This is the point of anamorphic lenses - right? Has there been any empirical data for end quality of the result?
Upvote
0
Opinion: Love it or Hate it, Digital Correction is here to Stay
- Thread starter Richard-CN
- Start date
Yes, assuming that we can actually use both options and have an appropriate budget to suit.
Canon would clearly prefer that users only have one RF option though :-(
At least we have one option now!
Upvote
0
Yes. Regarding their original purpose on film it’s pretty clear that the significantly increased film area’s effect of reducing grain and increasing resolution overcame the softness created by optically stretching the image back out to correct the squeezed distortion. I’m sure there will be some calculations somewhere if you looked for it.
However, due to digital’s much greater resolution per area and cleaner output the anamorphic technique isn’t as relevant for improved quality. In fact because it is digital the desqueeze has to be done through interpolation, similar to distortion correction that this thread is discussing, and my understanding is that unlike film, this produces a technically inferior result, and so it’s used to specifically produce a less perfect ‘look’.
It’s worth noting that desqueezing is going to require much more interpolation than tweaking the distorted corners in our RF lenses.
Upvote
0
I agree the size and weight of my RF 85/1.2L DS are not an issue in use, though it’s not my most used lens (that would be the RF 24-105/2.8L Z, which weighs a little more).
But I don’t travel with either of those lenses. I will be traveling with the RF 14/1.4, and will have no qualms about packing it. The Sigma 14/1.4 is the same weight as and longer than the EF 11-24/4L, and when I traveled with the latter lens it was always a carefully considered decision that usually meant not taking some other lens. The RF 10-12/4 and the RF 14/1.4 combined are about the same weight as the EF 11-24/4 or Sigma 14/1.4, and don’t take up too much more space.
Upvote
0
I turned on my 2024 EOS R1's DLO for my 2006 EF 85mm f/1.2L II USM & 2008 EF 200mm f/2L IS USM.
All the noticeable CA, LoCA and other image quality oddities were absent from my SOOC JPEGs.
It was like having a newer gen EF lens without extra money much less RF L lens money.
All the noticeable CA, LoCA and other image quality oddities were absent from my SOOC JPEGs.
It was like having a newer gen EF lens without extra money much less RF L lens money.
Last edited:
Upvote
0
Same here, I need the DAM portion of Lightroom as well so I use DxO PureRaw as a plugin for LR. That gives the the lens correction and noise reduction capabities of DxO and all the good bits of Lightroom. Both DxO PL and PR have a trial period,
Upvote
0
DLO has basically reinvigorated my entire EF lens stable. I have yet to see the need by means of final image quality or mechanical function to upgrade to RF. I do enjoy playing with RF lenses as my friends bring them by, but my EF stable has been quite fit for purpose for what I enjoy photographing when combined with the R cameras. My EF stable covers the focal range from 20mm through 600mm in primes and zooms.
In fact, I'm starting to doubt that many of the RF lenses truly have a quality or mechanical difference that justifies the multi-thousand dollar price hikes vs their EF peers. Yes, some clearly outperform their predecessors — but not the majority. I feel that many RF lenses are living off of hype from the cream of the crop and the fact that EF is starting to fade frm the shelves. Not that I blame Canon for making a buck, but I think many (but not all) photographers would better off buying cheaper yet more reasonable priced glass than RF glass, especially with the inventory clearing deals popping up.
To be clear, I'm not saying the RF glass is bad — it's great on its own merit. I'm saying mm for mm most of it is not worth a premium for the performance difference, for my experience, when mm equivalent EF glass is on the shelf.
Just last year in Canada the EF 50mm f/1.2 and 24mm EF 1.4 II when on sale for ~ 50% off, new with warranty, placing them in the $1,500 +/- range. They are more than adequate for all of my family and friends, whether they print or post. And I'm willing to bet that most photographers aren't so extreme in their needs that the size or focus speed differences are justified in job-accomplishing terms. At the Olympics? Sure! The new 14mm lens for astro? Yup! Photos of children at graduation, gymnastics, hikes, picnics, family reunions, or travel? Not a chance. Any need for RF in those moments for the typical well heeled person would be gear-head need (and yes, I can fall into that category too) or lack of EF glass availability due to time.
I say this as a guy who can afford the newer stuff. But most of my family and friends cannot, and I'd rather see the cost-benefit balance of Canon's EF lenses return to the fold. If smaller is what Canon needs to do to put good glass in the hands of family and friends going forward, well so be it. But history and competition shows this is more a modern Canon choice, and one self-imposed, than anything else.
Last edited:
Upvote
0
OK, that semi-rant aside — to return to the topic of this thread, digital correction being here to stay, I have a few more thoughts:
First, I think that what the VCM line represents is a good thing. I very much enjoyed the classic Canon EF prime line-up of the 20mm, 28mm, 50mm, 85mm, 100mm, and 100mm macro EF gold-ring USM series. I now very much enjoy the L series upgrade for the equivalent prime line (but I kept the 20mm -- the rest went to my kiddo). With the addition of the 14mm VCM and the hybrid-shooting qualities of the set I do feel these are spiritual successors for that gold-ring line, but with an L-leaning bump in quality thanks to the passage of technology and time (in the same spirit anti-lock brakes were passed on to the consumer from airplanes and F1 cars). For the modern lens investor, the VCM line should bring decades of joy.
The fact that the image circle does not encompass the full sensor and digital correction is needed to make the full resolution image as promised by Canon in terms of the camera sensor is an implementation detail. You can like it (as some do) or not (as I am prone), but it is functional and the results seem overall good -- as least as good as a corrected image from a gold-ring prime, if not better. I think it's fine for Canon to pursue this strategy. Especially since it doesn't seem like Canon's software is making up anything that a standard stretch and lighten achieves elsewhere.
But I have been thinking about the lack of sensor coverage in another way. Canon promises a resolution based on a sensor when it sells a camera: 45mp for the R5 line, and 20-32mp for the R6 line depending on when you buy-in (I ignore the exact dot count here but do assume I am focused on the exact amount in the brochure). Here's the rub: if Canon were to move its total consumer accessible lens line to less than sensor coverage then I bet (and it's a very educated bet) that a class-action lawsuit could be brought to bear with the claim that Canon makes a false promise with its camera line as the advertised resolution of the camera is not being provided to the consumer. It likely wouldn't matter if the extreme lenses did provide coverage, because — and I speculate — likely most people buying an R6 camera are not buying $5k+ lenses. In other words, the economically-aligned lens options would not satisfy the marketing claim. It would probably come down to whether or not Canon could argue that the software transformation engine provided a de facto equivalent to the marketing claim, but were that true then there would be no need to advertise a megapixel count for the camera because AI can remarkably resize any photo to astounding outcome (indeed, a feature of the R1 and yet it still claims a megapixel count). So, I agree that software transformations are fair game for being a part of lens design, but I think that in the case of compensating for unused pixels the camera manufacturers producing lenses do need to be somewhat cautious. Third party lens makers would not be subject to the same issue as they aren't also selling a camera with declared capabilities to go with their lenses.
Put another way, Canon promises on the box and in the brochure that if you buy a 32mp R6 III it will take pictures using its 32mp sensor; Canon is not promising that it will take 31 or 30 or 29 mp pictures that get transformed into 32mp pictures using some software and a specific lens. But the VCM line with its smaller coverage and the software engine being applied is exactly that second scenario. If that approach were to become prolific then at some point someone will notice and make an expensive complaint.
First, I think that what the VCM line represents is a good thing. I very much enjoyed the classic Canon EF prime line-up of the 20mm, 28mm, 50mm, 85mm, 100mm, and 100mm macro EF gold-ring USM series. I now very much enjoy the L series upgrade for the equivalent prime line (but I kept the 20mm -- the rest went to my kiddo). With the addition of the 14mm VCM and the hybrid-shooting qualities of the set I do feel these are spiritual successors for that gold-ring line, but with an L-leaning bump in quality thanks to the passage of technology and time (in the same spirit anti-lock brakes were passed on to the consumer from airplanes and F1 cars). For the modern lens investor, the VCM line should bring decades of joy.
The fact that the image circle does not encompass the full sensor and digital correction is needed to make the full resolution image as promised by Canon in terms of the camera sensor is an implementation detail. You can like it (as some do) or not (as I am prone), but it is functional and the results seem overall good -- as least as good as a corrected image from a gold-ring prime, if not better. I think it's fine for Canon to pursue this strategy. Especially since it doesn't seem like Canon's software is making up anything that a standard stretch and lighten achieves elsewhere.
But I have been thinking about the lack of sensor coverage in another way. Canon promises a resolution based on a sensor when it sells a camera: 45mp for the R5 line, and 20-32mp for the R6 line depending on when you buy-in (I ignore the exact dot count here but do assume I am focused on the exact amount in the brochure). Here's the rub: if Canon were to move its total consumer accessible lens line to less than sensor coverage then I bet (and it's a very educated bet) that a class-action lawsuit could be brought to bear with the claim that Canon makes a false promise with its camera line as the advertised resolution of the camera is not being provided to the consumer. It likely wouldn't matter if the extreme lenses did provide coverage, because — and I speculate — likely most people buying an R6 camera are not buying $5k+ lenses. In other words, the economically-aligned lens options would not satisfy the marketing claim. It would probably come down to whether or not Canon could argue that the software transformation engine provided a de facto equivalent to the marketing claim, but were that true then there would be no need to advertise a megapixel count for the camera because AI can remarkably resize any photo to astounding outcome (indeed, a feature of the R1 and yet it still claims a megapixel count). So, I agree that software transformations are fair game for being a part of lens design, but I think that in the case of compensating for unused pixels the camera manufacturers producing lenses do need to be somewhat cautious. Third party lens makers would not be subject to the same issue as they aren't also selling a camera with declared capabilities to go with their lenses.
Put another way, Canon promises on the box and in the brochure that if you buy a 32mp R6 III it will take pictures using its 32mp sensor; Canon is not promising that it will take 31 or 30 or 29 mp pictures that get transformed into 32mp pictures using some software and a specific lens. But the VCM line with its smaller coverage and the software engine being applied is exactly that second scenario. If that approach were to become prolific then at some point someone will notice and make an expensive complaint.
Last edited:
Upvote
0
Megapixel count is not the same thing as resolution. Some of your long following fantasy stems from this misapprehension to an extent. But Canon doesn't even "promise" in the way you claim. Anyhow, with these lenses the camera does indeed produce an image of the MP count that it is "supposed" to - just that it is sometimes done with more digital manipulation than some other lenses. Is it not enough to simply dislike something - even in the face of evidence that it doesn't do what people fear? Why must what-ifs and unreasonable future scenarios be concocted?
Upvote
0
Indeed. My R1 and R8 give at least 24 MP images from all my lenses. If I use DxO and turn off the 3:2 constraint, some lenses give me >24 MP.
Most buyers of lenses with this characteristic aren’t even aware of it, and of those that are (likely from spending too much time on forums like this one), many don’t care.
Class action lawsuit? LOL. That would be as frivolous as me suing those complaining about this problem for the emotional pain and suffering of reading their complaints. If you know about this issue, and you don’t like it, don’t buy the lenses. Too simple, I guess. Better to go litigious.
Upvote
0
OK, let’s see what Canon writes down:
Canon Canada’s R6 III page:
www.canon.ca
Effective Pixels: Approx. 32.5 megapixels
And at Canon Europe:
32.5 megapixel resolution
(The word approximate does not appear near that text.)
Wax on about the difference between megapixels and resolution as we might, but Canon uses the terms simultaneously. Lay people don’t care about the difference and Canon marketing plays to that.
My R6 box has appropriate versions of those numbers stamped on as well. In terms of sales, yes that forms a promise in countries such as Canada due to truth in advertising laws.
I appreciate that the end image is of the proper size (thank goodness). I’m not going to sue them. I’m just speculating on what one danger of leaning hard on small image circles with digital compensation might lead to if taken to the extreme. I figured in part that’s part of the nature of this thread.
I do agree with what I think neuroanatomist implies in that the promise is serviced by the final image. Canon would have to fight on that premise, and the ruling would have industry implications for design to be sure.
Anyhow, I won’t shake the tree too much further on this. Just looking at this in the spirit of the thread from the perspective of a tech COO who prior was in charge of R&D pipelines for a large US based tech company.
Surely everyone will keep looking at the tech marvel part of this story.
Canon Canada’s R6 III page:
Canon EOS R6 Mark III | Mirrorless Camera
The Canon EOS R6 Mark III is built to meet your demands and deliver stunning results. Capture 32.5MP stills with impressive detail at up to 40 fps and enjoy the comfort of pre-continuous shooting to aid you during those unpredictable moments. With Canon Log 2 colour, 4K 120p slow motion, high...
Effective Pixels: Approx. 32.5 megapixels
And at Canon Europe:
32.5 megapixel resolution
(The word approximate does not appear near that text.)
Wax on about the difference between megapixels and resolution as we might, but Canon uses the terms simultaneously. Lay people don’t care about the difference and Canon marketing plays to that.
My R6 box has appropriate versions of those numbers stamped on as well. In terms of sales, yes that forms a promise in countries such as Canada due to truth in advertising laws.
I appreciate that the end image is of the proper size (thank goodness). I’m not going to sue them. I’m just speculating on what one danger of leaning hard on small image circles with digital compensation might lead to if taken to the extreme. I figured in part that’s part of the nature of this thread.
I do agree with what I think neuroanatomist implies in that the promise is serviced by the final image. Canon would have to fight on that premise, and the ruling would have industry implications for design to be sure.
Anyhow, I won’t shake the tree too much further on this. Just looking at this in the spirit of the thread from the perspective of a tech COO who prior was in charge of R&D pipelines for a large US based tech company.
Surely everyone will keep looking at the tech marvel part of this story.
Upvote
0
As much as I sing praises for in-camera DLO's ability to correct optical aberrations sadly it cannot fix hardware limitations. Modern sensors like the 45MP R5 Mark II and the high-density 32.5MP R7 reveal softness in older EF glass. The EOS R1 uses a high-speed bus that EF lenses cannot fully saturate. Native RF lenses provide 12-pin communication for faster data transfer and more precise AF tracking than the 8-pin EF system.
The price of RF lenses includes new tech like Voice Coil Motors (VCM) and Nano USM. These motors move heavy glass faster and quieter than old Ring USM motors found in lenses like my 2006 EF 85mm f/1.2L II USM or 1999 EF 500mm f/4L IS USM. Many older EF lenses use "Focus-by-Wire" systems. When these electronic motors fail the lens cannot be focused manually. Canon typically stops making parts 7 years after a lens is discontinued. If I bought the last 1988 EF 200mm f/1.8L USM with the 1st batch of 2003 EOS 10D before its 2004 discontinuation and 1989 EF 50mm f/1.0L USM when it was discontinued in 2000 I wouldn't be able to get brand new spare parts in 2026 as they're past this support window.
RF lenses are designed for the short flange distance of the RF mount. This allows for larger rear elements that hit the sensor with straighter light rays. This reduces purple fringing and corner softness. Older EF wide-angle lenses such as your 2007 EF 14mm f/2.8L II USM or 2007 EF 16-35mm f/2.8L II USM show more distortion on mirrorless sensors compared to native RF versions. The R1 and R5 Mark II also offer Coordinated Image Stabilization. This combines sensor movement with lens movement for up to 8 stops of shake correction. Older EF lenses often provide only 3 to 4 stops or zero correction if they lack lens IS.
Focus speed is a critical difference for professional work. The EOS R1 can shoot at 40 fps. Most EF lenses released before 2006 cannot move their focus elements fast enough to keep up with this rate. Using a 2001 EF 70-200mm f/2.8L IS USM on an R1 will result in fewer sharp shots during fast action compared to the RF version. For revenue-generating work the risk of a motor failure on a discontinued lens like the 1999 EF 300mm f/2.8L IS USM represents a total loss of the tool with no path for official repair.
The size and weight of RF lenses improve ergonomics for long workdays. Newer designs move the center of gravity closer to the camera body. This reduces fatigue for photographers. While EF lenses are cheaper on the used market they require an adapter which adds length and weight. For hobbyists this is a fair trade. For mission-critical professionals the lack of spare parts and slower communication speeds make pre-2006 EF lenses a liability.
As we approach the 10th year anniversary of the RF mount in 2028 more and more EF lenses will cease getting the 7 years of spare parts support with no path of repair except from donor lenses.
This is why I wish by 2017 I unloaded 30% of my oldest EF lenses and 2008 EOS 5D Mark II & 2009 EOS 1D Mark IV. By Q1 2024 unloaded 2014 EOS 7D Mark II & 2015 EOS 5Ds R to get the R1 & R5 Mark II released months later with these RF lenses
- 2019 RF 50mm f/1.2L USM
- 2020 RF 15-35mm f/2.8L IS USM
- 2020 RF 24-70mm f/2.8L IS USM
- 2020 RF 70-200mm f/2.8L IS USM
- 2021 RF 100mm f/2.8L Macro IS USM
- 2023 RF 135mm f/1.8L IS USM
- 2023 RF 28mm f/2.8 STM
- 2024 RF 35mm f/1.4L VCM
From 2019-2023 I hardly did enough photogrpahy to merit any upgrades.
Upvote
0
The argument that digital correction compensates for sensor coverage fails to account for data loss. When a lens like my 2009 TS-E 17mm f/4L or 2008 EF 24mm f/1.4L II USM is adapted to the R1 or R5 Mark II the sensor records the native image circle. If a lens requires extreme digital stretching to fix vignetting or barrel distortion it effectively discards peripheral pixels. This lowers the actual resolved detail compared to a native RF design. For revenue-generating work this "software stretch" reduces the quality of large-scale prints.
The EOS R1 and R5 Mark II utilize advanced "In-Camera Upscaling" and Neural Network Processing. These tools work best with high-frequency data provided by modern RF glass. Older designs like the predecessors of the 2007 EF 14mm f/2.8L II USM and 2011 EF 8-15mm f/4L Fisheye were built for film or early digital sensors. Their glass coatings and internal baffles often cause "sensor flare" or internal reflections when used on the high-gloss surfaces of modern CMOS sensors.
Mechanical reliability is the primary concern for mission-critical use. The 2011 EF 600mm f/4L IS II USM, 2011 EF 400mm f/2.8L IS II USM and 2011 EF 300mm f/2.8L IS II USM rely on specialized internal motors and focus groups. Once Canon ends the 7-year support window for these specific versions a single mechanical failure renders the lens permanent scrap. Unlike the newer VCM or Nano USM motors these older Ring USM units are no longer in production. For a professional using a lens that cannot be repaired within 48 hours is a significant business risk.
The EOS R7 features a very small and dense pixel pitch. Older zooms like the 2002 EF 24-70mm f/2.8L USM or 2003 EF 17-40mm f/4L USM often fail to "resolve" the 32.5MP density of an APS-C sensor. While DLO can sharpen the edges it cannot create detail that the glass failed to project onto the sensor. In contrast the RF "VCM" and hybrid lines are designed to exceed the resolution requirements of current 45MP and 50MP sensors.
The "Gold-Ring" EF era (including my 2000 EF 100mm f/2.8 USM Macro (non-L))represented a peak in value. However these lenses lack the 5-axis communication required for the R1’s most advanced AF tracking modes. While they work for family photos or "decades of joy" for hobbyists they lack the build redundancy and electronic speed required for modern professional revenue streams.
Upvote
0
Whereas I agree with much of which you write, I am not sure about this. The flange diameters of EF and RF are the same and so the rear elements of both have the same limits on size. Also, the further away the rear element from the sensor, the "straighter the light rays" so the longer flange distance EF should not be a disadvantage in that aspect. (This is different from designing uwa lenses.)
Upvote
0
Hello AlanF, long time no hear! Thank you for your input on lens design.
The EF and RF mounts share a 54mm inner diameter but the flange focal distance is significantly different. The EF mount requires a 44mm gap to clear the mirror box while the RF mount is only 20mm from the sensor. This shorter distance allows RF lens designers to place much larger rear glass elements closer to the sensor. This design principle called telecentricity ensures light rays strike the sensor at a more perpendicular angle. This reduces the "smearing" and chromatic aberration often seen in the corners of older EF wide-angle lenses like my 2007 EF 16-35mm f/2.8L II USM or the 2007 EF 14mm f/2.8L II USM.
Regarding the size of the elements the EF mount's 44mm distance forces a "retrofocus" design for wide lenses. This requires many glass elements to stretch the image back to the sensor. By contrast the RF mount allows the rear element to be almost as large as the 54mm mount itself. You can see this in the RF 100mm f/2.8L Macro IS USM. While my 2000 EF 100mm f/2.8 USM is an excellent lens the RF version uses its larger rear glass and shorter path to achieve 1.4x magnification and significantly faster AF tracking on the R1.
Furthermore the RF system uses a 12-pin electronic connection instead of the EF's 8 pins. This allows the R1 and R5 Mark II to handle much higher data bandwidth. This speed is required for Coordinated Image Stabilization where the camera sensor and lens elements move in perfect sync. Older EF lenses (even when adapted) cannot communicate fast enough to reach the 8-stop stabilization limits or the 40fps burst speeds of the R1.
Finally while long flange distances can lead to straighter rays in theory the physical mirror box of a dSLR limits how large that rear element can be. In practice modern sensors like the one in the R7 have very dense pixels that are sensitive to the angle of light. RF lenses are designed specifically for these digital sensors whereas EF lenses designed before 2006 were often optimized for film or lower-resolution digital sensors.
What I described above is why I wish I sold all my pre-2006 lenses by 2017. RF L lenses are awesome sauce!
Upvote
0
I've played with the in-camera upscaling as I feel like I was probably the only person on earth who used the IBIS-operated high resolution mode on the R5, and I've found that it adds detail that's not present in the original scene (even on the RF 85 1.2L). As with most AI-driven technology, you're not actually adding anything real; it's just guessing at best, and the results honestly look horrid to my eyes. The sensor shift tech was great - I've used it a few times in an Olympus body to produce high resolution raw files and it does the job quite nicely as long as the scene is still.
In terms of "sensor flare", I see this issue on the R1 with modern RF lenses including the RF 85L, 28-70 f/2, and even the RF 35 VCM. And sensor bloom where you get vertical flaring with mechanical shutter modes - which wasn't present on the 1DX II - is present on all RF bodies to date including the R1, R3 and R5 II (but oddly not on the 1DX III).
Back on the digital correction argument - Sigma just released a new 35mm f/1.4 DC DN II ART lens. This thing is almost perfect in terms of geometry, sharpness, CA control, is about the same size and is lighter than the 35L VCM. I would pay the same price as the Canon RF 35L VCM for that Sigma lens if there were an RF version. And you bet it performs better in terms of CA, sharpness etc than the RF 35L VCM.
I've said this before - if the RF VCM lenses were geometrically correct, I would have bought everything below the 50 (and probably the 50 as well - the 50 is however, geometrically appropriate). I actually cancelled my first-batch-in-the-country RF 35L pre-order after seeing the reviews, and affirmed my decision with an evaluation loan - but my EF II developed an issue that CPS were struggling to repair, so they offered me the replacement of either the EF II or the RF brand new - I should have gone for the EF, but I figured the increased contrast and size on the RF will be more practical in reality (but my photos are just not up to the same level of quality).
Perhaps there is a compromise on optical performance made to facilitate the use of the VCM AF system, or perhaps to allow for more economical repair or more resilient build quality to stand up to the rigours of professional use (something the Sony 35 GM absolutely fails at).
Upvote
0