I recently ended up going down a bit of a rabbit hole comparing lenses from 10-15 years ago with lenses from the past 5 years, and I noticed some trends.
Sharpness - Has greatly improved
Coma - Has greatly improved
Chromatic aberration - Has improved quite a bit
Bokeh - Has gotten smoother
Flaring - Has greatly improved from major manufacturers
Optical Vignetting - Seems to have gotten a bit worse overall
Distortion - Has gotten worse as more lenses rely on digital correction
Vignetting - Has gotten much worse overall. We seem to be accepting a bout a stop more now with digital corrections
On the positive side, the difference in sharpness and coma was quite noticeable. On the negative side, the difference in Vignetting was quite noticeable. We used to call close to 2 stops high and almost unacceptable, and now close to 3 stops is becoming common. Lens designers are probably relying on digital distortion and vignetting corrections to keep lenses smaller. Worse optical vignetting is also probably a sign of smaller lenses.
Of course, there are lenses which are outliers, but I found the overall trends interesting and thought I would share for anyone curious.
My gut is that sample variation has improved for all the major manufacturers. Once we hit 30MP+ sensors, people started paying much more attention to sample variation. But I haven't really seen anything concrete comparing brands. If anything, it semas like if a lens from a big brand has high sample variation at this point, it's model specific not brand specific.
@DWOfPaul you forgot to mention the most important lens quality: weight+size. That is where the most progress has been made. I recently held my Sony 24-70 f/2.8 Mk2 in one hand, and my old Canon EF 24-70mm f/2.8 in another. The delta was quite dramatic. Looks like we traded vignetting and distrotion for less weight.
Another positive trend is the return of aperture rings. #1 reason why I left Canon and mostly shoot E-glass on Nikon bodies. They still didn't get the memo.
Noted the differences especially with CA control and bokeh of today’s lenses vs 15 years ago. I almost bumped the 5 year old thread of EF 50L vs 50/1.2 GM. I have the former and it’s an amazing lens but no match for today’s modern lenses. The good news is today’s software ie Photoshop and DXO PL 10 Elite are equal to the task of removing purple and green fringing. So I’m staying the course. However I might still consider the upcoming Viltrox 50/1.2 if the price is right.
DWOfPaul wrote:
I recently ended up going down a bit of a rabbit hole comparing lenses from 10-15 years ago with lenses from the past 5 years, and I noticed some trends.
Sharpness - Has greatly improved
Coma - Has greatly improved
Chromatic aberration - Has improved quite a bit
Bokeh - Has gotten smoother
Flaring - Has greatly improved from major manufacturers
Optical Vignetting - Seems to have gotten a bit worse overall
Distortion - Has gotten worse as more lenses rely on digital correction
Vignetting - Has gotten much worse overall. We seem to be accepting a bout a stop more now with digital corrections
On the positive side, the difference in sharpness and coma was quite noticeable. On the negative side, the difference in Vignetting was quite noticeable. We used to call close to 2 stops high and almost unacceptable, and now close to 3 stops is becoming common. Lens designers are probably relying on digital distortion and vignetting corrections to keep lenses smaller. Worse optical vignetting is also probably a sign of smaller lenses.
Of course, there are lenses which are outliers, but I found the overall trends interesting and thought I would share for anyone curious....Show more →
The root of the “this is now better but that got worse” is understandable: there is no free lunch in optical design. For example, it’s really hard to get rid of coma without retaining some distortion and transmission falloff; correct the latter two and the coma resurfaces. Then you can correct a lot of aberrations using exotic glass, aspheric surface grinds and complex optical designs, but now weight and size increase dramatically.
What is nice is that things like transmission falloff and optical distortion are relatively easy to correct after the fact. Much more difficult to increase resolution or remove some aberrations like stigs after the fact.
Optical design tools have improved out of sight in the last 5 or so years, but lens makers have definitely leaned heavily into computational software to correct some aberrations in chasing smaller and lighter lenses. A highly optically corrected UWA lens for example would be much larger and most people seem not to care about built-in lens profiles doing the heavy lifting and prefer the smaller lens. Look at a lens like the Tamron 12-20 f/2.8, in the DLSR days such a lens would have been a beast.
Jack Flesher wrote:
The root of the “this is now better but that got worse” is understandable: there is no free lunch in optical design. For example, it’s really hard to get rid of coma without retaining some distortion and transmission falloff; correct the latter two and the coma resurfaces. Then you can correct a lot of aberrations using exotic glass, aspheric surface grinds and complex optical designs, but now weight and size increase dramatically.
What is nice is that things like transmission falloff and optical distortion are relatively easy to correct after the fact. Much more difficult to increase resolution or remove some aberrations like stigs after the fact. ...Show more →
Interesting, I never realized coma correction could end up hurting vignetting, but it makes sense.
Personally, I find 3+ stops of vignetting excessive. Our sensors have definitely gotten better, and we can probably recover a good 5 stops of detail today, but if you're losing 3+ stops to Vignetting correction, that only leaves you about 1-2 stops for shadow / exposure adjustments, especially for something like astrophotography where vignetting is rarely wanted. Also, for low light photography, you are usually already pushing the ISO. If you are at ISO 6,400 and have to push 3 stops for vignetting, your corners are now effectively ISO 51,200 before any other adjustments.
DWOfPaul wrote:
Interesting, I never realized coma correction could end up hurting vignetting, but it makes sense.
Personally, I find 3+ stops of vignetting excessive. Our sensors have definitely gotten better, and we can probably recover a good 5 stops of detail today, but if you're losing 3+ stops to Vignetting correction, that only leaves you about 1-2 stops for shadow / exposure adjustments, especially for something like astrophotography where vignetting is rarely wanted. Also, for low light photography, you are usually already pushing the ISO. If you are at ISO 6,400 and have to push 3 stops for vignetting, your corners are now effectively ISO 51,200 before any other adjustments....Show more →
It's why the Sigma 15 fisheye is marketed as a dedicated "astro" lens. Without rectilinear correction, you virtually eliminate coma and vignetting issues as well as keep round objects round at the corners of the frame.
I think there has been a relaxing of discipline on apertures.With DSLRs, a zoom lens hardy ever exceeded f/5.6 on the long end. F/6.3 was considered a major compromise. With mirrorless cameras having better AF, and improved performance at higher ISO, lens designers seem willing to accept f/6.3, f/6.7 and even f/8 without question. However, those mirrorless camera features don't change depth of field
Nikon boasted that the Z-mount was the widest and therefor could gather more light, then issues lenses like the 16-50 f/3.5-6.3. Really? f/6.3 at 50mm!
Jack Flesher wrote:
It's why the Sigma 15 fisheye is marketed as a dedicated "astro" lens. Without rectilinear correction, you virtually eliminate coma and vignetting issues as well as keep round objects round at the corners of the frame.
Thanks for the info. I may have to pick up the Sigma 15mm f/1.4 fisheye next year, it's been tempting me since it came out.
mklass wrote:
I think there has been a relaxing of discipline on apertures.With DSLRs, a zoom lens hardy ever exceeded f/5.6 on the long end. F/6.3 was considered a major compromise. With mirrorless cameras having better AF, and improved performance at higher ISO, lens designers seem willing to accept f/6.3, f/6.7 and even f/8 without question. However, those mirrorless camera features don't change depth of field
Nikon boasted that the Z-mount was the widest and therefor could gather more light, then issues lenses like the 16-50 f/3.5-6.3. Really? f/6.3 at 50mm!
I think this more has to do with AF in the DSLR days. My understanding, and somewhat my experience using TCs on the Nikon 300mm f4 PF, is that DSLR AF got much worse after f5.6, I can definitely confirm F8 made AF take a hit even on my D500. Since mirrorless focuses on the sensor, aperture does not seem to affect AF performance as much. With my sony 200-600mm f6.3 + TC 1.4 AF is fine at f9 on my Sony cameras and adapted to my Nikon Z cameras.
On the size side, f6.3 is about 1/3 smaller then f5.6 for a corresponding focal length. So especially at longer focal lengths, this can make quite a big difference
I just watched analog insights youtube on the v4 50mm Summicron. In it he quotes the leica handbook (I owned that book, have no idea if I still have it) and in it it talks about about the cost effectiveness of the planar cemented doublets in the lens. It must be much easier to make and more reliable to make having flat surfaces rather than curved ones to cement. In his video he talks about how Leica referenced economic benefits. A twofold benefit, a less complicated optical design levels up all the variables to a much more consistant product.
Compare that the Tessar design, Kingslake famously said was one of the most difficult to produce. The tolerances were so tight. I might have owned a dozen or more of this design and all bar two (now that I think about it) were average performing lenses. Needing to be stopped down 4-5 stops to be usably sharp. The two that were good were outstanding.
I doubt the lens lottery will ever change. Robert Monaghan made curated newsgroup postings specifically on this subject. And that was in the 90's.
I've pretty much given up on fancy lenses and go for classic lens designs now.
The trend towards smaller, lighter, and slower aperture lenses, many with with severe distortion and CA is sickening. As well there is the retroscopic trend of manual focus Chinese copies of mid-20th century primes and cameras with controls like the 50s/60s/70s. Video centric has complicated things for those of us that just shoot stills. The trend towards mediocrity will likely continue.
With all the meds people are supposedly talking to lose weight, maybe people could manage reasonably sized bodies with finger grips and lenses fast enough to not suffer so much diffraction with high-res sensors.
I'm still hoping against hope that Nikon will find someone to make a high-res sensor, not the same old 45MP with less resolution than we had in 2015.
old-gregg wrote:
I recently held my Sony 24-70 f/2.8 Mk2 in one hand, and my old Canon EF 24-70mm f/2.8 in another. The delta was quite dramatic. Looks like we traded vignetting and distrotion for less weight.
You do take on almost a stop more vignetting at the wider end. Distortion's close enough to call them the same, but the Sony is a tiny fraction better at any setting. At least comparing vII to vII. Still, you get less weight and significantly higher resolution. That's gotta be worth something.
Can I ask for clarification and examples of 'the *retroscopic* trend of manual focus Chinese copies of mid-20th century primes' please? My thesaurus does not know this word, and I would like to hear more about your concerns, and why it may be sickening or disturbing, or even noteworthy.
The move to what the old lenses delivered is well underway in the light cine field at present, it's very widespread among Cooke, Zeiss and Leitz (Leica). That's my context here, but would like to know more.
This video talks about weight saving for teles , and the secret is in the motor which provides more torque to move the optical elements thus allowing the designer for more freedom in designing the optics
&t=1205s
(see Sony vs Nikon section)
EB-1 wroteThe trend towards smaller, lighter, and slower aperture lenses, many with with severe distortion and CA is sickening.
I have the same attitude. For example, the smaller and slower Sony 24mm f/2.8G is the right lens for me because I value weight above all else, and I don't need fast aperture for wide angle. But its barrel distortion was so enormous, that I got the GM instead.
At the same time, when I am looking at software-corrected output from the G... and I see nothing wrong. The lens+software together produce great results, and yet I can't bring myself to own a lens like that. The contradiction is killing me.
Maximilian may not be thrilled with Leica's APO-Summicron-M 35/2 and Noctilux 35/1.2, both of which use a 10/5 design of decidedly curved elements in their three doublets and one triplet; both use UV-cured epoxies.
AI tells me that 'three decades of chemical research' provide confidence that the early-generation synthetic UV-cured adhesives that rendered some Summilux-M v3 lenses as irreversibly damaged will not be repeated.
old-gregg wrote:
I have the same attitude. For example, the smaller and slower Sony 24mm f/2.8G is the right lens for me because I value weight above all else, and I don't need fast aperture for wide angle. But its barrel distortion was so enormous, that I got the GM instead.
At the same time, when I am looking at software-corrected output from the G... and I see nothing wrong. The lens+software together produce great results, and yet I can't bring myself to own a lens like that. The contradiction is killing me.
Both optical and digital distortion correction will reduce image sharpness, although they do so for different reasons. Optical correction involves compromises in the optical design, while digital correction requires geometric warping and resampling of the image, which can reduce fine detail toward the edges.
A lens design is simultaneously trying to control spherical aberration, coma, astigmatism, chromatic aberration, field curvature, distortion and so on.
These don't behave independently so optimizing a design for very low distortion requires compromises elsewhere.
If the lens design accepts more distortion in order to achieve higher resolution, and then let software correct the distortion. It can still mean a net resolution gain.
