universal walking zoom, whether the RF 28-70mm F2.8 IS STM or maybe the RF 24-105mm F4L IS USM, I would definitely recommend buying the RF 28-70mm F2.8 IS STM, even despite its unusual lens locking mechanism.

I bought a JJC LH-73D lens hood for the lens, a replacement for the original Canon EW-73D. Knowing me, I'll probably never use it, but it's better to have it than not, because you never know when it might come in handy.