Image quality

There are all sorts of misconceptions about image quality in digital photography. For instance, people commonly believe that the resolution, or the number of megapixels, define image quality.

I’ll illustrate this with a screenshot from dpreview’s comparometer:

As you can tell, the top left image is from Sony A7II, the one I’ve been using since 2016. The bottom left is from the one I just ordered, the Sony A7RV. The top right is from the Canon 5d, the camera I’ve been using since 2006 and which Biljana was using until very recently. The bottom right is from the Pentax 645Z, the medium format camera.

As you can tell, other than some white balance differences, they are all basically the same image with different amounts of magnification. This means that the difference in resolution determines how big you can print the image without perceivable loss of fine detail.

This means that doubling of the resolution means that the image is printable on double the paper size, and we happen to have a standard of paper size measurements, in fact two, A and B:

Basically, every larger size (smaller number) is produced by mirroring the smaller size along the longer side, thus doubling the surface.

If, for example, a 12.7MP image from Canon 5d can produce a high quality B2 print (which I have done), an image of double the resolution, 24-26MP, can produce a B1 print of same apparent quality. This, of course, assumes that everything else, like noise and the amount of actual resolution measured in line pairs, scales equally.

Other than printability on large paper sizes, image quality is not affected by sensor resolution. There are, however, several other factors that determine image quality: noise, color depth, and dynamic range. Noise is obvious – it can degrade the image in appearance if it is excessive. Dynamic range is also easy to understand – it’s the ability to resolve greater number of brightness levels. In essence, one ev (exposure value, or aperture value) is twice the amount of light. With every ev of dynamic range, there’s a 100% increase in the level of brightness. This means that the dynamic range is 2 to the power of n, the same way binary numbers are defined by the bit depth of the variable type; 8-bit means 256 possible values, 16-bit means 65536 and so on. Today’s sensors can resolve over 14 ev of dynamic range, where slide film resolved 5 ev, and best BW and color negative emulsions resolved 10 ev. This means that everything above 10 ev is excellent, but using it in a real picture might require tonal compression in processing.

Color depth, however, is somewhat less clear as parameters of image quality go, but I would define it the same way I would dynamic range, because it’s the same thing: the ability to define gradient of primary colors, where every pixel is defined by three binary components of certain bit depth, for red, green and blue. 8-bit color depth means a color gradient of 256 shades for each of the 3 components. 16-bit color depth means 65536 shades for each of the 3 components, and so on; again, it’s the 2 to the n-th power. Of course, the ability to convert a signal from the sensor into a n-bit format doesn’t mean there’s actuall n-bits of data in the source, assuming the analog to digital converter doesn’t introduce its own issues. You can read the analog data from a small smartphone sensor into the 16-bit numberspace, but there won’t be 16 bits of color data in there. So, the ability to define discrete shades of colors across the large dynamic range is what differentiates between sensors with “thin” and “thick” colors. The difference in color depth is visible at any image size and is much more important for the perception of image quality than resolution, which only becomes relevant when you enlarge the image. So, the luminance and chrominance dynamic range is what defines the number of brightness levels and color tones a sensor can capture. When we introduce the noise, which contaminates both luminance and chrominance data, we get all the parameters of image quality.

So, what does this mean, translated to the world of actual cameras? It means that the pictures from my current A7II and A7RV will look exactly the same, unless I decide to print over a meter wide, in which case the A7RV images will look more detailed if you come so close that you no longer perceive the whole picture. As for the color depth and dynamic range, there will be no perceivable difference, because both cameras are extremely capable.

The difference is that the autofocus on the new A7RV is extremely capable, while the autofocus on the A7II is rudimentary and unable to deal with things that move. Also, the viewfinder on the A7II is adequate, while the viewfinder on the A7RV is excellent, which contributes nothing to the image quality, but should reduce my eye strain significantly, which matters to me since my eyes are not what they used to be. Also, the fact that A7RV has 60 MP resolution means that it has 26MP of resolution within the APS-C circle, which means I can magnify the telephoto range by the factor of 1.5x and still retain the same print size that I have on the A7II, which is a much more tangible functional difference than the ability to print larger than a meter in width, which I almost never do. The ability to turn 400mm of range into 600mm is extremely useful.

Now for the drawbacks. The old camera is free since I already own it, while the new camera cost 3500 € used. This is a significant cost, since all of my lenses probably cost less than that; alternatively, I could get several GM grade lenses for that amount of money. This means that I needed to have very good reasons for the upgrade. Also, the new camera produces bigger files, and more of them because it’s faster, which means greater requirements on memory cards and storage drives, not to mention computer processing power. My computers and storage drives are already adequate, but I had to buy an extremely expensive cf-express memory card, which is a NVMe gen-3 1 lane drive. Yes, they now have memory cards that are NVMe drives, because apparently you need that in order to record video and clear the buffer quickly. In essence, all the drawbacks are a matter of money, while all the benefits are a matter of user comfort and the ability to actually get the kind images that I otherwise wouldn’t be able to get, for instance by turning a 400mm 35mm system into a 600mm APS-C system temporarily and tracking a bird in flight so accurately that its closer eye is continuously kept in focus.

Reasons

One could rightfully ask why the hell am I buying almost 6000 € of photographic equipment, on top of 2000 € of stuff I’ve already bought recently, if I expect serious disasters that will end the world as we know it.

One could ask with equal right why I’m mowing my lawn, or brushing my teeth, or servicing the car. It all assumes the kind of continuity I don’t, in fact, believe in. However, I don’t know the timing, which means I have to behave as if the things are going to outlive me, and on the other hand be ready to leave today if God calls. This means that I function in a way that is both detached, and involved. I’m performing all kinds of duties on a daily basis, and yet I’m ready to leave every single second.

The reason why I ordered the equipment is actually detached from any expectation to use it; I merely decided to pay respect to my photographic art and skill. It is more of a sacrificial offering than anything else, because in this world one needs to support things that he sees as valuable, because what you don’t support dies of neglect by default. So, it’s a matter of philosophical consistency, rather than some investment in the future or what not. No; rather, it’s a respect to what is and was. Biljana got new stuff for the same reason. It is important to pay respect to that which is good and valuable, the same way it’s important to keep uprooting the weeds.

Respect

I love the 16-35mm Zeiss. Since I bought it, I had a burst of creativity with it, making a deluge of wide-angle shots that look as if they were queued up somewhere for years. Its perfect image quality helps, since everything turns out as I envision it, as long as I keep the flare in check.

Which makes me think: why didn’t I get it before? I knew that I wanted it, since 2016 when I bought the Sony camera. It was kind of expensive, that’s true, and I already had the 17-40mm Canon which is very similar, and I used it with an adapter. But since Biljana used it so much with her Canon system, I very rarely shot anything with a wide-angle anymore, and it was not a good thing. I should have bought the Zeiss earlier, but I was putting all the money into gold, and I cut all the “unnecessary spending”, which included photography.

I think that was a mistake, however. To me, photography equipment is not just another gadget that essentially does nothing, like a fancy watch. It’s a creative instrument, something that allows me to produce and develop my photography. Similarly, a computer is not a gadget, it’s an extension of my mind. I was, however, smart enough not to skimp on computers – that’s something I use every day, and if there’s something wrong with how it works, I feel it. I think I put a pause on photographic spending quite a while ago, when I was broke and in debt, and I basically just used what I had, and this continued as things got better; I simply didn’t revisit the concept, and I had other things on my mind as well. But then, the reaction I had when I recently bought the 50mm f/1.8, and much more when I got the Zeiss, pretty much surprised me, because I didn’t expect that kind of a creative outburst connected with getting the equipment I needed, because I thought it would be like getting some gadget that does nothing and you get tired of it soon and don’t even notice that it’s there. No; this is not like getting a new car when you already had a decent car; it’s like getting a car when you didn’t have a car and you really needed it, but you kept arguing against it to yourself – cars are expensive, they are just another thing to worry about, you’re better off without it, the less things you have the better and so on, and then you finally get the car and the whole world of possibility opens up before you. Suddenly new places are in reach, and when you go to those places you find out whole new things that open up your mind to things you didn’t think of before. In this case, it’s things that were in my mind but I couldn’t create pictures from them because I didn’t have the adequate gear.

And then I decided: fuck it, I’m going to get the rest of the stuff I know I need, but I kept myself from getting because it’s expensive, and I was being rational with money. I’m getting the FE 100-400mm GM and the A7RV body. Something clicked – buying the gear you use for creative purposes isn’t excessive spending or buying gadgets you projected desires into, only for it to become empty and meaningless a week after you bought it. It’s more like respect being paid to important things in your life, and if you don’t, it’s going to die from neglect. It’s not just relationships with other people that need respect in order to flourish, it’s also parts of your own life – ability to write, create visual art, and so on. Also, there’s a difference between stupid ideas like “if I only had that lens/camera, I’d be taking better pictures”, and “if I only had a wide-angle lens, I could take the wide-angle composition that I have queued-up in my head”. The latter is actually a manifestation of a genuine creative impulse.

I’ve been thinking about the telephoto shots I took with borrowed equipment, and it’s not like I don’t want to take that kind of pictures. It’s just that I didn’t have the money for it, because telephoto photography is one of the most expensive technical parts of the craft, and even when I did have the money, I still blocked it off – nah, that’s a money pit, let’s just steer clear. At some point, this stopped being financially responsible and realistic thinking, and became a sign of disrespect to myself.

Diversity

I’ve been going through my library of old photos and thinking.

Before 2006, I’ve been using standard zoom lenses by default, and when you ask people why those lenses are good, they will tell you it’s because they are universal, and allow you to take all kinds of pictures – from landscapes to portraits and details and so on. However, when I look at my photos taken with Olympus E-1 and the ZD 14-54mm standard zoom, over 90% of them conform to the pattern of “extend to 54mm, aperture wide open”. I was not a “diverse” photographer at all, and in fact I could have used the ZD 50mm f/2 macro instead of the standard zoom, without any adverse effects. Even then, I was very specialised for isolation-based closeups, and from what I can tell, I produced very “mature” work in that area. I knew what I was doing and the results turned out the way I wanted them:

However, there was a reason why I used the lens almost exclusively at 54mm: I didn’t know anything about shooting landscapes, or wide angle anything in general. It’s not that I didn’t try, but the results were crap, in a sense that I couldn’t control the scene in such a way that would capture the feeling of calm stillness that I learned to capture with closeups. When I think of it, I tried to follow a formulaic approach for shooting landscapes, and the images sucked. Also, when I would use wide angle, the scene felt cluttered and full of distractions that created something that was the exact opposite of what I did with closeups. Also, the 5MP camera lacked the level of detail that would be required for a wide angle landscape shot in which everything is supposed to be sharp.

And then I decided I’m going to learn landscape. It certainly was a learning curve; my early attempts were crap, until the point I was reviewing the Olympus ultra-wideangle, the ZD 7-14mm f/4, and at some point it clicked: I stopped trying to remove things from the scene in order to simplify it. I embraced the chaos in the scene and just arranged it into a flow. When I think about it now, it’s not that I learned to use wide angle; rather, I changed my attitude towards Chaos as a principle, by no longer trying to eliminate everything chaotic and thus create order, and instead felt the wild spin of the Chaos in a scene and freeze a moment that feels right.

It took me years to get comfortable with the concept of infinite depth of field, chaos, suggested motion, people in the frame, random things in the frame, non-obvious composition, and, sometimes, intentional motion blur. But, how else do you take a portrait with a fisheye lens in dense woods? 🙂

I must admit that the technique required me to pretty much abandon my usual style and methodology, and initially the equipment more-less dictated what I did; essentially, the camera took the pictures it wanted to take. It took me a while to first control the process, then get comfortable with it, and eventually extend my style through it. At some point, wide angle shots I took started looking as just my normal stuff, and that’s when I became happy with it. Even if it’s not nature, and if it’s black and white.

 

Kit lens

If there’s anything constant in photography circles over the decades, it’s the universal contempt for the “kit lenses”, basically the standard zooms that come with the camera. They are criticised for being plasticky, poorly made, having terrible variance between samples due to poor manufacturing, unsharp, having lots of vignetting, chromatic aberrations, being too slow, and so on.

The additional problem is that when you try to look for samples of photos made with such lenses, you are invariably showered by terrible snapshots made by people who are very unskilled at photography, and usually quite new to it, so they don’t know what they’re doing. The images produced are thus universally terrible. This creates some kind of confirmation bias – yeah, the lens is terrible, what do you expect from a kit lens, just get a proper one if you want your photos not to look like those beginner snapshots.

I must admit that I once believed something similar, and had an aversion to kit lenses, especially since my favourite kind of photography was to use shallow depth of field to isolate the subject, and kit lenses are generally quite slow – often f/5.6 at the long end – and they also often perform poorly wide open, at least that’s what I saw from the samples online. Sure, I used a Minolta MD 35-70mm f/3.5 lens when shooting film, and I thought it was a very good lens, but this one seemed to be quite highly esteemed online, since Minolta produced lenses for Leica R system, this one being also produced as Leica Vario Elmar R 35-70mm f/3.5. The fact that it behaved well was thus unsurprising – it is a Leica design executed by Minolta, after all.

The second “kit lens” I had direct experience with was the Zuiko Digital 14-54mm f/2.8-3.5, and it was excellent. If anything, it was even better than the Minolta; quite a stellar piece of optical design, with only a few minor drawbacks, such as the onion-circle bokeh and some green chromatic aberration on contrasty areas when shot into the light. Since this lens was only a “kit lens” because it was sold in a kit with Olympus E-1 professional camera, and was otherwise known as a optically superb piece of gear, this was also not unexpected, and my experience with “kit lenses” remained great.

I had no reason to doubt my understanding of kit lenses until I got to test the Olympus E-500 camera with its ZD 14-45mm f/3.5-5.6 kit lens. While was not as good for closeups as my ZD 14-54mm, it was otherwise almost identical. The colors were clear, sharpness was excellent, and other than the cost cutting being visible from the cheaper materials, optically speaking this lens was great, and especially great for the money.

I was pretty shocked, but shrugged my findings off – it was Olympus, after all; they are known for making excellent lenses, and even their cheap plasticky kit lens must be superior to those of other manufacturers. After all, they don’t produce them by the boatload like Canon for their entry-level cameras.

By that point, however, a conclusion was starting to form, because I made quite a bit of poor images with good lenses when I didn’t know what I was doing, and I also made quite a bit of good images with stuff that would be deemed “entry-level”, but the true shock happened when I bought a Canon 5d, the camera with the best and sharpest sensor at the time, and just for shits and giggles I bought a used EF 35-70mm f/3.5-4.5 lens for 40 EUR or so. It was very old (made in 1987 as a kit zoom for the EOS 650, the first one Canon made for the lineup) and very heavily used, also obviously very plasticky. I quite rightfully expected it to perform terribly, especially on a high-resolving sensor of the 5d. I took it one day for a walk to take pictures of my kid.  I quite expected the results to be fuzzy, resolving less detail than my 5MP Olympus kit, because, after all, the Olympus lens was very sharp; I even called it “the shit lens” before even trying it out.

What I was shocked to see is one of the most highly detailed, crisp-sharp portraits I have ever seen, anywhere, by anyone with any kind of equipment, period. (note: the illustrations here are not representative of the actual raw files due to compression and reduced file sizes)

This is not shot from a tripod, stopped down, or in any other way cooked up to make the lens perform well. I basically opened it up, extended it to 70mm, put the ISO at 400 because the light was dim, focused on the eye and took several pictures. The shocking part happened at home, when I opened the files in the raw converter and saw the detail on the zipper, eyelashes and the fabric of the cap. I quite literally never saw anything this sharp, and I already tested the EF 85mm f/1.8. Far from being a plasticky thingy that would do a disservice to the mighty 5d, I could actually bet that it outresolved the sensor by quite a margin.

Still in shock, I went online to see if I can find any reviews of this lens, and of course, it was universally poorly regarded – low resolution, CA, vignetting, all kinds of blah. It was actually so poorly regarded that very few deigned it worthy of even reviewing it.

And so, I went on to use this “shit lens”, as I continued to call it, in part due to its plasticky design, but now mostly in mockery of the online photographic community, and took some of the sharpest, most colorful and atmospheric images I ever made. It was my favourite landscape lens on the 5d, because of both its range and the incredible color and details it resolved. Sure, it has very nasty flare and bokeh is quite harsh at portrait distances, so I had to keep it within the range of parameters it liked, but when I did, it was absolutely stellar.

There was a conclusion that brewed in my mind, and at some point it came out: “hey wait, those people online are actually completely full of shit”. The fact is, you can make any lens look bad. All lenses can take shitty pictures. However, it’s not actually hard to get even the cheapest lenses to take great pictures. Also, the myths about “sample variance” are also likely caused by very inexpert use.

At some point I decided to test my hypothesis. I went online to check which camera-lens combo was the least well regarded and considered unworthy of even an amateur who wants his pictures to look decent. The consensus seemed to be the Olympus Pen E-PL1 with its kit lens, the 14-42mm f/3.5-5.6. The camera had “high noise” and the lens had the plastic mount, collapsible ultra-cheap plasticky design, poor aperture and range, and low resolution, according to the tests. So of course I proceeded to take some of the sharpest 12MP nature shots with it, rivalling the 5d.

To add insult to injury, that lens also had excellent closeup capability, which made it incredibly versatile for almost-macro shots:What about the high noise on the sensor? Well, yes, the sensor is very noisy at ISO 1600 and above, and I don’t recommend using it at those settings. However, there’s something else that I noticed in the dpreview lab test: the on-sensor CMOS noise reduction is very conservative, meaning that it doesn’t aggressively de-noise the RAW files; essentially, it doesn’t remove chroma noise by desaturating the files and killing all the color. As a result, the colors this sensor produces are deep, “meaty” and quite suitable for my kind of nature photography. The sensor is also very sharp and detailed. As a result, something that looks like a toy and is routinely dismissed by the “discerning”, “advanced” users is actually a great camera.

“I can’t see anything on this display, this is terrible.”

Sure, it has flaws – no viewfinder, which means you can’t see shit in bright light, which is when I used it the most. Slow electronics. Display with very poor outdoor visibility. Terrible ergonomics worthy of Sony. All in all, the stuff that doesn’t contribute to image quality, but does contribute to your ability to actually use the camera to take pictures. So, I retired it, and, having proved my point about kit lenses and entry level cameras, bought Sony A7II full frame mirrorless camera with the FE 28-70mm f/3.5-5.6 kit lens, which, you guessed it, is universally poorly regarded online, and proceeded to use it to shoot the most detailed, colorful landscape shots I have ever made.

The lens itself, of course, is excellent. Colors are excellent, contrast is excellent, sharpness is excellent, and any defects are minor. Is it a perfect lens? Of course not. It’s quite slow, and has poor close focusing distance, making it quite poorly suited for closeups and portraits. It is, however, a stellar performer for landscape photography and for portraits in context. But if you’re ever seen with one in public, your reputation as a photographer will never recover. 🙂

As a conclusion, people are full of shit.