Tongfamily (Richtong.com Sync)

I haven’t gotten the latest A9 or the A7 series but have an older A7R3 with a 40MP sensor. With the pandemic, I have basically taken zero photos for the last two years. Now with the A7R4 out with a 60MP sensor at $3.5K and the really remarkable Sony A1 which is a crazy $6K for the body that is 50MP and which can continuously shoot at 30 fps.

But, the main goal is that while you get new bodies, the main investment is really in the “glass” which is a great lens that will live through several camera generations. As an example, I’ve had a Canon 70-200 F/2.8 zoom lens that I’ve used for 12 years across two generations of camera backs from a Rebel Ti to a Canon 5D Mark II, and while not as great as the latest is still remarkable and you can sell it back used.

And, the truth is that the overall sensor quality of DSLR systems is pretty remarkable. As an example, the DXO Mark review of sensors has no less than 15 cameras with an overall score of 97 or hight. And many are 40MP and above.

So given this, what are lenses that I would recommend now. Well, the first thing is that with such high-resolution cameras, you won’t really see this super-resolution unless you are shooting from a tripod, this is such fine-grained stuff that even at 40MP, you need one. So a great tripod is a must and this is really for landscapes or if you need to take sports and zoom in. Remember that a 135mm lens as an example at 40MP will yield cropping a 270mm equivalent image with 10MP which is very good for all but really large poster-sized prints. And this is done at F/1.8 so much less light than a typical F/2.8 that a zoom lens has.

TL;dr Sony 24mm Sony 50mm and 135mm for Pros,

So to get to the punch line, if money is really no object and you want less which is going to last decades, Sony 24mm f/1.4 GM, the Sony Zeiss 50mm f/1.4, and the Sony 135mm G Master are the ones to get and give you wide, prime and a good telephoto. Note that at the long end with a 135mm G, if you crop you can get a 10MP image at 270mm which is pretty amazing.

For more reasonable budgets, you just dropping a tiny bit in terms of speed saves a lot and the lens are quite good, so the Sony 50mm f/1.8, which is literally 10x less expensive, is a great choice or the Sony FE 55mm f/1.8 at $998 is very good as well. Then on the long side, the Sony FE 85mm f/1.8 at $600 is a great choice.

More details

So given this what are the lenses that I would recommend now. For this, I like to use the scientific measurements that are done a DXOMark and especially from folks that are doing true LWPH (line width per height) measures measurements like Roger Cicada does at Lens Rentals or that Optical Limits does. With The-Digital-Picture has the best MTF charts for you to stare at which they get from an OLAF testing machine. The DXO Mark is OK, but their measurement is proprietary so it is hard to know what they really mean, but in general, all these are pretty correlated and they are also not doing any more updates. I find that since DXO Mark has a nice sorting tool I start with that system and then move to confirm them by looking at other “bench reviews” as well as subjective ones as well from folks like Ken Rockwell.

Why Prime (and not Zoom) Lens?

One quick note is that while I started with a system of using 24-70 F/2.8 and then 70-200 F/2.8 because with just two lenses I had the ideal travel system, today because prime (or fixed) lenses have so much higher image quality, I instead have settled on a three-lens prime system, that is wide at 24mm, then regular at 50mm and finally a long 135mm. This works great except when you need to change the lens and that is slow, but normally I find that when looking at a subject like a landscape that’s pretty easy and I have time to switch. It is harder with sports like volleyball, but I find then I take some shots wide, then some at zoom and regular. And yes, someday, it probably makes sense to just get two bodies (eg buy a Sony A1 and keep the Sony A7R3 around) so that you have two lenses at the ready.

Modular Transfer Function (MTF) vs resolution

Note that now that you get MTF it is actually quite hard to figure out what effective resolution really means. You have to basically say at a particular MTF that’s acceptable, what effective resolution do you get. To do a little math with Sony A7R3 sensor a 42MP 35.9mm x 24mm sensor is 7976 pixels wide x 5317 pixels high. So the ultimate resolution with a perfect lens would have 5317/2 or 2,658 line pairs across 24mm or a maximum resolution with a perfect lens of 110 lp/mm. That would give a perfect 48MP with a perfect resolution which is impossible with existing technology. This means that resolving at 50lp/mm means the lines are just 2 pixels wide and so you can see that having an MTF of 0.7 is pretty incredible at that level.

When I was first buying lenses for the Sony A7R3, lots of the Sony lenses were consumer-grade, but in the last four years, they have really filled out what they call their G Master line and these are generally the best of the bunch. That being said I do have a Zeiss which I love, but today here is what I would suggest for the 40MP and above Sony cameras. Today those are A1, A7M4, and A9M2 judging first by sorting all the Sony FE mount lenses in DXOMark. Note that DXOMark has kind of left the station in terms of doing lens, most of the reviews are now five years old, but they are still useful. The best site was called photozone.de but is now called Optical Limits (although they have also stopped doing many reviews and are at 42MP for their Sony FE mount reviews.

Camera Labs and Lloyd Chambers also do MTF testing and it is helpful that instead of doing LW/PH (line widths per picture height, it calculates this as a percentage of the overall sensor, so you can see what overall resolution you are getting). The other site is Ephotozine which does do the MTF in LW/PH, this is basically the resolution of the lens on a bench. That site is really hard to sort through, but you can sort by the vendor to get the best Sony lenses for instance.

Interpreting an actual MTF chart is actually pretty complicated, but basically here is how they work typically using Imatest or OLAF software:

1. The X or horizon axis starts with zero which means the center of the lenses and then in millimeters typically moves out to the corner, so you are seeing how the resolution drops off as you go from the very sharp center all the way to the corner. All lenses drop off a bit, but the more it maintains sharpness (that is the straighter the line the better, but every real lens will show roll-off). For a full-frame 35mm lens, these will typically be 0mm from the center, 5mm, 10mm, 15mm, and then 20mm from the center.
2. The Y or vertical axis is the amount of contrast that you are getting. The test is basically a set of black and white lines. So 100% contrast means that the modular transfer function is completely sharp. That is the line transition goes from 0 to 1 like a step function. If you have 50% contrast that means you have a ramp and if it is 0% that means you don’t get any change at all. The perfect lens will be 100% contrast change. You never in a real lens get to 100%, but 90% is really good. And 70% is very good.
3. Then you will see a series of lines, this is done at different line widths per millimeter. That is 30 lines per mm of height as an example is very fine detail and it should be the lowest, that is it is so tight. Here a little math is useful, a full-frame camera has a sensor that is 35mm wide and 24mm high (35mm get it, btw this is not the diagonal measurement that is common in lots of televisions, because it sounds big, but just the height). To translate this 30 lw/ph is 30 * 24mm
4. Now the last piece is Sagittal and Meridional lines basically are the lines radially from the center out edges or lines which are perpendicular to that, there are different resolutions depending on the line positioning
5. Then you have different apertures, every lens wide open (that is the smallest f-stop, like f/1.8) is going to be less sharp, then they will get sharper and sharper until a maximum (typically f/4 or f/5.6) then as you close the lens, you will get diffraction effects so the sharpness decreases (typically it’s bad at f/16).
6. The final thing is that you can take all of this and end up with a chart that shows you line width/pixel height, which is different than line width per millimeter, this just tells you to have many lines you get for the full horizontal resolution. As an example, if you are getting 4100 lw/ph then this translates into getting 4100 lines vs the sensor vertical height. For a 40MP camera (assuming a 4:3 aspect ratio so the 40MP = 4/3* height^2 is 5500 pixels height so this lens resolves at 4100/5500 at 75% of what is possible. See as an example the ephotozine review vs. the digital picture which shows the modular transfer function (MTF) at 50 lp/mm which is the typical measure of “can you resolve it”
7. The way to think about it is that at low MTF you lose resolution, everything becomes basically grey and you can’t resolve anything. Theoretically when you get to an MTF of 10% then you have in effect lost all resolution. So put another way, if you care about the effective resolution then you want to see how things do at a very high lp/mm and see when it hits MTF of 0.1
8. But most of the time the pros are looking at the MTF and for regular humans, there is some definition of minimum MTF that give you effective resolution. But I think that looking at 50 lp/mm and what MTF threshold you get gives you a much better sense.

Understand F-stops

This is a strange thing, but because F-stops of f-number are reciprocal, it seems like going from F/22 to F/16 is massive and from F/1.8 to F/1.4 is just a little, but the reverse is true. Technically what this means is that it is the ratio between the focal length and the diameter of the aperture of the lens.

And, the traditional numbers are set so that each change is a halving of the light available. Since we are talking about diameter, you take the square root of the number which is \frac{1}{\sqrt{2}} or \frac{1}{0.7071}, so the normal sequence that is halving the amount of light each time is, so the right way to think about it is that you get 4x less light as you go from f/1 to f/2 to f/4 and f/8 (which makes sense the amount of light in is going to vary as the 2 \pi (\frac{D}{2})^2 or the square so the in-between figures are 40% larger. So you can see the F-stops are really round so F/1.4 should really be F/1.41 etc¨

\begin{aligned}
f/1 &= f/2^{0.5 * 0} = f/(2^0) = f/1 \\
f/1.4 &= f/2^{0.5 *1} = f/(2^{0.5}) = f/1.41\\
f/2 &= f/2^{0.5*2} =  f/(2^1) = f/2 \\
f/2.8 &= f/2^{0.5*3} = f/(2^{1.5}) = f/2.82 \\
f/4 &= f/2^{0.5*4}= f/(2^2) = f/4 \\
f/5.6 &= f/2^{0.5*5} = f/(2^{2.5}) = f/5.65 \\
f/8 &= f/2^{0.5*6} = f/(2^3) = f/8 \\
f/11 &=  f/2^{0.5*7} = f/(2^{3.5}) = f/11.313 \\
f/16 &=  f/2^{0.5*8} = f/(2^4) = f/16
\end{aligned}

What it also means is that some of the very small numbers are actually big changes, so if you have an f/1.2 or an f/1.8 lens, the changes are actually pretty large, so instead of double from 0 to 1, if you increase by half that so that compared with an F/1.4 lens, an f/1.2 light can sense 50% more light and f/1.8 is 50% less:

\begin{aligned}
f/1.2 &= f/2^{0.5*0.5} = f/2^{0.25} = f/1.189 \\
f/1.4 &= f/2^{0.5*1} = f/2^1 = f/2 \\
f/1.8 &= f/2^{0.5*1.5} = f/2^{1.75} = f/1.682
\end{aligned}

Part of the way to understand what this means is to realize that f-stops tradeoff against exposures, so to get the same amount of light you need to half the exposures, so these give you the same amount of light. Exposures in photography are also done backward as fractions and you need to go 4x longer

• f/1 at 1/1000 seconds
• f/2 at 1/250 seconds
• f/4 at 1/62.5 seconds
• f/8 at 1/16 seconds

And then if you add the intermediates, you get 2x differences

• f/1 at 1/1000th seconds
• f/1.2 at 1/750 seconds
• f/1.4 at 1/500 seconds
• f/1.8 at 1/375 seconds
• f/2 at 1/250 seconds
• f/2.8 at 1/125 seconds
• f/4 at 1/64 seconds
• f/5.6 at 1/32 seconds
• f/18 at 1/16 seconds

So you get a sense that having an f/2 lens is actually much slower than 1.4, basically, an f/1.4 lens is twice as sensitive as an f/2 lens.

Lens Focal Length vs Magnification

I have to say the camera business is all over the place in terms of understanding things, the f-stops are inverted vs the amount of light, the shutter speeds are inverted vs the amount of light, and then for focal lengths, these are inverted vs the degree of magnification. Put another way, by doubling the focal length, you get 2x the magnification. And the thing to realize is that a 35mm full-frame actually has a diagonal of 50mm which you get with the hypotenuse of a 24x35mm sensor (24^2+35^2)^0.5 which is 43.2mm but the industry settled on 50mm which is actually 1.15x magnification as being standard 1x zoom

• 50mm. This is considered the standard length but gives you a 1.15x magnification which is called 1x which is the standard so stay with me on that.
• 25mm would be 25mm/50mm half that so 0.5x zoom or 2x wide-angle
• 20mm. This doesn’t seem like much change, but it is the ratio so actually quite a bit 50mm/20mm or 2.5x wide-angle
• 12.5mm is 12.5mm/50mm is 0.2 is 4x wide-angle

Or put the other way:

• 50mm. 1.15x magnification or 1x zoom
• 100mm 2.3x magnification or 100/50mm is 2x zoom
• 135mm is 135mm/50mm or 2.7x zoom so quite different
• 200mm 4.6x or 4x zoom
• 400mm 9.2x or 8x zoom
• 600mm is 600mm/50mm or 12x zoom or 9.2*1.5 or 13.8x zoom

And there are some intermediates that you see a lot, but basically, a small change at the lower end results in big changes in how wide-angle it is and what to get (tl;dr if you can about image quality get the 28mm if you really want wide, then 24mm is about the lowest you should go).

Sony FE Mount Recommendations for Prosumers

These are expensive lenses, there is no doubt about it, but if you are want something that will last decades, then here are some choices:

1. Carl Zeiss Batis 25mm F2 Sony FE. This is an older lens, but with a Sony A7R2 and is relatively slow, it had a swapping effective 36MP rating for a 42MP sensor. This is what I actually have and it works great, but see the new Zeiss 28mm Otus or if you really want a big landscape then the Sony 24mm below which is what I would currently recommend as it is much faster. And on the 60MP Sony A7R4 it is a 48MP effective resolution (again I’m not sure what minimum MTF is being used here. From Camera Labs, we can see the actual MTF across the center, and it’s about the same 80% of the total possible. Also, this lens works best at F/4 where it gets to maximum edge sharpness. The Digital Picture testing shows that at 50 lp/mm you get 0.6 MTF at the center which is pretty decent although the lense. But that it does fall off badly at the edges so even the pretty course 20 lp/mm falls to 0.4-0.4 MTF at the edges
2. Sony FE 24mm f/1.4 G Master. This lens came out in 2018, so I got the older Zeiss one, but if I were starting again, I’d go with this one and it is 50% faster than the f/2 above. The way to read the charts is to look at the number of line widths per pixel height (LWPH) at the center and at the edges to figure out how sharp it is. In this case, the peak is 4,050 lines at F/5.6, it’s pretty typical that you get better sharpness stopped own this comes out. And it really shows when you look at the Lens Rental review that the performance is just spectacular (remembering that it is very hard to design high-resolution wide lenses like this, with 50 lp/mm it is just 0.55 at the center, but this is significantly better than Canon and Sigma for instance that start at 0.4 or 0.35 MTF.
3. Sony Carl Zeiss Planar T* 50mm F1.4 ZA. This lens is 8 years old and it is still one of the best at 40MP against a 42MP imager. As an aside, this shows how long a prime lens can last. This is just a phenomenally sharp lens, particularly in the center. As an example from Digilloyd at 50lp/mm it has a 0.7 MTF so to turn that into English, that means it can resolve 100 lines per millimeter and the lense is 24mm high, so you get 2400 lines and it means nominally if 0.5 MTF is the minimum there is still more resolution at the center. As a comparison, Nikon 50mm f/1.4 at 50lp/mm is at a 0.3 MTF so can’t resolve it, and is the same at 20 lp/mm so quite a bit less.
4. Sony FE 135mm F/1.8 GM. Another Lens Rental review shows an incredible 0.8 MTF at 50 lp/mm at the center. That is truly an incredible result where 0.5 MTF for 50 lp/mm is considered very good. It does drop to 0.4 MTF at the edges, so it is the center that is amazing. It doesn’t say what F-stop is being used, but I assume this is at maximum resolution F-stop which is usually f/4 or f/5.6

As an aside, you can see in DXO Mark, the tradeoffs between zoom and prime, so the very expensive but wonder Sony FE 70-200 F2.8 GM OSS at $2600 is still remarkable giving 38MP/42MP effective, but not quite as sharp as these other systems and it is also much slow (half the speed) which is the big tradeoff.

There are some other choices if you really want to go wider than the 28mm, but there are tradeoffs in terms of the resolution but 24mm might be OK if the resolution is less important and you need a little big bigger wide-angle 50/24 is 2.08x wide vs 50/28 is 1.78x wide:

The Sony FE 20mm f/1.8 G is actually the lens that I would get, it is very wide and when doing mountains, you really want that. This is superwide so you should expect a lot of falloff in MTF, but it is really more of a specialty lens, with a significant falloff so that at 50 lw/mm it falls to 0.15 MTF at the edge. So I would get this if you really wanted this specialty lens.

Budget offerings

This is big heavy glass, so the more budget-friendly options have some “pixel peeking tradeoffs”, but in the real world, they are going to work really, really well:

1. Samyang AF 35mm f/1.8. Going to budget typically means moving to their party lens folks. This 35mm is a moderate wide-angle and very fast at f/1.8 or the Tamron 35mm f/2.8. It is much slower, but for most modern cameras that’s not much of a problem if you can get ISO 3200 photos cleanly as an example. Personally, I would get a fewer number of better lens, so I’d rather get a great 50mm (see below) than a cheaper wide-angle.
2. Sony FE 35mm f/1.8. Yes, Sony has a version of this as well.
3. Sony Zeiss Sony FE 55mm f/1.8. This has some lens flare issues, but is much cheaper than the 50mm f/1.2 as an example.
4. Sony FE 85mm f/1.8. This compares really well with the way more expensive Sony FE 85mm f/1.4 GM