Jump to content

Evaluation of Auto focus systems in current cameras

Recommended Posts

Requirements of stills and video are quite different though. What Sony does very well is tracking the subject, something that the Panasonic Mu43 unfortunately don‘t do that well. AF is certainly the most difficult topic you‘ve tackled in your spree of recent articles.

Share this post

Link to post
Share on other sites

I am afraid that using laboratory test poorly reflects the actual performance of the cameras in real life situation, even less in underwater situations.


NOTE: This low light test is conducted with a stationary subject, and the camera mounted on a sturdy tripod. Most digital cameras will fail miserably when faced with a moving subject in dim lighting. (For example, a child's ballet recital or a holiday pageant in a gymnasium.) Digital SLRs like the Nikon D7500 do much better than point & shoots, but you still shouldn't expect a quick autofocus lock with moving subjects.

Imaging Resource warns that these low light AF are only relevant in stationary situations.

Indeed, if you are shooting a reef in wide angle, the performance of the autofocus will not be an issue.

In the other post, examples were given of situations where camera autofocus is challenged:

  • black water diving: this is perhaps the most challenging. My experience of black water diving is that there is constant movement and very low light. 
  • macrophotography: here the issue is not that the subject is moving quickly, but that the depth of field is so shallow that the slightest move (current, etc.) will have a huge impact on the focus
  • juvenile fish portraits: those little devils typically keep moving very fast. Another challenge for autofocus

Performance in these real life situations cannot be measured in stationary laboratory tests. In term of difficulty of focusing, I am not sure that shooting a bird in flight is more difficult that focusing on a pigmy seahorse in a wavy sea ...

Autofocus is an extremely complex topic, and different test protocols may get very different results. For example a French photo magazine (Chasseur d'Image, excellent publication, really independent minded and thorough test methodology) is using a moving target at 50 km/h and measure the ability of the camera to make shots as the target move from 50 m to 0m. Most modern cameras are able to focus accurately, at 5 to 9 i/s depending on their speed. But as the target gets near, some AF start to fail, as the nearer the target, the greater AF adjustment is needed between two shots. How relevant for our UW photography? Not much probably (more relevant for car racing maybe), but it shows the difficulty to test such complex features as AF. It also shows that thanks to progress, most good modern camera perform well in normal situations. However in UW photography, we are often far from "normal" situations.

Therefore I would not draw any conclusions from these tests for UW photography.

You are right to highlight the need to know the camera features, and how to select the right parameters depending on shooting conditions. It probably matters much more than the difference of performance between otherwise very good cameras.


Share this post

Link to post
Share on other sites

The other factor that comes into play is the lens, some lenses are faster at AF than others.  Most systems slow down with macro AF as you get down towards life size.  Not an easy topic by any means, but you have to start somewhere. 

Share this post

Link to post
Share on other sites
Posted (edited)

Frankly I have zero issues with focus with my cameras. The worst performance I have ever had is with the Nikon D7100 that had a talent in focusing behind and in front of the subject, this happened regularly on sport and action shots.

Today  UW I use autofocus single point for wide angle and never miss a shot. For macro I use single point  AF or back button for super macro, again I do not have issues of any sort other than myself being stable.

I think it is fair to say that nobody does research starting on how well autofocus performs and nobody goes anymore to a shop to test let along the availability of real life situation for an underwater shooter.

I think the key point that is coming up is that underwater photography requirements are actually not that high on a camera. Yes imaging resource tests are not indicative for much, I have written it myself in the article, however you never reach that level of darkness. I have a light meters and I sat in a room until I had 16 candela it was very very dark and the camera still focussed. There are aids underwater so actually I do not think that low light focus is really a consideration. I do like the low light low contrast target test though as this demonstrates what happens when the autofocus moves into contrast detection. Contrary to what most think when there is no light all cameras switch to contrast detection and there you can see that some manufacturers do not really model that into their thinking.

For sure the most important metric of all is autofocus speed in normal conditions and I had two key learnings:

1. Hybrid systems are slow. You can see the Sony performance drop and I think using adapted lenses this is even worse. So for my Sony remains a concern for stills especially long focal. Yes it may be good for video but then Sony camera have all sorts of other codec issues and therefore really not a choice for me.

2. Autofocus technology is a separate topic on its own. I have two cameras right now both Panasonic GH5 and G9. On a single shot the GH5 is actually faster and you can push it in low light more however the G9 has superior artificial intelligence, it can detect several animal shapes and even a person when is turned on their back or the face is sideways. There is no doubt that for land photography this by far is the most important thing for action and wildlife shots. For fish there is no AI so I would not trust predictive focus that much.

I have also learned that some manufacturers like Panasonic and Olympus had to fight hard to demonstrate their capabilities when you half press the shutter there are a whole set of tricks that kick in. On a DSLR the camera simply uses the fastest aperture to find focus (which is the reason why on phase detect accuracy fails and you have back or front focus). On MFT the camera does all sorts of thing, aperture wide (less problem as more depth of field for accuracy), ISO bump, boost of the EVF and LCD, peaking overlayed on AF! you can clearly see the camera goes 4-5 Ev higher to focus I have not seen this on DSLR and I think it is only possible if you perform phase or contrast detect on the sensor itself

In conclusion I think that saved the obvious issues with hybrid systems, autofocus is another very overrated myth of underwater photographers and almost all pro or semipro camera are good enough and we should stop making excuses and learn how our own equipment work and go and take some shots

Re @ChrisRoss comment I fully agree however at high level of magnification back button focus is almost the norm even for people that normally use tracking. The issue there is steadiness unless you are rock solid you can't afford to wait for AF to work its magic whilst you are struggling to frame the shot, let alone possible particles etc etc. The point about performance with specific combination is actually specific to MFT due to the fact the lens are heavily programmed and NOT entirely interchangeable between manufacturers (which should not be). We need some third party macro lenses from Sigma to break the Panasonic vs Olympus sabotage

Edited by Interceptor121

Share this post

Link to post
Share on other sites



  • Create New...