Canon 5D Mk4 Review – Conclusions

So, this has been a long time coming, but I like to be thorough you know.

The question everyone wanted answering was “is the Canon 5D Mk4 ‘better’ than the 5D Mk3 Andy”..?

The short answer is, ‘in my opinion’ a very affirmative YES.

But of course I’ve got to justify the ‘yes’ and that can be done by stating the important improvements – better image quality and autofocus.

Just the same as with its high performance cousin the 1DX Mk2, the Canon 5D Mk4 has had an impressive IQ boost brought about by one thing above all else – the SENSOR and its recorded Output.

The auto focus system has had the same overhaul found on the 1DX Mk2, and so there’s another big improvement.  Nope, the 5D Mk3 AF was NOT the same as that found on the original 1DX….

If you are a ‘tech slag’ then you’ll love the ‘touch screen’ menu, and the GPS too.

The touch screen drove me nuts when I first got hold of this camera – I hated it.  But I’ve gotten so used to it now that when I turned it off the other day I soon turned it back on – changing settings is tedious without it!

The SENSOR.

Nikon have had the lead over Canon for quite a while when it comes to RAW recording:

• Lower noise levels – especially at low to mid-range ISOs (100 to 6400)
• Better shadow recovery
• Option to shoot fully uncompressed 14bit RAW

Ok, so Canon (stupidly in my opinion) still refuse to allow you to shoot true uncompressed RAW, but on the other two counts they have at long last just about caught up with the boys from Minato.

For Canon users the sensor and its recorded RAW output on both the Canon 5D Mk4 (and 1DX Mk2) is something quite revolutionary; while as a Nikon user I’ve been used to it for ages!

What is it I’m talking about?  The benefit of having the ADC ‘on sensor’ or ‘on die’ to give it the correct terminology.

Canon have previously had their ADC circuitry buried deep in their DIGIC chips which are separate from the sensor, and so require wired connection.  This leads to two distinct problems:

1. Number of connections is physically limited.
2. Signal Entropy!

I’ll do a separate blog post covering sensor makeup shortly.

But now with the above two camera body sensors they’ve gone the Nikon way, using ADCs integrated within the sensors themselves.

It’s a well known fact that Nikon have used Sony sensors, or made sensors of Sony design by ‘special arrangement’, for ages.

In output terms, the Canon 5D Mk4 and Canon 1DX Mk2 sensors do bare such a spookily strong likeness to the Sony Dual Pixel Exmoor design – the coincidence is staggering!

What this means is basically:

• A lower noise floor.
• Greater potential shadow and highlight recovery over its predecessor.

Canon 5D Mk4 Dynamic Range

There are all sorts of reviews/claims plastered across the web that claim the Canon 5D Mk4 has a greater dynamic range than the class leader Nikon D810.  These claims, all by ‘third party idiots’ mind you, not Canon, are based on test results published by DXO Mark.

If only the idiots could read a graph!

According to the graph, at base ISO the Nikon D810 kills it by well over a stop, and doesn’t fall behind until base + 2.5 stops – 300 ISO indicated.

For landscape and other high definition/resolution photography you are going to be using your camera at base ISO to maximize DR, so basically the 5D Mk4 doesn’t even come close in this respect.

Having said all that, the way DXO Mark conduct their testing is somewhat circumspect in a lot of folks opinions – mine included.  Nonetheless, these results are being regularly misinterpreted and misquoted  everywhere!

When it comes to actual ‘tripod on the ground’ dynamic range you will always and without fail find that the ‘real’ DR is lower than the ‘oft quoted’ version – why?  Because the ‘testers’ try too hard and use complex methodologies that involve maths, or ‘scaling’ techniques that look test images as 13″x19″ prints – crazy!

All I’m interested in is how much of a scenes brightness range can I record on the sensor with one single exposure; and will I need to bracket exposures.

So let’s have a look at the performance of the 5D Mk4 sensor and see how much we can milk it for:

So here’s a scene outside ‘Chez Andy’ on a dull and rather overcast day – this gives the camera a better fighting chance than it would have on a bright blue sky day will full bore sunshine.

Evaluative metering gives a manual exposure reading of 1/30th sec for f8 at 100ISO (base ISO from what I can gather).

The two main regions of interest are obvious in any test of dynamic range – brightest highlights and darkest shadows, the areas indicated by the red circles, together with their spot metered values.

The indicated spot for the sky is a bit misleading – I actually pointed the camera straight up at the sky with the lens defocused and nothing but ‘sky’ in the frame!

Also, bare in mind that camera meters give you an exposure to record a tone as 50% grey!

So I shot a bracketed sequence from 1/250th to 4 secs, at 100ISO and f8.

The scene brightness range runs to a metered 11 stops, so if DXO Marks published test DR of 13.59Ev at 100 ISO (64 ISO as they would call it) is correct then one or more of these frames WILL contain detail in both the bright highlights and darkest shadows.

We might have to ‘recover’ that detail in post, but it should all be there within the recorded sensor output.

Guess what – it isn’t.  Very nearly, but not quite. Feel free to download the raw files yourself by clicking here (approx 2mins download).

To save a ton of typing and image uploading I’ll run a short video on how I do a quick assessment of the images to obtain a ‘real world’ ball-park DR value:

And purely as an exercise, what can we pull out of this single frame?

Looks somewhat HDR-ish because of the dramatic highlight and shadow recovery settings, but it just goes to show what you can pull back on this Canon 5D Mk4 sensor – you’d never pull this off on a single frame shot done with a 5D Mk3.

If I run the same type of rough analysis on the Nikon D810 and a descent bit of Zeiss glass I get a DR approximating 11.5 stops, and pretty much the same for the D800E.

More importantly, for the Canon 5D Mk3 the result is no more than 9.5 stops, but I’ve only tested it using the older 16-35mm f2.8 Mk2.

Just to clarify the DXO Mark ‘thing’ – while I either question or argue the numerical value of most of their sensor tests, the ‘trends’ identified within those results are pretty much spot on.

A good place to view more realistic DR values for a large number of sensors/cameras can be found here.

And as a final caveat regarding ANY sensor DR test – the test is based on the RECORDED SENSOR OUTPUT.  This is solely comprised of the ADC and image processors ‘digitised interpretation’ of the true ‘analogue output’ of the sensor. 

Is this a distortion of reality?  Maybe, but for the moment it’s what we’re stuck with!

So I think the Canon 5D Mk4 does pretty good on the dynamic range front, but the crazy high values the ‘third party idiots’ bandy about are just pie in the sky.

Frankly DR values of 13 to 14+ stops from a 14 bit ADC and a 36×24 sensor are something of a ‘step beyond’.  A 16 bit ADC on a medium format sensor on the other hand……but then that’s what you pay the big bucks for!

But just so we’re clear, the Canon 5D Mk4 DR is very noticeably greater than that of its predecessor.

Autofocus Performance

Now I’ve already posted about this HERE. So if you haven’t already read that then do so first.

I find the Canon 5D Mk4 noticeably faster in AF acquisition the the Mk3, and a lot more responsive when tracking subjects moving directly towards the camera.  It’s not a 1DX Mk2 under these circumstances, but I was surprised at just how close it came to its big brother in this respect.

However! Unlike the 1DX Mk2 which ‘sticks to a subject like glue no matter what’ in the tracking department, the Canon 5D Mk4 can sometimes chuck its toys out the pram when subjected to lens flare.

This means that back lit subjects CAN sometimes present a bit of a problem.

Back lit compositions against a dark background and without flare cause zero problems.

But introduce a bit of flare and things can go pear-shaped very quickly:

Please note: I said ‘can’ not ‘does’ – it doesn’t happen all the time.  But when it does, even keeping the AF tracking active and on target doesn’t help you when it does ‘stuff up’ – if it’s not focused in the desired plane at frame 1 it stays that way for the entire frame sequence.

This can most likely be cured with a firmware update, but as it stands at the time of writing then this shot, done with the 1DX Mk2 could be problematic:

…when you consider it’s just one frame from a long action sequence with lens flare where every frame is sharp.

But then again, the 5D Mk4 isn’t trying to be a 1DX Mk2; it’s just trying to be better at everything than the 5D Mk3 is/was.

ISO Settings – Noise

Hopefully you will have already read my post Camera ISO Settings – The Truth About ISO

If you haven’t then may I suggest you do – pronto!

ISO, or ‘post exposure applied gain’ is all relative to the number of photons passing through the lens and being collected by the photosites on the sensor.

The net result is that a shot at base ISO can look like crap if you are trying to photograph the ubiquitous ‘black cat in the coal house at midnight’, and 10,000 ISO can look epic in the presence of huge photon counts:

The Canon 5D Mk4 IS less noise at any ISO setting than its predecessor 5D Mk3, again simply because of the ‘on die’ or sensor-integrated ADC.

As I said earlier, the older Canons – and that includes the crackpot 5DS and SR – have off dye ADC components, and this limits the number of connections between the sensor and the ADC. This number was (I’m fairly certain!) limited to 8 with cameras fitted with a single Digic processor, and 16 in those with twin Digics.

In order for the system to turn a respectable image processing time this low number of communications channels or buses had to carry all the sensor data to ADCs that needed to chew it up and spit it out at a great rate of knots – in other words they are high frequency ADCs.

And here is the kicker; there is a rigid and inflexible bond between operating speed/frequency and noise.  This is the noise seen in your shadows – especially when you try to recover them by even a modest amount.

Moving the ADC ‘on die’ allows for more connections. This in turn allows for the use of ADCs with lower operating frequencies, which in turn results in a lower noise floor.

I’m not going to produce a raft of comparison shots between the Canon 5D Mk4 and its predecessor – hell, this post is long enough as it is, and there are plenty of them already on the net.

In Conclusion – Major Improvements over the 5D Mk3

The Canon 5D Mk4 IS a better camera than its predecessor in the two major attributes of a stills camera:

1. Faster Auto Focus with greater flexibility and control.
2. Improved Dynamic Range, Noise Floor and post-process latitude – all of which can be attributed to the switch by Canon to ‘on die’ ADC circuitry.

These above two improvements are major, and possibly more far-reaching than a lot of you may imagine.

Other Improvements:

1. More megapixels if that floats your boat.
2. Frame rate increased from 6fps to 7fps – though I don’t like a fixed fps personally.
3. Touch screen menu system.
4. Built-in GPS – which can drain the battery BTW if not set properly in the menu.
5. Built-in Wi-Fi – which I have yet to get working!

Things I Don’t Like:

1. Dual Pixel Raw – God in Heaven what a crock!  Dual pixel tech was created to give phase detection AF for video. But Still Camera Setting 2 on page 1 is like Canon thought “how can we turn this into a USP for the gullible stills-only camera buyer”.
2. SD media slot – come on Canon – twin CF (not twin C-Fast) or switch to XQD.
3. It eats batteries if you forget to turn off WiFi and GPS.
4. Pathetic lack of proper viewfinder blind – seriously Canon!
5. The persistent refusal of Canon to offer uncompressed RAW recording. It would take the smallest of firmware updates.  To me it just seems ridiculous not to give the user the choice as Nikon and others do.

So yes, in my opinion, the Canon 5D Mk4 is a better camera than the 5D Mk3.

If you own a 5D Mk3 have you GOT to trade it in?  That depends on what you want out of your camera and only you know that.

Would I trade in my D800E for one?  Hell NO!

But if you do fancy the upgrade from the Mk3 then, based on the review example I have here, you will see a considerable beneficial difference in your images – unless of course your name is Neil Burton!

Become a patron from as little as $1 per month, and help me produce more free content. Patrons gain access to a variety of FREE rewards, discounts and bonuses.