SIGMA again announces EF-M lenses, and mount conversion service

[AlanF]

The Canon **EF** 50mm's are all pretty bad for sharpness. I've had all but the STM, which I think is the same as the MkI 50/1.8 that I have had. Also the 50/1.8 has a 5-bladed aperture whie the 1.4 might be 6-bladed, both pretty old-fashioned.

The 1.8 STM is very sharp. This is what opticallimits/photozone writes:
"Classic 50mm primes tend to be rather soft at max. aperture setting but they get extremely sharp when stopped down to around f/4 or at least f/5.6. This is mostly true for EF 50mm f/1.8 STM as well. However, unlike its predecessor, it is actually very sharp in the image center at f/1.8. The corners/borders have a decent resolution but the contrast falls apart here. The quality increases continuously towards medium aperture settings. The center is already excellent at f/2.8 and there's also a lift in contrast in the outer image regions. The real boost occurs at f/4 with very good borders/corners. The sharpness is outstanding across the image frame at f/5.6. As usual diffraction has am impact beyond f/8 but f/11 remains perfectly usable."

Canon EF 50mm f/1.8 STM - Review / Test Report - Analysis

Canon EF 50mm f/1.8 STM - Review / Test Report

opticallimits.com

It's a great little lens - try it (and especially before writing it off),

 

[melgross]

Canon wants to make money, and the question is whether there is any money for Canon in making more aps-c primes. Sigma can spread costs and build volume with marketing the same lens with different camera mounts, and Sigma may be willing to chase smaller volume than Canon. So, as always, we shall see what happens.

The interesting thing is that Canon’s M Series is increasing in sales. We can’t count Nikon’s mirrorless line, or Canon’s R line, because they’re new, and so starting from zero sales, they’re growing. But for pretty much everything else for everyone else, the situation is not how popular a line is, but how unpopular it is.

i say that, because we look at every company, and see sales dropping fairly rapidly. So lines that are dropping less rapidly are called “more popular”. That means that the M Series is one of the rare, these days, lines that are actually popular. Which is why, after all these years, Sigma is making lenses for it. Fast lenses that Canon isn’t yet making. My belief is that if these lenses sell well, Canon may look at that and decide that this is something they should be doing.

 

[SteveC]

Fast lenses that Canon isn’t yet making. My belief is that if these lenses sell well, Canon may look at that and decide that this is something they should be doing.

Precisely this. Canon's decisions with respect to the M line are justified by many here because of "marketing" but marketers are capable of failing to see a potential (or in other instances, making catastrophically poor decisions--e.g., new Coke). If this is actually a growing-in-absolute-terms segment of the market, hopefully they will say "Oh, we expected it to do fairly well, but didn't expect it to do this well! Let's see how far we can take it."

 

[ColinJR]

One of the major reasons I bought a Fuji camera (to accompany my canon) was because I got fed up with canon‘s lack of interest in making really good lenses for APS-C. Maybe these sigma lenses will inspire canon to step up and bring some L glass to the crop line. It would make much more sense for me to use a canon M camera as my small and light/travel/street photography kit than use a completely different system. I hate not being able to use my tilt-shift when I’m out with the Fuji, for example.

 

[LensFungus]

Maybe these sigma lenses will inspire canon to step up and bring some L glass to the crop line.

Meanwhile here is me waiting for Canon to bring some L glass to APS-C...

 

[SwissFrank]

It's a great little lens - try it (and especially before writing it off),

Did the optical formula even change, though? If not, then in effect I HAVE tried it. When EF started, they thought pros would either use the 50/1.0, or the 50/1.8. By not having a 1.4 I think they were pushing people towards the 1.0. That first 1.8 wasn't an L but it was built for pro use, if anything a bit stronger than the 1.4 that followed it. Once they had the 1.4, they re-issued the 1.8 in a 100% plastic version, but with the same glass. And I think the STM version was the same glass, although maybe with more modern coatings giving say less glare. I had the 1.0, 1.2, 1.4, and Mk I 1.8. I liked them all but they weren't especially sharp.

To compare to the other lenses, the RF 50/1.2 is from another planet. 1) It is sharper in the corners wide open than the others are in the corners at ANY aperture. 2) it is sharper in the corners wide-open than the others are at the center wide-open. 3) generally, it's 30lp/mm contrast line is above the EF 50's 10lp/mm line, meaning 3x the linear resolution, or actually 10x more area resolution. Once this lens is an option for you, there's simply not going to be something "worth trying" on an EF lense, unless you're looking for portability.

 

[AlanF]

.. 3) generally, it's 30lp/mm contrast line is above the EF 50's 10lp/mm line, meaning 3x the linear resolution, or actually 10x more area resolution...

Lensrentals has measured the MTF of the EF 50/1.8 STM to be 0.92 at 10lp/mm and 0.68 at 30lp/mm in the central regions. For the RF 50/1.2 to have 3x higher linear resolution, it would have to have values of MTF of 2.7 at 10lp/mm and 2 at 30lp/mm. As the maximum value of an MTF is 1.0, your statement that the RF has 3x higher resolution does seem somewhat puzzling.

Let me give a theoretical example of the comparison you made. Suppose you have two absolutely outstanding lenses. One has an MTF of 0.99 at 10lp/mm and 0.98 at 30lp/mm, and the other 0.979 and 0.96, respectively. The first is only marginally better than the second despite its MTF at 30lp/mm being better than the other at 10lp/mm. There is not a difference of a factor of 3 in their resolution, both are getting close to perfection.
So what is the difference? Crudely speaking, the first lens resolves 9.9 lp/mm at the 10 lp/mm level, and 29.4 lp/mm at the 30 lp/mm level. The second lens resolves 9.79 lp/mm at the at 10 lp/mm level, and 28.8 lp/mm at the 30 lp/mm level. That is, tiny differences.

 

[Sharlin]

The Canon **EF** 50mm's are all pretty bad for sharpness. I've had all but the STM, which I think is the same as the MkI 50/1.8 that I have had. Also the 50/1.8 has a 5-bladed aperture whie the 1.4 might be 6-bladed, both pretty old-fashioned.

To be pedantic, the 1.4 has eight aperture blades, like most Canon's 90s and 00s midrange and L lenses. The 1.8 STM has seven, and the newest L lenses almost all have nine, many-pointed sunstars being in fashion these days. Even the latest 18–55mm kit lenses have seven blades.

Meanwhile here is me waiting for Canon to bring some L glass to APS-C...

Yeah, that's not going to happen.

 

[stevelee]

I got the 50mm f/1.4 to use as a “portrait“ lens for my T3i. It was great for that.

When I first got the 6D2, I used the 100mm macro for that purpose. I found it not very pleasing. (It is a wonderful lens for other purposes, obviously especially macros.) So when Canon had a refurb sale, I bought the 85mm f/1.8. If I were still shooting crop cameras, I‘d still use the 50.

 

[Travel_Photographer]

I think it's great that are more options for Canon M cameras, but just to be accurate, and don't shoot the messenger, these are old lenses that Sigma just decided to slap an M-mount and electronics on.

The title of ALL these blog articles should be

"Sigma adapts their existing F1.4 Contemporary line of Sony and Four Thirds mount lenses to Canon"

Two Of the 3 lenses were introduced YEARS ago. Sigma just recently decided well I'll guess we'll just tap into the M market. Here's the 2016 release of the lens 3 years ago, the exact same lens "released" recently:

Sigma 30mm F1.4 DC DN Contemporary for Sony E-mount lens review

The Sigma 30mm F1.4 DN DC Contemporary was announced in February 2016, and sits atop Sigma's line of DN mirrorless lenses. It's currently the only APS-C F1.4 autofocus lens available for Sony E-mount and has an MSRP of $339. Can it best the Sony 35mm F1.8 OSS? Read more

www.dpreview.com

Google the 16mm and you'll see it's old too.

Sigma did nothing but take their old existing line of Sony and Four Thirds lenses and put on the Canon mounts and electronics and announce it like it's something amazing.

Don't get me wrong. I think that's great. If these lenses are useful for Canon shooters, awesome. But Sigma DID NOT create or cater to Canon M-series. You know how they're old lenses? You can take the 2016 version of the Sony and have them put on the Canon mount. And it's the same exact lens that you would buy today in Canon M. That's great. I love the idea of switching brands and having a lens manufacturer take your old lenses and switch mounts.

The point? Sigma did not create new lenses for the Canon M system because they thought there was a market. They just took their years' old lenses and put a Canon mount on it because they figured hey, cool the M series is doing well, might as well put in the SMALL INCREMENTAL DOLLARS to put a Canon mount on our existing old lenses.

I'm only writing because I don't want anyone to see the Sigma post and think that 3rd party manufacturers are designing lenses JUST for the Canon m-mount. To my knowledge, no 3rd party one has created a lens just for Canon M.

I can't be clearer though... I think it's GREAT that these lenses are available to those who want them. I just don't want anyone thinking this is indicative of where the market is going.

 

[scyrene]

maybe we are tired of this blind defense of small size.
if fuji can, so does canon.

Blind defence? By me or by them? I'm just telling you how it is. Sure we can imagine things being different, but not all our dreams come true. If you want wide aperture, adapt older lenses, buy a different brand, or move up to full frame. Not every system can offer everything you want.

 

[scyrene]

And yet even the most basic telephoto kit lenses in Canon's EF-S line exceed 200mm. An EF-M equivalent to the 75-300 would even be a quantum leap forward. Doesn't stop me from longing for, well, longer, of course. Hence my statement about imagining a tiny 100-400 equivalent. Doesn't make it any more likely. But still...

No 100-400 lens for APS-C is ever going to be 'tiny' no matter how much you stretch that word. Sorry.

In truth, Canon has given the entirety of the APS-C lineup the cold shoulder, and sold a bunch of cameras in spite of themselves. [...] Canon COULD do more. And they haven't. Whether by laziness or (more likely) perceived lack of market, the end result is still the same.

I look at it like this: they have found a position where they sell a lot of this type of camera. Adding the extras you and some others desire wouldn't push up sales enough (in their view) to justify the extra effort. It's not laziness or malice, it's just good business.

 

[SwissFrank]

Lensrentals has measured the MTF of the EF 50/1.8 STM to be 0.92 at 10lp/mm and 0.68 at 30lp/mm in the central regions. For the RF 50/1.2 to have 3x higher linear resolution, it would have to have values of MTF of 2.7 at 10lp/mm and 2 at 30lp/mm. As the maximum value of an MTF is 1.0, your statement that the RF has 3x higher resolution does seem somewhat puzzling.

For the RF 50/1.2 to have 3x higher linear resolution, it would need to be 0.92 at 10*3lp/mm = 30lp/mm. It does, or thereabouts.

You're tripling the contrast ratio for a given resolution and of course getting nonsense results. Instead you should be tripling the resolution for a given contrast ratio.

Let me give a theoretical example of the comparison you made. Suppose you have two absolutely outstanding lenses. One has an MTF of 0.99 at 10lp/mm and 0.98 at 30lp/mm, and the other 0.979 and 0.96, respectively. The first is only marginally better than the second despite its MTF at 30lp/mm being better than the other at 10lp/mm. There is not a difference of a factor of 3 in their resolution, both are getting close to perfection.

OK, I can see you understand my point. And yes, by my method of calculation, the one that's .98 at 30lp/mm would be 3x sharper linearly, about 10x sharper by area, than one that's .979 at 10lp/mm. That's because you could blow up a picture from the first by 10x and have .98 contrast where the second lens only has .979 contrast without being blown up at all.

Now, it's a bit academic because we don't have infinite-resolution film/sensors to do this with, but if we did, my point would hold.

And what makes it academic isn't that my method is too theoretical in all cases for a real-world comparison, but rather that you've made lenses with utterly impossible resolutions. In real life, we're not comparing .98 vs. .979 but rather .92 to .68 to .4 and so on.

Lensrentals doesn't have figures for the RF 50 yet? If not, I suspect they may give it .92 at 10lp/mm in the corners, and give it .92 at 30lp/mm in the center. I'm not a fan boy and have no skin in the game. I'm just saying the RF 50 completely redefines the amount of detail you'll get with a Canon lens. It's not merely twice or three times as sharp. OTOH it is so huge I doubt many will use it much, and it's got vignetting from hell. (I hardly use mine; I use the 24-105/4IS or my Leica M's 35/1.4ASPH.)

 

[AlanF]

For the RF 50/1.2 to have 3x higher linear resolution, it would need to be 0.92 at 10*3lp/mm = 30lp/mm. It does, or thereabouts.

You're tripling the contrast ratio for a given resolution and of course getting nonsense results. Instead you should be tripling the resolution for a given contrast ratio.



OK, I can see you understand my point. And yes, by my method of calculation, the one that's .98 at 30lp/mm would be 3x sharper linearly, about 10x sharper by area, than one that's .979 at 10lp/mm. That's because you could blow up a picture from the first by 10x and have .98 contrast where the second lens only has .979 contrast without being blown up at all.

Now, it's a bit academic because we don't have infinite-resolution film/sensors to do this with, but if we did, my point would hold.

And what makes it academic isn't that my method is too theoretical in all cases for a real-world comparison, but rather that you've made lenses with utterly impossible resolutions. In real life, we're not comparing .98 vs. .979 but rather .92 to .68 to .4 and so on.

Lensrentals doesn't have figures for the RF 50 yet? If not, I suspect they may give it .92 at 10lp/mm in the corners, and give it .92 at 30lp/mm in the center. I'm not a fan boy and have no skin in the game. I'm just saying the RF 50 completely redefines the amount of detail you'll get with a Canon lens. It's not merely twice or three times as sharp. OTOH it is so huge I doubt many will use it much, and it's got vignetting from hell. (I hardly use mine; I use the 24-105/4IS or my Leica M's 35/1.4ASPH.)

I do understand your point 100% - it is completely wrong, and it's not academic. And it's nothing to do with my choice of lenses in the second example that are theoretically very good - they were chosen as another example to show the mathematical flaw of your analysis. I had pointed out first that, using the actual measured numbers for the MTFs for the 50/1.8 STM, for the RF 50/1.2 to be of 3x higher resolution it would have to have impossibly high MTFs of much greater than the theoretical maximum of 1.0. The RF 50/1.2 might be a much better lens, but there is absolutely no way it has 3x the linear resolution at 10-30 lp/mm.

 

[SwissFrank]

I do understand your point 100% - it is completely wrong,

If so, why were you multiplying contrast by three, not resolution, when you wrote

they were chosen as another example to show the mathematical flaw of your analysis

For the RF 50/1.2 to have 3x higher linear resolution, it would have to have values of MTF of 2.7

There's no mathematical flaw. The first lens in your example really WOULD have (slightly) higher contrast at 3x the resolution, given an ideal sensor.

it would have to have impossibly high MTFs of much greater than the theoretical maximum of 1.0

I'm really puzzled: you actually made a good example, but again you're totally misunderstanding how math works, even though I've corrected you already on this. Until you say that you now understand why this "triple the contrast number" is not how to do this, I can't spend any more time talking to you.

 

[BillB]

The EF 50mm f/1.4 is a 1993 design, great for the 35mm film cameras of the day, but sorely lacking when compared to modern lens designs. I don't need to see the Sigma to say that it's probably going to be a better lens. [And I'm saying this as someone who dislikes Sigma lenses and swore never to buy another!]

The Canon 50mm f1.4 is pretty sharp stopped down to f2.8. So, if you can be happy with a 50mm f2.8, the Canon is worth thinking about.

 

[AlanF]

If so, why were you multiplying contrast by three, not resolution, when you wrote
There's no mathematical flaw. The first lens in your example really WOULD have (slightly) higher contrast at 3x the resolution, given an ideal sensor.
I'm really puzzled: you actually made a good example, but again you're totally misunderstanding how math works, even though I've corrected you already on this. Until you say that you now understand why this "triple the contrast number" is not how to do this, I can't spend any more time talking to you.

Don't waste any more time talking to me if you don't wish to. However, you might look at measurements of the acutance of the RF 50/1.2 on the R compared with the EF 50/1.8 on the 5DIV that has the same sensor and see if you can spot the 3-fold difference. I am not saying that acutance = resolution, but DxO does calculate it from the MTFs.

Canon RF 50mm f/1.2L USM on Canon EOS R vs Canon EF 50mm f/1.8 STM on Canon EOS 5D Mark IV

www.dxomark.com

 

[SteveC]

No 100-400 lens for APS-C is ever going to be 'tiny' no matter how much you stretch that word. Sorry.

But would it be significantly smaller than it would be for full frame? And, hypothetically: if you went down half a stop or even a full stop in aperture over the full frame, would it become smaller still, enough to justify the loss of aperture? (I wonder how fast they can "back of the envelope" questions like that, internally?) Maybe you could get something small by comparison with that 100-400L, even if it's not 'tiny.'

Those are technical questions, to be sure, and no such lens would ever come to fruition if Canon doesn't perceive a significant market for it, even if the lenses are significantly smaller. So long as the M series is thought of as entry level/must be tiny, the answer will be no.

 

[blackcoffee17]

Meanwhile here is me waiting for Canon to bring some L glass to APS-C...

View attachment 186966

Canon always stated that the L designation is for full frame lenses only. Except some cine lenses and the compacts with fixed lenses.
So i don't we will ever see an L lens which is not full frame. But i don't care about L as long as they produce some quality lenses like the 32mm 1.4.
That lens is basically L quality, except for build quality which is ok.

But it's annoying that while every other system has at least one higher quality standard zoom, Canon only got the mediocre all-plastic and
dark 15-45 and the all-plastic 18-150.
Annoying that you have a 32MP amazing sensor and $1000 M6II and you can only pair it with cheapo plastic lenses with F6.3 aperture.

 

[blackcoffee17]

But would it be significantly smaller than it would be for full frame? And, hypothetically: if you went down half a stop or even a full stop in aperture over the full frame, would it become smaller still, enough to justify the loss of aperture? (I wonder how fast they can "back of the envelope" questions like that, internally?) Maybe you could get something small by comparison with that 100-400L, even if it's not 'tiny.'

Those are technical questions, to be sure, and no such lens would ever come to fruition if Canon doesn't perceive a significant market for it, even if the lenses are significantly smaller. So long as the M series is thought of as entry level/must be tiny, the answer will be no.

At 400mm the size of the lens is 95% defined by the aperture and not the sensor size. A 400mm 5.6 lens needs a 71mm diameter front element, no matter the sensor size. And all other lens element need to be sized accordingly.
Smaller aperture, like 6.3 and less corrections would make a lens smaller but dropping 1 stop would be too much. Don't think anyone wants a 400mm F8 lens.