Do YOU need all those megapixels?

A2_194595

Recently there have been quite a few articles appearing online and in print, talking about whether (or not) you need the latest super-high-megapixel cameras. First there were the 36 Mp cameras, while today there are also 42 Mp and 50 Mp cameras. I’m not going to go into the details of the medium-format options, which today go up to 100 Mp in a single frame. But similar issues apply to all of them. When the Nikon D800 was announced, 36 Mp seemed huge. Today not so much (but still very respectable!) but the newer cameras keep pushing the boundaries.

Example “35mm full-frame” cameras:

36 Mp: Nikon D800/D800E/D810, Sony A7r, Pentax K-1.
42 Mp: Sony A7r II, Sony RX1R II.
50 Mp: Canon EOS 5Ds(R).

Incidentally, the difference between 42 and 50 megapixels is sometimes misleading when it comes to the number of pixels across the image. In terms of pixels these cameras have:

D810: 7360 x 4912
A7r II: 7952 x 5304
5Ds(R): 8688 x 5792

When you read any of these articles you of course have to wonder how much “buyer’s bias” is involved. Sometimes it seems the author is trying to convince themselves that they’ve made the right purchase in buying (or in not buying) that new camera. And of course the same question can apply to this article.

So to be clear: I’ve used Canon DSLRs of 20-22 Mp in my work since the end of 2008 (having started with lower-resolution EOS cameras in 2001) and am very comfortable with that being a useful level of detail. When the EOS 5Ds and 5Ds R came out in mid 2015 I looked at them with interest but wasn’t sure I could justify changing up from the 5D Mark III for several reasons. Finally in late 2015 I decided that yes I could, and got myself a 5Ds R before heading to Antarctica in January 2016 to lead a photography workshop. On my previous Antarctic trips I had used 5D Mark II cameras as well as a 40D and a 7D, so I knew what to expect. Yes I am very happy with my 5Ds R (with over 7000 images made with it so far) but while it’s working for me I’m not here to tell everyone they need to upgrade.

Incidentally, I did look at the Sony A7r II carefully (including using one) when it appeared, but so far for me this style of camera simply does not work for my photography. It has a few interesting technical advances, but overall the package doesn’t suit me.

By the way, you can click on any of the images in this post to see larger versions.

How many pixels do you need?

This is an obvious question that is easy to start with. If you’re only ever displaying your images on HD TV screens then your final result will be only 2 Mp. 4K UHD is around 8 Mp.
If you make 8″x12″ prints then at 360 ppi this only needs 12.4 Mp. At 300 ppi only 8.6 Mp, and at 200 ppi (a low-end rule of thumb as to how low you can go before people might start noticing issues) that’s only 3.8 Mp.
If you step up a level and make 12″x18″ prints (e.g. on A3+ paper) then at 200 ppi this is 8.6 Mp, at 300 it’s 19.8 Mp, and at 360 ppi it’s 28 Mp. A reasonable 22 Mp image (e.g. from an EOS 5D Mark III) should have no problem producing a killer print at 12″x18″ (or even larger such as 24″x36″).

In episode 944 (“Practical Pixels”) of his LensWork podcast Brooks Jensen opined that for work producing images for books, 16 Mp is a reasonable number. I concur that for most purposes the 16-24 Mp range is very useful for most people.

The Milky Way points down to an auroral glow over Bass Strait. Equatorial mount allowed a 2-minute exposure.

Milky Way and Aurora
EOS 5D Mark III, Samyang 14mm/2.8
20 Mp (slight crop cleanup)

The 5Ds R has a 28 Mp mRAW mode, but I find myself sticking with the full RAW format. I never know when I’ll want or need the extra detail, and it’s not worth losing those opportunities. That would be like someone shooting in JPEG “except for those shots where I need the extra features of RAW”. It’s usually too hard to know beforehand!

Do you need to crop?

Of course this all assumes that you don’t crop your images significantly. However I find I often do crop. Sometimes to fix a wonky horizon (I often have the excuse of being on a rocking boat). Sometimes to use a better aspect ratio that suits the subject. I regularly produce horizontal and vertical work in ratios ranging from 1:1 (square) through 5:4, 4:3, 3:2, 16:9, 2:1, 21:9, 3:1, etc and this usually involves intentionally throwing away pixels from the original image. I do also stitch images for some of the panoramas, but that’s not really significant in this discussion.

Near Petermann Island, these seals were just taking it easy.

Crabeater Seals
EOS 5Ds R, Sigma 120-300mm/2.8 DG OS HSM S @ 120mm
32 Mp (full-width “Cinemascope” 21:9 crop from original)

Sometimes I crop because I find an effective composition that is only a small part of the original scene. In an ideal world I would notice all these options before taking the photo and then use the optimum lens to minimise the cropping required, but especially in a fast-paced world of wildlife photography as well as taking landscapes from moving vehicles (e.g. ships) this doesn’t always work. I do often use zoom lenses for this reason, but the further I can crop in to an image and still end up with useful image quality, the happier I know I’ll be when processing the images.

What will it cost?

But with all those extra pixels there’s a cost. Sometimes (always?) there’s a monetary cost, but there are also flow-on effects that you have to be prepared for.

A2_197992

Blue sky surprise
EOS 5Ds R, EF 24-70mm/4 L IS USM @ 39mm
50 Mp (no crop)

Can you get sharp pictures?

Obviously with the ability to capture finer detail the question comes up as to whether your camera technique and your lenses are up to the job. Yes these cameras are pushing the envelope: to get the best value from them you’ll want to have everything optimised. But that isn’t just something for the high-megapixel cameras: improving your technique will help with all cameras.

After shooting with 5D Mark II and Mark III cameras for several years, in 2014 I added a 20 Mp EOS 7D Mark II to my kit. Due to the smaller pixels (and the magnifying side-effects of the 1.6x crop) I had to refine my shooting techniques (including choosing faster shutter speeds and in some cases changing aperture choices) to get the image quality I was used to. Simply put, I had become lazy even though I had used an 18 Mp EOS 7D in the past. Over time I also upgraded several of my older lenses.

When I considered getting the 5Ds R I knew I was going to face the same issues (the pixels on the 5DsR are almost the same size as those of the 7D Mark II, although there are a lot more of them). As I had already got used to the 7D Mark II, I was confident this wasn’t going to be an issue, and in the end it wasn’t. I’d made the relevant tweaks to my equipment and technique over a year before.

Pauline's serenade

Antarctic sunset
EOS 5Ds R, EF 24-70mm/4 L IS USM @ 24mm
85 Mp (3 frames stitched)

Incidentally even though the pixel density on the 5Ds R and 7D Mark II are similar, if I crop a 5Ds R image down by 1.6 to the 20 Mp level, then as long as I’ve exposed it well the 5Ds R image is usually coming out sharper. As expected the cancelled-out anti-aliasing filter is making a big difference. I doubt the non-R “5Ds” would have the same advantage, and I have not yet noticed any moiré in my images.

Yes my AF is calibrated

Because the AF in a DSLR uses an AF sensor independent of the main imaging sensor, with the fine detail these cameras can capture I’ve found it’s essential to calibrate the “AF Microadjustment” (in Canon’s terms) for each camera with each lens, even for the wide-angle lenses. I do this for my own gear and for clients using a combination of Reikan’s FoCal and Michael Tapes’ LensAlign/FocusTune products, including for internal calibration of some Sigma lenses. This isn’t because of any particular fault in the cameras or lenses, but more that the required tolerances are so fine that all the combinations of equipment can’t be perfectly aligned out of the factory(s). Note that the mirrorless Sony cameras use a different AF architecture.

A2_203079

Leopard Seal
EOS 5Ds R, Sigma 120-300mm/2.8 DG OS HSM S @ 300mm
22 Mp crop

Do you need long lenses?

The theory goes something along the lines of: if you can zoom in by simply cropping, do you really need a super-long lens? I hadn’t really paid this idea much attention as I already had a set of lenses I’m very happy with, but in retrospect my experience with teleconverters does play into this.

When the full-frame EOS 5D Mark III was my main camera, I decided I was happy with the quality of the latest 1.4x and 2x teleconverters when combined with my 120-300mm/2.8 lens. These make it a 168-420mm/4 and 240-600mm/5.6 lens. But when I also started using the EOS 7D Mark II, I decided that with the smaller pixels I was no longer happy with the quality I got with the 2x (this was with both the Canon EF 2x III and the Sigma TC-2001 teleconverters), although the 1.4x was still quite acceptable. I did do very careful testing with long lens supports, etc to minimise/eliminate any technique errors. The 120-300mm/2.8 lens with 1.4x teleconverter and the 7D Mark II becomes a 270-670mm-equivalent (“mm-e”) f/4 lens anyway, which is quite respectable.

However although I had the 1.4x teleconverter with me in Antarctica, and used the lens on both a 7D Mark II and a 5Ds R body (having previously calibrated the AF on all those combinations) I found myself using the 1.4x on only one day. When working in a Zodiac I tend to use all zoom positions (admittedly with a lot of use at the “long” end) and being able to zoom out to 120mm instantly was great, while I’ve also been able to crop in to some of the 300mm shots by significant amounts and still end up with killer images. Without the 1.4x the lens also gave me a lot of flexibility in aperture, which was very useful in the challenging low-light situations where losing one stop of light would be noticeable. I will keep and use the 1.4x in the future, but the ability to zoom by cropping is greatly appreciated!

A2_198449

Antarctic Tern
EOS 5Ds R, Sigma 120-300mm/2.8 DG OS HSM S @ 300mm
4.6 Mp crop (just to see what would happen!)

Do you need high ISO?

This is an interesting question. While the EOS 5D Mark III has cleaner images (lower noise) at high ISOs than the 5Ds (and the Mark III has higher ISO choices available although some of those usually aren’t worth using) this is offset somewhat by the fact that if you’re going to use the images at the same size then the noise in the 5Ds images will be finer and not as intrusive.

So far I haven’t found the 5Ds R’s theoretically lower ISO range to be a major issue for my work.

A2_198041

Paradise Harbour
EOS 5Ds R, Sigma 120-300mm/2.8 DG OS HSM S @ 171mm
292 Mp (stitch of 16 vertical images) @ ISO 1600

A2_198041

1:1 crop (796 x 597 px) @ ISO 1600

Can you manage all those gigabytes?

Obviously, with more pixels there are more bytes. Some of these cameras produce huge files (up to around the 80 MB mark per image). This means that the size of your memory cards may have to increase, and you’ll probably want to make sure the cards you choose provide the best write speed in your camera (see resources such as the Camera Memory Speed website) in order to reduce frustration when taking photos.

While some people have been afraid of having 32 GB or 64 GB of files on a single memory card “in case something happens to it”, if those sizes are re-framed in terms of holding 400 or 800 images per card then that can change the perception of size and risk. An 8 GB card that can only hold 100 images doesn’t seem as large as it used to. Mind you I usually try to download and backup my files regularly, and in general only need the cards in the cameras to hold a day’s shooting at most. In busy environments such as African safaris and polar expeditions I often download several times per day.

At the same time you may want to invest in a faster card reader in order to reduce frustration during downloads. I had previously optimised my travel system with multiple SSDs and fast USB3 readers, and when downloading from two cards concurrently and backing up to a second drive I usually get well over 130 MB/s sustained downloads (depending on the particular card models of course). So again I was already set up, and on expeditions the download and backup process is quite fast.

Storage/backup is another obvious concern: 80 MB files are a lot bigger than 25 MB files. Luckily 1 TB portable drives are common today and that should offer storage for more than 12,000 files. One issue I found on my 2016 Antarctic trip was that Lightroom’s Preview database was much larger than on previous trips, and threatened to fill up my laptop’s internal SSD (having 1:1 previews available is a huge speed-up, but obviously there are side-effects). Similarly the storage requirements for images and for the Lightroom database on my desktop server are growing, but that was happening anyway and doesn’t seem like a quantum leap.

Then there’s processing power. Your laptop or desktop machine will have to process more data, so it will take longer. Luckily the speed of recent CPUs is quite decent, and the biggest impact is not in terms of processor speed but rather in terms of RAM. My desktop machine which has 32 GB of RAM has hardly skipped a beat, but my mobile MacBook Air (which I wrote about last March) has only 8 GB has definitely noticed the difference when generating Lightroom previews.

A2_203695

An iceberg shows its colours
EOS 5Ds R, EF 24-70mm/4 L IS USM @ 35mm
31 Mp crop

What will the next camera have?

In the Canon range the 22 Mp EOS 5D Mark III is still a great camera, and presumably a 5D Mark IV (which might get 24 or 28 Mp) will be even better. It’s not just about the pixels, presumably there will also be developments in terms of focus, handling, etc. I’m not yet hesitating to use the “smaller” cameras for general shooting, but for many uses I’m very happy with the 5Ds R. It has opened up new options for me: both in terms of cropping flexibility and in terms of achieving large prints.

On expeditions I often work with a pair of camera bodies with different lenses mounted, and having a second 5Ds R would certainly be very convenient. But given the frequency of use and the cost of the camera, so far I’m still comfortable with having a second body with 20-ish Mp. Even if my 5Ds R was to fail catastrophically on an expedition, I’m comfortable with the backup camera being lower resolution.

A2_204207

Iceberg sunset in the South Shetland Islands
EOS 5Ds R, EF 70-200mm/4 L IS USM @ 173mm
32 Mp (full-width “Cinemascope” crop from original)

So what should you do?

Whatever makes you happy of course!

If you feel your photographic skills are up to it then by all means go for the high-megapixel bodies, but be prepared to upgrade/optimise all of your photography workflows. But for most people I think the 16-24 Mp range is still plenty, and gives you a tool that can allow you to develop your skills and achieve wonderful images without having to make the extra investments.

More than the camera body, there are at least two things that can be more important in terms of improving your images. Your own skills (which is where practice and education comes in: always things worth investing in), and camera lenses: choose good ones and they should last you a lot longer than that fancy camera body will!

Leave a Reply