Download MP3

For the past couple of years, I’ve posted articles talking about my impressions about most higher end camera. This time I’ll be doing this both as a text post and as a podcast. These notes are not exactly a summary of the podcast, but they should be close. I’m actually writing this independently of recording, not transcribing my recording.

I don’t get early hardware from any manufacturer, so this is mostly my impressions based on past experiences and published specifications.

Video

I’m going to start this discussion by talking a bit about the video capabilities of the 5D mark IV. If some of the discussions I’ve seen on line is any indication this is a point that’s somewhat contentious. That said, video is probably the least interesting aspect of this camera to me.

On the 5D mark 4, Canon elected to up the recording resolution to 4K, true DCI 4K, not ultra-HD that everyone else seems to be using. DCI 4K video can be recorded at up to 30 FPS. Stepping down form 4K, 1080p (FullHD) video can be recorded at up to 60 FPS, and 720p (HD) video at up to 120 FPS.

The contention part is the codec that’s being used for the 4K video. Both 1080p and 720p are recorded in h.264. However, 4K is stored using MJPEG (motion JPEG). The complaint I’ve seen seems to be that MJPEG is old and not as efficient and somehow bad for editing.

I’ll have to differ on the editing side of things. But MJPEG is an old style of coding that isn’t as efficient as H.264 in terms of space. On the other hand, it is an “ALL-I” format, as each frame is stored essentially as a JPEG without any dependance on any other frames.

The question in my mind is why Canon went this way.

That said, according to the documentation, 4K video will be 4:2:2 chroma subsample, while 1080p and 720p video will be 4:2:0 subsampled. Given the relative bit rates for both 4K MJPEG and 1080p h.264, I suspect the best quality video will be at 1080p.

As far as compression methods go, the 5D mark IV has 4, up from the 2 in the 5D mark III. One of these is MJPEG, which is used for 4K video and is not available for 1080p or 720p video. 1080p and 720p video are only compressible in h.264 and the options are:

• ALL-I – Only i-frames are created, frames have no temporal dependance on other frames, files are larger than IPB files, but better for editing especially when there are fast cuts.
• IPB (standard) – H.264 files are created with all frame types, i, p, and b, frames. These files are smaller than ALL-I files so there’s more recording time, but they can be more problematic when fast or accurate cuts are needed in editing.
• IPB (lite) – A lower bit rate version of the IPB setting. Produces even smaller files, so it’s good for long documentary type recording or when small files are needed, but the quality will be lower than the other options.

The one video aspect that interests me is high frame rate (HFR) shooting at up to 120 FPS in 720p mode. I’ll admit, if there was one feature that I found disappointing in the camera it’s this. I would really have liked to have had 120 FPS HFR video at 1080p.

At the same time, my comment to all the people bitching about 4K video performance in the 5D mark IV applies to my complaint about HFR video as well. If you want these features, Canon has a camera to sell you that will give them to you, the Cinema EOS C300 mark II. And to be honest, I’m betting that this is Canon’s position on this as well. And truthfully I don’t really fault Canon here. I started this discussion with video, but as far as I’m concerned as a still photographer, video is the least interesting aspect of the camera.

Auto Focus

While I don’t care so much about video, auto focus is one of the areas that really interests me. While Nikon went whole hog with loads of AF points on the D5 and D500 — most of which are just assist points to boot — Canon has gone for supporting narrower apertures. With the 1DX mark II, and now the 5D mark IV, this means all the AF points are sensitive to lens and teleconverter combinations at f/8 instead of just the center 5 points.

To me this is a big deal. While I fully appreciate the value of a fast long lens, and fast lenses in general. When you get to 300 mm and up, fast lenses are massively expensive. A 300 mm f/2.8L IS II USM is in the $6000 range, the 500mm f/4L IS II USM is $9,000, and so on.

In addition to being expensive, they’re also big and heavy too. While they are certainly great lenses, and deliver the best possible image quality, that comes with the trade off of their weight being a lot more to carry in the field. Plus as primes, they’re not as flexible as zooms.

Instead what I favor are Canon’s smaller lighter zooms, namely their EF 100–400mm f/4.5–5.6L IS II USM. It’s small, relatively light at around 3 pounds, and delivers pretty good image quality either bare or with Canon’s 1.4x III TC.

And the 1.4x TC is the catch. It pushes that lens’s min aperture to f/8, and on my 5D mark III, that limits me to only the center AF point.

For static subjects, only having the center AF point is not that big of a deal. Focusing and recomposing works. But for dynamic subjects, where tracking AF is needed, it significantly limits the ability to compose an image.

With the 1DX mark II, and now the 5D mark IV, having the full AF field available at f/8 this situation is dramatically alleviated.

There is a catch though. Canon’s always designed their AF systems to be conservative and deliver what they’re claiming. They don’t leave things up to chance, and when an issue with image quality could hamper the operation of the camera they take measures to insure that situation isn’t allowed. This was most obviously seen in the huge swath of available AF point modes in the 2012 AF system of the 1DX and 5D mark III, and it continues with the 1Dx mark II and 5D mark IV.

The effect of these limitations will of course depend on the equipment you own. For example, I still use Canon’s Extender EF 1.4x II, not the newer 3rd generation one. With the 1.4x II on my 100–400, because of the optical quality of the lens, Canon limits the AF to only 9 AF points, one of which being cross type, instead of the full 61 points with 21 being cross type. An improvement for sure, but still limited.

Admittedly I’m not complaining that much, a TC 1.4x III isn’t that much of an investment and it will improve my image quality as a whole anyway. Plus 9 points is better than 4 as it stands, especially since they reach out towards the edges of the frame.

GPS

GPS, or specifically geotagging, has been something of an on going complaint of me since I went to Alaska in 2015. I have a slew of images that I can’t readily place because I don’t have really good landmarks in them, and I didn’t have a GPS with me at the time.

I’ve been further struggling with the prospects of GPS and geotagging leading up to a trip to the American southwest I should be making the fall.

After having looked at a number of approaches to geotagging images, it is beyond clear to me that the best possible situation is having the GPS unit built into the camera.

On the up side, having the GPS in camera means that you don’t have to worry about syncing data or times after the fact. It also means you don’t have to worry about an external GPS unit — even if it connects to the camera.

On the other hand, if you have multiple cameras, it does mean you have to have the GPS running on all the cameras, not on just one. Plus it uses some of the camera’s battery and therefore places limits on the number of images that can be made.

Personally, I think the pros of seamless tagging and less equipment to deal with out weight the potential disadvantages.

As far as multiple cameras go, as long as your clocks are relatively in sync, you can always copy the GPS data from one cameras images or tracklog to the non-GPS equipped camera images.

Oh yea, the 5D mark VI can be configured to keep a tracklog not just tag images.

WiFi

Unlike GPS Wifi is a feature I’ve never really found myself really wanting, but I’m increasingly coming to the conclusion that it was my own perceptual blindness that made me not care. Well I shouldn’t say I didn’t care, it just wasn’t anything that I really felt was pressing.

In thinking about it, Wifi is actually quite useful, at least for me. I do a lot of product photography for this site, and for the vast majority of that, the process goes something like this: take a handful of pictures, pull the card from the camera and import them into Lightroom, examine them critically, return the card to the camera and take a couple more images.

Now I could shoot tethered, and in fact, tethered shooting exists largely for this reason. My problem with tethered shooting is the cord. With the way I work, I’m constantly afraid that I’m going to either trip on the USB cable, pull my laptop off the table, destroy the port on my computer by yanking the cord the wrong way, or worst of all, send my camera crashing to the floor.

Wifi completely eliminates the problem of having a wire to trip over or yank equipment off a table.

Between being able to get images from the camera to my computer, remote control the camera from a phone, and so forth, my apathy towards wireless may have been misplaced.

The one concern that I have that still lingers, is how secure all of this ends up being. At least on my EOS M3, it appears that the entirety of security is dependent on the encryption provided by the Wifi network that the camera is connected to. Moreover, I haven’t looked into what level of encryption is available when operating the camera in stand alone network mode.

I’ve said this in the past, security is hard, and there’s been few if any instances of a non-internet connected industry transitioning to internet connected products without serious security issues. At least, there is an option to completely disable the WiFi subsystem in the menus of the 5D mark IV. And honestly, this is probably where I’d recommend keeping it unless you’re activity using it.

Image Sensor

While it’s better late than never, Canon has finally moved to on chip ADCs. The 5D mark IV joins the 1DX mark II and 80D in this category. The move to on-sensor ADCs has been a long time coming, and directly impacts image quality and dynamic range.

While there aren’t yet, at least at the time of publishing this, any dynamic range tests of the 5D mark IV, I expect that it will have somewhere above 13 stops at ISO 100. At least that seems to be the most reasonable assumption at the current time.

The 1DX mark II sees 13.5 stops at ISO 100, and the sensor has larger photo sites. Bracketing it on the bottom the 80D sees 13.2 stops at ISO 100, and has much smaller photo sites. That leads me to believe that the most reasonable conclusion is that we should see somewhere in between those to values. At least I see no reason that dynamic range should be lower than the 80D.

In any event, the peak dynamic range should be much better than the barely 12 stops the 5D mark III gets.

One other interesting note is that the ISO range is largely unchanged from the 5D mark III. The 5D mark IV has a base range of 100–30,000, with an expanded range of 50 to 102,400. The only difference from the 5D mark III is the incrase in base ISO from 25,600 to 30,000.

I find the 1/3-stop increase to be interesting. In my experience, many photographers use the max ISO as a proxy for image quality. Cameras with higher ISOs tend to be perceived, rightly or wrongly, as having better sensors and better image quality than cameras that don’t have as much range. For the most part, historically this could be argued to be true. However, the Nikon D500 an D5 demonstrate that this should really be taken with a lot of skepticism especially when the numbers get preposterously big.

This change, however, isn’t preposterously big. So it may be that Canon did this as an indication that the 5D mark IV is better than the 5D mark III, or there may be some other motive.

With that said, I’ve been rather concerned that Nikon’s absurdly high ISO numbers on the D5 and D500 would precipitate a race to the bottom in the camera industry — you could think of it akin to the so called megapixel wars. That if Nikon can do ISO 3.2 million, then Canon or Sony needs to one up that and hit 6.4 million, and so on.

In practice, the results of Nikon’s absurd ISOs are pretty crap, and they’ve let the bottom fall out in tears of image quality to get there. Moreover, there’s no reason to believe the Canon or Sony could do much better at the same levels.

Fortunately, either because of either intelligent designers or the camera being too far along in development when the D5 and D500 were released, the 5D mark IV doesn’t go flying off the stupid high ISO deep end. ISO 102,400 is more than high enough for the vast majority of situations, and you can always dial in negative exposure compensation and boost the images in post if you absolutely need more shutter speed when shooting in the dark.

But at least for the moment, reason still seems to prevail in on Canon’s ISO front.

Resolution has also been a long running balancing act in my opinion too. More resolution is good for getting more detail and making bigger prints, but the cost is file size and therefore needing more storage.

At 22 MP the 5D mark III was a pretty good balance in 2012 in my opinion. Files were about 25MB, which not small wasn’t tremendously huge either.

At 30 MP the 5D mark IV pushes the bar up a little, but not so much that you end up even coming remotely close to the 5Ds’s 50 MP 50+ MB files.

8 MP is a 36% increase in resolution, which is both a big enough to be noticeable upgrade and not something that’s so big you end up having to throw out all your storage cards because they’re too small.

Real Time Lens Optimizations (JPEG files only)

Update: I’ve verified in Lightroom what I was starting to suspect. I was incorrect, the 5D mark IV doesn’t alter RAW files with diffraction or distortion correction settings. There is no change in quality in software like Lightroom that doesn’t implement Canon’s corrections.

Note: Given the lack of ability to test the results of in camera lens optimizations, I currently do not have a good understanding if these are applied in raw or just JPEG. Moreover, the text in the manual is not entirely clear — it indicates for example, that digital lens optimizer will reduce shooting speed, but why would it be even run if the RAW image data is not going to be modified?

The one down side to a higher resolution sensor is that it lowers the diffraction limited aperture. This is the point where diffraction becomes a limiting factor and beyond which the image quality will traditionally degrade.

One of the suit of tools available in Canon’s Digital Photo Professional software, at least in recent versions, has been diffraction correction. I took a brief look at its performance in a previous article. In my limited testing with a 5D mark III, it seemed like the desktop diffraction correction could drag an image shot at f/22 back to a similar level of quality as you had just after diffraction set in (~f/11) or even when the lens was wide open (f/4).

The trouble is, like a lot of photographers, I don’t use DPP. I use Lightroom. And Lightroom doesn’t offer diffraction correction — and probably won’t any time soon.

My point is, being able to do diffraction, chromatic aberration correction, and peripheral illuminate correction in the camera and just have corrected raw files to start post on is a boon.

Now the caveat to this is that Canon says that the diffraction correction applied in camera isn’t going to be as good as the diffraction correction applied in DPP on the computer. This point stands as one thing I’m very interested to test out. I suspect that some is better than nothing, but at the same time, I’m also curious to know whether it’s possible to fully correct the in camera corrected images in DPP after the fact.

Dual Pixel Sensor Architecture and Dual Pixel Raw

Arguably the biggest architectural change in the 5D mark IV’s sensor, aside from on-sensor ADCs, is the use of Canon’s dual pixel architecture. Though the dual pixel architecture was primarily develop as a mechanism to enable phase detection like AF without having to have dedicated phase detection pixels for video use. Canon seems to have decided to expand the reach of the technology into actual image quality aspects.

Part of this comes from the 5D mark IVs dual pixel raw format, and part comes from a few new processing options in DPP. Specifically Canon is touting dual pixel raw as a way to adjust bokeh and optimize image sharpness.

In some ways, dual pixel raw seems like the first actual real serious implementation of Lytro’s light field capture technology in a meaningful way for photographers. Instead of offering users the promise of being able to refocus the image in post processing — and carrying with that a massive file size versus resolution penalty — this is a much more limited approach.

I freely admit, when Lytro announced their camera I was intrigued by the technology, but very skeptical of the sales promise. Most amateur photographers have a hard enough time composing an image where there’s one interesting subject in focus and Lytro was proffering the idea that they should have to compose with lots of subjects that could be in focus.

On the other hand, as a serious photographer, focus and depth of field are something that I want deliberate control over as part of the creative design of my images. Lytro’s premise of being able to focus anywhere later, would, for me at least, turn into focus once and lock the image down.

Canon’s solution isn’t nearly as “flexible” but it appears far more reasonable. For starters, dual pixel raw files aren’t 10 times bigger than the resulting images; dpr files should be slightly less than twice the size of standard raw files (66MB versus 36MB per Canon’s estimates).

When shooting you’re still focusing and composing as you normally would. You’re still picking the depth of field as you normally would. And you’re not paying an astronomical penalty in file size for unlimited but ultimately unnecessary flexibility. And for that you get the ability not to shift focus around anywhere, but to make small but critical improvements in focus, sharpness, and resolution.

It’s too early for me to speculate how much of a draw this feature really would be, or how well it will work in practice.

What I don’t see anywhere in Canon’s discussion about the dpr format is on improving dynamic range. Magic Lantern, the supplementary 3rd party firmware mod for Canon’s camera, has had a feature called dual ISO raw, where they alternate between two ISO settings to expand the dynamic range of the resulting image. The technique works, but it does have limitations — specifically vertical resolution in the highlights and shadows is reduced by half.

Presumably Canon could implement a similar one-shot HDR-esque technique at the dual pixel level. This would, in theory at least, be able to resolve at basically the same resolution as sensor does normally, but still pull in the dynamic range increase associated with the two ISOs being able to reach further into the shadows and highlights.

Usability and UI

Most of the time I do these writeup I don’t spend a lot of time talking about the nuts and bolts of the cameras and a lot more time talking about usability changes and improvements. The thing with the 5D mark IV is that there aren’t a lot. The body is largely this same as the 5D mark III and 5Ds [R].

Three major external changes on the camera are that the cable release terminal has been moved from the side of the camera to the font, the side ports have been further sub divided, and there’s now an AF area selection button.

As a habitual user of Really Right Stuff L-plates, moving the cable release port is something that makes me happy as there’s no long any interference with the L-plate and the terminal to worry about.

The reorganization of the side ports isn’t too bad either. With the cable release port moved, you no longer expose the mic and PC ports to plug in a cable release. The new organization is in many ways better grouped too. The PC sync port has it’s own cover. The mic in and headphone out are both under the same cover, as presumably if you’re recording from an external mic, you’ll probably also be using headphones to monitor the recording. And the HDMI and USB3 ports are under the final cover.

Of all of that, my only real complaint would be the combined HDMI/USB3 port cover. When shooting video, at least for me, I frequently use an external (Small HD) monitor to get peaking and other focus assists, and simply to have the monitor at an angle that’s more conducive to my viewing. Using that leaves the USB port uncovered. Not a huge deal I guess.

Viewfinder UI

Canon’s intelligent viewfinder is one place where I’m not convinced the design is as intelligent as the name implies. Honestly, I would argue that the peak of the intelligent viewfinder was with the 1DX. The problem I see is what information Canon chooses to make available and how.

In the viewfinder there are two areas where information can be displayed. One is the yellow backlit LED panel along the bottom. The second is as a superimposed LCD image in the frame.

Being that the LED panel is illuminated and readily visible I would strongly argue that the most important setting information should be displayed there. To that extent, the shutter speed, aperture, ISO, buffer remaining, exposure compensation, AF lock, exposure lock, and flash sync mode are.

The secondary information is provided via overlays on the viewfinder itself. These include the AF points, a grid, the level, and now a whole slew of new icons. At least in the case of the 5D mark III, these icons are black against whatever background your image is — so if the image is dark, then you can’t see these at all. Now not yet having a had the opportunity to mess with newer intelligent viewfinder displays, these icons may be illuminated now, I’m not sure. The 5D mark IV’s added viewfinder overly icons include:

• battery remaining
• shooting/exposure mode
• white balance mode
• drive mode
• AF mode
• metering mode
• jpeg/raw
• digital lens optimizer
• dual pixel raw
• flick correction
• warning
• AF status

While I’m not going to poo-poo the value of any of these specifically, I’m not sure I really see how they warrant being in the field of view — if at the bottom edge. The grid, af points, and level are all things that useful when composing an image. A little icon indicating that the camera is in spot metering maybe not so much — double so given that the spot metering circle could simply not be drawn when the camera isn’t in spot metering mode.

It seems to me like Canon is including stuff in the intelligent viewfinder overall now just for the sake of including it. And it’s not like they freed up a considerable amount of status bar area to make the text for the exposure information bigger or anything like that.

At least the viewfinder doesn’t have redundant meter/exposure level indicators like the 1DX mark II does.

Okay, when I sat down to write this I had no intention of ending up at more than 4000 words. So I’m going to end this here. When I get the camera, I’ll be looking to write a review of it — which will probably end up being longer than the 25,000 word missive I wrote on the 5D mark III — and probably will end up posting some more detailed articles on various aspects. Heck, this was meant to be a podcast not a full on article.

Until next time.

Pre-order one on Amazon^{(Affiliate Link)} .