The Full Frame Debate

(commentary)

At least amongst the serious enthusiast market, there seems to be this growing groundswell of "I'll be shooting full frame some day." This chorus grows with each new rumor or announcement of a full frame camera. 

At this point, Nikon has five full frame cameras (D4, D3x, D800, D600, Df). An absolutely terrible lineup if you ask me, as it pretty much dictates that you need to stay with the same body as a backup as a pro due to the various accessory issues involved*. Meanwhile, Sony has four (RX1, A99, A7, and A7r). Also a terrible lineup, as you're not going to upgrade from one of those to another easily, except maybe the A7 to A7r, but then why didn't you buy the A7r in the first place? Canon has four: 6D, 1Dc, 1Dx, and 5DIII; another slightly strange lineup with a 4K video camera slipping in there. 

*Note that "lineup" is not "product." LIneup refers to the options a user has to choose from, and while Nikon's FX lineup has lots of options, they feel extremely disconnected from one another now, with no real connection between the products. Compare D800, D4, D3x to the old D700, D3s, D3x line. Today: no connection; previously: solid connections. So, the "products" can be excellent, but the lineup they create can be weak. That's my point.  

If you look around, you'll find rumors of Fujifilm, Pentax, and even Olympus eventually having a full frame option or two. 

One has to ask: what's the big deal about full frame? Why are all the makers taking scatter shots at the full frame market and why are all the enthusiasts so fast to endorse this strategy? A cynic might say: because they're hoping you'll bite into a more expensive treat. If your unit sales are dropping 15% but you can get 20% of your top end DX buyers to purchase a much more expensive camera, your actual sales numbers might be okay.

As a reminder, the most expensive part in a DSLR is the sensor. While full frame sensors have come down in price, every time they do, a crop sensor should come down in price by as much as a factor of eight. I've written before about sensor costs. Back at the start of full frame, it was pretty clear that a 36x24mm sensor was topping US$500. Last time I looked (around the time of the D600 intro), the best guess I could get out of my semiconductor friends was US$350. Today, I think the number may have stretched down to US$300. So large sensor prices have been coming down. 

Still, they don't come down as fast as some of the Web fora are suggesting they will. Moreover, even if you can get a perfect yield (no defects on the wafer, no defects in building the circuitry on the wafer, no defects in cutting the wafer into chips and casing them), the full frame price will probably never drop below 3x that of an APS/DX sensor. Why? Because full frame uses more silicon and it wastes more of the wafer, and the time on the fabricator is longer and more complex. 

So producing a full frame camera with the same body and components as a crop sensor one (e.g. D610 versus D7100) will always result in the crop sensor product being less expensive. 

Why the heck are the camera makers trying to get you to buy more expensive cameras? As I noted earlier, it's because unit sales are going down. To keep their overall sales and profit numbers up, they need to get you to buy upscale from where you were. 

"That's okay," you say, "because full frame is better." Technically, yes that's true in some measurements, but are you sure it's actually better in practical application for you? There are two "bars" in image quality, I think, the toe and the shoulder:

quality.jpg

Below the toe at point A we have cameras that a serious shooter probably wouldn't consider because they don't meet the "good enough" factor. At the other end of the quality curve we have cameras that extend beyond point B, which is the point where "difficult to tell a difference" comes into play. 

I would argue that for most people, full frame as opposed to crop goes beyond point B. For example, which one of these ISO 3200 shots is 24mp DX and which 24mp FX?

dxvsfx.jpg

Okay, I'm cheating a bit because you're looking at a reduced shot in this Web page, but I can tell you that looking at these pieces of 24mp images full size on my big monitor, I prefer the one on the left ever so slightly. It has more visible noise, but it's very small and granular and not at all obtrusive. The image on the right has slightly clumpier noise, still small and unobtrusive, but just enough different that I like it less. 

So which was which? The left hand photo is a D7100, the right-hand photo a D600. Basically the same camera, but one with DX (D7100) and one with FX (D600) sensors. 

It's even possible that my liking the D7100 better here is because I have a really good sample of the D7100 and a really bad sample of the D600. In other words, as you get to point B shoulder on the image quality curve, even sample variation may be enough to tip the nod to one product over another. 

But wait a minute, you're saying, doesn't FX have a one-stop advantage over DX? Maybe, but can you actually see a one stop advantage? That's a question you should really ask. For example, DxOmark ranks the D610 and D7100 this way:

  • D610: 25.1 bits color depth, 14.4 EV dynamic range, 2925 ISO
  • D7100: 24.2 bits color depth, 13.7 EV dynamic range, 1256 ISO


I'd argue that 99% of you reading this can't see the differences in color depth, 95%+ can't see the differences in dynamic range, and the only way most of you will see the difference in ISO handling is if we push to an extreme (I'm guess we'd have to get out to ISO 6400 before most of you would clearly see a difference after my post processing). 

Now, there is a difference in DOF capability. All else equal again, you can get a one-stop shallower depth of field out of the FX sensor for the same shot if you've got the right lenses (more likely on FX than DX, unfortunately). Most of you could see that difference, I suspect. But how many of you actually need or use it? ;~)

Let's say for a moment that we want to take the exact same shot (the basketball example, above). I mean exact. Same DOF. Let's see:

  • FX — f/4, 1/1000, ISO 6400
  • DX — f/2.8, 1/1000, ISO 3200


This is where things get tricky. Really tricky. The minute I want more DOF in certain situations, the DX starts to equalize very quickly with the FX camera. That's because to keep the exposure the same for the basketball shot I have to up the ISO on the FX camera because I also have to use a smaller physical aperture. The amount of light available isn't different, after all. 

Technically, the FX camera is going to be better but some small margin. It tests better in the deep pixel quality tests, it can shoot into light about a stop lower in light than the DX camera, and if you have a fast enough aperture you can isolate the focus plane more dramatically than DX. But if you're going for the same exact shot with both cameras, sometimes the FX and DX cameras are literally about the same! You boosted the aperture and ISO on the FX camera to match the depth of field on the DX camera while keeping the ball still frozen. 

Plus, if you're coming from any body more than about three or four years old, a new DX camera is going to be significantly better than where you were, too, so don't judge DX on what you've been using. It's really easy to see that in comparing D300 versus D7100 images, for example. Sensors came a long way between 2007 and 2013. More and better pixels, with gains pretty much across the board. In fact:

  • D7100: 24.2 bits color depth, 13.7 EV dynamic range, 1256 ISO
  • D300s: 22.5 bits color depth, 12.2 EV dynamic range, 767 ISO

Plus you have double the pixels now ;~). 

Here's the thing: for most people, Point A on their quality bar ("good enough") is probably low enough that we now have smartphones that pass it. Then we have a lot of cameras crowding Point B on their quality bar. I would guess that if you let me do the shooting and processing, a whole slew of cameras would look like they're performing at or near your "difficult to tell the difference" mark. Certainly all the Sony NEX models, all the Fujifilm mirrorless models, the Canon EOS M, the forgotten Pentax K-01, virtually all the crop sensor DSLRs, and, of course, the FX DSLRs are going to be crowding point B or be beyond it. The most recent m4/3 mirrorless bodies (E-P5, E-M1, GX-7) would all be bunched up there, too. 

To me, this is an exciting time in photography. The Olympus OM-D E-M1, for instance, is above my "quality needs" bar (which is lower than my "difficult to tell the difference" bar, but not by a lot). It's also small, light, versatile, and a joy to take on long hikes compared to my D4, which also happens to be 16mp. If I need 36mp in the backcountry, I'll just stitch a couple of shots together ;~). 

So we're about to circle around to where I started: do you really need a full frame camera? My answer would be "probably not." The camera makers want you to buy one probably because their gross product margins are about the same no matter which DSLR you buy, so they want you to buy the most expensive one you can afford. Somewhere along the way, people got it into their heads that full frame was so much better than anything else, that any time a company mentions that they're coming out with another full frame camera, especially an "affordable" one, the entire Internet gushes with lust over the newcomer. 

If you remember all my "want" versus "need" articles over the years, basically what's happening right now is that camera makers are preying on your "wants" and not your "needs." Put another way, if everyone wasn't so lusting over full frame bodies Nikon would have had to have come out with the crop sensor D400 and some more DX lenses by now ;~). 

Don't get me wrong. I shoot right now mostly with FX and m4/3. For my critical work, I want the very best I can squeak out of pixels, and with a truly portable, versatile body, that means FX. I've had offers from a couple of companies to explore Medium Format, but that pushes me up into cameras not quite suited to all the things I need to do photographically. Thus, FX is what I chose for my professional work and what I pick up when I need absolute maximum quality. 

But if were a more casual shooter having to pick one system that was more than good enough and difficult to tell the differences on, m4/3 through APS/DX would probably be where I was at. As it is, for days when I'm putting on the miles I've picked m4/3 as my format to keep size and weight down. So please don't call me a hypocrite for using FX for pro work: I'm living what I preach for some of my work, and I suspect that as time goes on and more cameras get up to Point B or further on my image quality chart, my absolute need for FX will go away. 

As I've written many times in the past 14 years (mostly in the past 5): Canon, Nikon, and Sony would make this a heck of a lot easier on you folk trying to decide between crop sensor and full frame if they'd just fill out their lens lineups and take that part out of the equation. After all, if you need a 24mm equivalent tilt-shift lens, you have zero choices except for full frame. That's one of the reasons why I was so early to applaud Olympus and Panasonic with m4/3: more so than for any other format other than full frame, they've delivered the lenses. If the EM-1 steals some Nikon DX customers, Nikon only has themselves to blame. Good thing Nikon has that line of five very different FX cameras then ;~).  

© Thom Hogan 2013 / All Rights Reserved bythom.com  @bythom #bythom