Interesting article: iPhoneX against the M100

Dpreview published an interesting article today about the iPhoneX compared to a Canon 100 with a 35mm 2.0 lens attached.

From the 10,000 foot level the iPhone images do pretty good - unless you wanted to do something other than viewing it on your phone.  The image breaks down as you look closely at the details.  Fast.

Computational photography is certainly a way these phone manufacturers can start to make a difference, and I have to start reading the logic patents a little more carefully to see what Canon is looking at doing in this area.  I have skimmed some patents that deal with it, however, these patents are extremely difficult to translate down to a common level of understanding.

The Canon DPAF sensor is in a unique spot - as it can work as an ILC sensor, and also at the same time be used for computational photography to increase or change background blur, focus point, DOF and do distance mapping to do all sorts of crazy things to the image.  As with the iPhoneX every single canon camera with a dual pixel sensor has a "distance map" of the image.  This is why from a pure speed aspect the EOS M's tend to lag Sony,etc.  There's a lot of heavy lifting taking 20 million some odd calculations for phase difference versus 400 some odd AF points.  As DIGIC gets faster, so will the EOS-M AF engines and the liveview AF in your DSLRs as well.

We have all seen the birth of ILC computational photography with the 5D Mark IV as the dual pixel RAW format is Canon's first foray into computational image photography.  Funny how it seems that Canon innovations always seem to get glanced over.  We hope this continues in DPP (Digital Photo Professional - Canon's own RAW image manipulation software) and also extends further as there are other applications of dual pixel raw that hasn't been taken advantage of (for example, there is 1EV of additional highlight headroom in a dual pixel RAW image).

Interesting times are ahead... 

As far as the article itself, dpreview concludes;

We're approaching a time of reckoning for traditional camera manufacturers. Not only are computational cameras getting better, but they're increasingly in people's pockets, at the ready whenever they're needed.

There are, of course, aspects of traditional cameras that phones can't replace; the form factor, the controls, the feel of the thing. But those are increasingly diminishing requirements for a broad range of photographers (especially since, as you well know, everyone these days is a photographer).

But to remain relevant, these sort of software 'tricks' are something that camera manufacturers are going to need to think more and more about. There may yet come a time when, finally, you don't absolutely need a bigger sensor for better results. And it's not necessarily a matter of 'if,' but a matter of 'when.'