First Looks at Canon RF 70-200mm F2.8 L IS USM Lens Image Quality

Del Paso

M3 Singlestroke
Aug 9, 2018
579
556
Maybe I'm wrong, but I think the "Image Quality" section is a bit disappointing, while contradicted by the excellent quality of the real pictures.
The sharpness and contrast I see on the portraits are actually convincing!
 
  • Like
Reactions: navastronia

koenkooi

EOS 7D MK II
Feb 25, 2015
527
312
Maybe I'm wrong, but I think the "Image Quality" section is a bit disappointing, while contradicted by the excellent quality of the real pictures.
The sharpness and contrast I see on the portraits are actually convincing!
The 70mm shots look very soft, as if they are out of focus.
 
Mar 14, 2012
2,305
193
Every image comparison between the R and SLR (5Ds R) favors the DSLR because of how the pixel information is processed by the camera. For example select the 200 f/2L IS or the 24-70 f/2.8L II for both cameras, and the DSLR will show sharper. The comparison to make is to compare something like the RF 24-70 IS to the EF 24-70 f/2.8 II using the R for both lenses, but unfortunately, there is no data for the EF 70-200 II or III on the R.
 

Del Paso

M3 Singlestroke
Aug 9, 2018
579
556
Till now, all RF lenses have been as good, often better than their EF counterparts.
Not so the 2,8/70-200 RF.
Still hoping the sample tested was an "Oktober Fest" lens...:unsure:
 
Last edited:
Mar 14, 2012
2,305
193
Can you explain further?
From Bryan's (TDP) review of the EOS R:


Unsurprisingly, the Canon EOS R vs. 5D Mark IV comparison shows the two cameras about equal with an important note. That note is that the 5D Mark IV's image is slightly sharper. While it immediately looks like the 5D IV is delivering better image quality, that is not necessarily the case.

We have long been using a low Canon Digital Photo Professional sharpness setting of "1" (on a 1-10 scale) for lens testing as sharpening quickly masks differences between lenses. While it seems that increasing the sharpness of an image is a cheap fix to poor lens quality, that sharpening quickly becomes destructive to details, ruining the fine image quality you were seeking in the first place. Camera manufacturers know that you like sharp images and they crank up the default sharpness in attempt to make you happy. Nearly always, that setting is at a very destructive level. Fortunately, you can adjust the sharpness level to your taste.

There is a long pipeline between the imaging sensor and the final image file and not all cameras utilize identical hardware and software (most often, they do not). More specifically to the point, the amount of sharpening showing in a final JPG image processed using the same settings is not always the same. And, that is the case here. Adding some headspace at the bottom of the sharpening scale doesn't seem like a bad idea.

If you zoom in your browser (try CMD/CTRL +, CMD/CTRL 0 to reset) while looking at the 5D IV noise results (discussed below), you will see some slight over-sharpening halos, slightly brighter colors around the border of the darker ones. I gave strong consideration to using a sharpness setting of "2" as the default for the EOS R (Canon defaults the EOS R unsharp mask strength setting to "4" vs. "3" in other recent EOS models), but the halos began showing at "2". So, we are staying the course with the site-standard sharpness setting of "1". You of course can increase your sharpness setting as desired. An EOS R setting of about "2" is approximately equivalent to the 5D IV at "1". Download the crops shared on this site and apply your preferred sharpening routine.
 

Mikehit

EOS 5D MK IV
Jul 28, 2015
3,229
418
From Bryan's (TDP) review of the EOS R:


Unsurprisingly, the Canon EOS R vs. 5D Mark IV comparison shows the two cameras about equal with an important note. That note is that the 5D Mark IV's image is slightly sharper. While it immediately looks like the 5D IV is delivering better image quality, that is not necessarily the case.

We have long been using a low Canon Digital Photo Professional sharpness setting of "1" (on a 1-10 scale) for lens testing as sharpening quickly masks differences between lenses. While it seems that increasing the sharpness of an image is a cheap fix to poor lens quality, that sharpening quickly becomes destructive to details, ruining the fine image quality you were seeking in the first place. Camera manufacturers know that you like sharp images and they crank up the default sharpness in attempt to make you happy. Nearly always, that setting is at a very destructive level. Fortunately, you can adjust the sharpness level to your taste.

There is a long pipeline between the imaging sensor and the final image file and not all cameras utilize identical hardware and software (most often, they do not). More specifically to the point, the amount of sharpening showing in a final JPG image processed using the same settings is not always the same. And, that is the case here. Adding some headspace at the bottom of the sharpening scale doesn't seem like a bad idea.

If you zoom in your browser (try CMD/CTRL +, CMD/CTRL 0 to reset) while looking at the 5D IV noise results (discussed below), you will see some slight over-sharpening halos, slightly brighter colors around the border of the darker ones. I gave strong consideration to using a sharpness setting of "2" as the default for the EOS R (Canon defaults the EOS R unsharp mask strength setting to "4" vs. "3" in other recent EOS models), but the halos began showing at "2". So, we are staying the course with the site-standard sharpness setting of "1". You of course can increase your sharpness setting as desired. An EOS R setting of about "2" is approximately equivalent to the 5D IV at "1". Download the crops shared on this site and apply your preferred sharpening routine.
Thank you for that.
My one comment would be that it may apply only to using DPP because a program like Lightroom does not as far as I know recognise Canon-embedded processing data whereas DPP will 'read' what the camera has done and apply it.
Or maybe I am missing something....
 

Act444

EOS 6D MK II
May 4, 2011
1,030
120
As I pointed out in the other thread - I don't think the comparison on that site with the EF version is apples-to-apples, and thus can't conclude much from it. OTOH, it seems to be the best native performer that covers 70mm for R users, edging out the long ends of both the RF 24-70 2.8 IS and the RF 28-70 2. Seems to more or less hold its own at 200mm as well

ETA: Differences are slight, though, so it's possible that real-world scenarios could mask some or all the apparent difference.
 

navastronia

EOS RP + 5D Classic
Aug 31, 2018
247
268
Looks like the RF has much less CA and slightly improved sharpness over the EF version III. Not leaps and bounds better for IQ and, IMO, worth the upgrade only if size and weight savings (and potentially autofocus and/or IS improvement) are important to prospective buyers.
 
  • Like
Reactions: YuengLinger
Mar 14, 2012
2,305
193
Thank you for that.
My one comment would be that it may apply only to using DPP because a program like Lightroom does not as far as I know recognise Canon-embedded processing data whereas DPP will 'read' what the camera has done and apply it.
Or maybe I am missing something....
Yes, that might be true. However, Bryan uses DPP and I'm assuming that the IQ files are processed that way. Therefore, it's relevant when comparing the images on his site.
 
  • Like
Reactions: YuengLinger

YuengLinger

EOS 5D MK IV
Dec 20, 2012
2,645
767
Southeastern USA
As a side-note, does it seem to others that this and other lenses are slow to be reviewed by numerous sites? I remember when any new lens had plenty of different reviews within about six weeks of release.

One I'm thinking about right now is the RF 15-35mm f/2.8L IS. Just seems to be a view silly youtubers doing little more than unpacking videos. Besides Bryan's thorough review, I see a kind of half-hearted effort on digitalcameraworld...Some "overviews," and some rehashes of rumors and Canon press releases.

Is Canon not sending out loaners? Is there just too much gear coming out? Is it that the reviewers can no longer make money? Sigh...

Did Dustin Abbott give up on Canon gear?
 
  • Like
Reactions: navastronia

shawn

EOS M50
Jan 28, 2019
26
25
As a side-note, does it seem to others that this and other lenses are slow to be reviewed by numerous sites? I remember when any new lens had plenty of different reviews within about six weeks of release.

One I'm thinking about right now is the RF 15-35mm f/2.8L IS. Just seems to be a view silly youtubers doing little more than unpacking videos. Besides Bryan's thorough review, I see a kind of half-hearted effort on digitalcameraworld...Some "overviews," and some rehashes of rumors and Canon press releases.

Is Canon not sending out loaners? Is there just too much gear coming out? Is it that the reviewers can no longer make money? Sigh...

Did Dustin Abbott give up on Canon gear?
I don't 100% trust any reviews or sites where they presume to be testing the resolving power of the lens. Simply put, resolving power is too complicated to test accurately. These sites can give you a general idea of how a lens will perform. Kind of a black and white is it good or bad.

One variable unaccounted for is that lens performance varies according to a variety of factors including distance to subject. For instance, telescopes are generally designed to focus on infinity, which is why they tend to outperform camera lenses for astrophotography.

A lot of people also don't know that there is not really a depth to depth of field. A lens comes into a focus at a point. That point needs to intersect a plane exactly.

The more megapixels the more exact the focus at 100% view needs to be. With today's high MP bodies I wouldn't trust anyone to nail focus over and over again to the extent that 100% views of the images for different lenses are actually accurate comparisons of two lens' resolving power.

Finally, photographing a perfectly flat surface is not necessarily an accurate representation of how a lens will photograph the real world.
 
  • Like
Reactions: Del Paso

SecureGSM

2 x 5D IV
Feb 26, 2017
1,358
330
To some degree you nailed it with regards to lenses perform differently at various distances to subject.
From my not so limited experience with Sigma Art lenses, they are not great at MFD. at all. They get much better at approx x30 FL and then excellent at infinity.
The area Where older Sigma Art lenses really suck is: transition from in focus to out of focus areas.
Modern Sigma Art : 85, 135 and especially 105 are much improved design and I suspect largely due to massively oversized front element. Not an optical engineer myself, just a gut feeling.
On another note, I would not hesitate recommending Modern Sigma art prime lenses to anyone, provided one lives in an area with adequate service coverage for the brand.
 

edoorn

EOS RP
Apr 1, 2016
243
162
apples and oranges, a bit hard to compare. I did check the 70-200 RF vs the 24-70 RF both at 70mm. They seem about similar there, and that's enough for me: I've tested the rf 24-70 myself and found it to be excellent, so if the 70-200 performs the same I'm more than happy with that.
 

navastronia

EOS RP + 5D Classic
Aug 31, 2018
247
268

Del Paso

M3 Singlestroke
Aug 9, 2018
579
556
I don't 100% trust any reviews or sites where they presume to be testing the resolving power of the lens. Simply put, resolving power is too complicated to test accurately. These sites can give you a general idea of how a lens will perform. Kind of a black and white is it good or bad.

One variable unaccounted for is that lens performance varies according to a variety of factors including distance to subject. For instance, telescopes are generally designed to focus on infinity, which is why they tend to outperform camera lenses for astrophotography.

A lot of people also don't know that there is not really a depth to depth of field. A lens comes into a focus at a point. That point needs to intersect a plane exactly.

The more megapixels the more exact the focus at 100% view needs to be. With today's high MP bodies I wouldn't trust anyone to nail focus over and over again to the extent that 100% views of the images for different lenses are actually accurate comparisons of two lens' resolving power.

Finally, photographing a perfectly flat surface is not necessarily an accurate representation of how a lens will photograph the real world.
I fully agree with your comment, lenses should be tested in real life.
If, for instance, you look at Leica M lens reviews ("chart testing"), the results are almost always disappointing.
Situation changes dramatically when taking "real" pictures.
My favourite quote, I think by Leica's former chief optical developer Dr. Mandler: "lenses are not made to photograph charts".
Reviews, yet, are reliable when they mention for instance CA behaviour of a lens, vignetting, or when a lens is actually a total "submariner".
Therefore, I'm waiting for comments by the RF 2,8/70-200's EXPERIENCED users...
 
  • Like
Reactions: edoorn

edoorn

EOS RP
Apr 1, 2016
243
162
I think in practice this could be a hit. IQ looks great; it's lightweight and compact to store and from what I hear, the first impressions are that the focus is very fast and silent. Event shooters like me find that very useful.

Furthermore, if there would be a 400mm 2.8 in R mount, I'd probably sell my EF 100-400 and 500 and get that to work in combination with the 70-200, and an extender (provided there's an R version coming). Would save me quite a bit of weight and some packing space on my wildlife trips!