This post will be a bit surprising for those that think I am an MFT partisan and despise any other format, as you probably imagine that is far from truth. This post will look at the strength of the APSC DSLR segment.
If you follow the rumours and announcements of Canon and Nikon you are probably aware that Nikon is not planning any new professional APSC DSLR and Canon just released the last model with the 90D and will not be releasing a new 5D camera having just released the 1DX Mark III.
This is going to be a significant blow to underwater photographers around the world as today most of competition winners shoot an APSC DSLR camera, in particular the Nikon D500 is probably the most popular camera of serious underwater shooters.
What makes APSC DSLR Unique for Underwater Photography?
In an image we can understand what has made this format such a great option
The Tokina zoom fisheye is simply the best native option for wide angle underwater photography. It is cost effective and despite the apparent low quality on land it takes some amazing underwater pictures.
What makes this lens even more interesting is that it produces decent results with a small 4.33″ dome.
There are several option acrylic and glass and if you want even better quality you can go for larger dimensions.
Today you can get the Tokina 10-17mm with his port for £1000 which is less than the cost of a Nauticam WWL-1 and much less than any WACP or Nikkor Nikonos vintage lenses.
Another extremely popular choice but this time for macro is the Sigma 105mm F2.8 EX DG OS HSM Macro Lens or otherwise the OEM Canon 100mm and Nikon 105mm macro lenses. Those have been taking some amazing super macro shots in the last 10 years plus thanks to 150+mm equivalent focal length.
APSC Mirrorless Cameras
Both Nikon and Canon have launched new mirrorless APSC and with that a new lens format. Sadly the Tokina 10-17mm autofocus will no longer work. This is a major blow and we need to understand if Tokina will delivery a Z mount version of their mythical lens.
Mirrorless cropped format has been the domain of Sony and Micro Four Thirds due Olympus and Panasonic for the last 5 years and it looks like there are no benefits at sensor level between 1.5, 1.6 and 2.0 crop that are meaningful.
The other issue is that Canon and Nikon mirrorless are also behind in terms of autofocus compared to MFT, while they already are matching or beating Sony.
As a new user would you buy an APSC camera from Nikon or Canon or prefer Sony? Would you just get a micro four thirds that at least has commitment from two brands and a complete set of lenses and ports for underwater use? Nikon themselves have branded their Z50 camera as a non professional unit and make self limiting design choices that are evidence that their commitment is for full frame, this is the only segment where they are making profits currently.
Future of Full Frame DSLR
Canon is definitely abandoning the DSLR ship and has some good mirrorless penetration with their 5DSR and have just announced the EOS R5 that will be the first unit with IBIS and have 8K video.
Nikon is still hanging to their upper range D series for full frame DSLR but it has been also moving strongly into full frame mirroless.
Both Canon and Nikon are no longer developing their DSLR lenses mounts.
Considering the domination of Nikon in the full frame underwater photography segment the full decline of DSLR will not happen for at least another 2 or 3 years but the time will come.
The extinction of APSC DSLR is not good news for underwater photography as currently no other format can match the choice of ports and lenses available to those shooters. There is a risk that a camera like a Nikon D500 becomes a precious 2nd hand commodity however shutters do wear out so this is not a sustainable path.
A few years ago we witnessed the death of compact cameras to phones and this was a first blow to entry level underwater photographers.
The upcoming death of APSC DSLR is going to hit deeper in the semipro user group however alternatives are available thought not matching the same flexibility of lenses and ports.
Our passion is getting increasingly more expensive as the digital camera market focusses on full frame and also more bulky and difficult to carry around.
You will notice that the featured image is actually a bird in flight. When we think about fast autofocus birds in flights is what is really going to test performance.
This image was taken by my wife with a Nikon D7100 and a Sigma 70-200mm lens in the Galapagos Islands.
I also shoot birds with my Panasonic G9 and have a direct experience of focus systems for moving subjects and I can comfortably say that today AI has become more important than anything else for those kind of shots. Artificial intelligence predicts movement and ensures that once the camera has reached focus the first time it reacts automatically to movement without the need to refocus.
Let’s start from the basics first.
Types of Autofocus
There are two types of systems for auto focus in digital cameras:
Both systems need contrast to focus despite the naming convention, so phase detection works on contrast too.
In situation of low light low contrast EVERY camera switches to contrast detection without exceptions.
Contrast Detection AF
This is the simplest and cheapest way to obtain focus and is what is typically implemented in compact cameras. Contrast detection moves the focus back and forth to find the maximum contrast and then locks on subject. This is sometimes perceived as hunting by the user when the camera fails to find focus.
Contrast detection is the most accurate method of autofocus as it looks for perfection without prioritising time. With exception of Panasonic no other major brands use contrast detection AF on high end or semipro models.
Phase Detection AF
With this technique the image goes through a prism and it is split then when the two parts match the subject is in focus and the focus locked.
Phase detection is less accurate than contrast detection in particular there are instances in which focus is achieved in front or behind the subject. This is the system implemented by Nikon, Canon, Olympus and Canon.
This system combines both methods, it starts with phase detect to determine the focus start and then uses contrast detect to make sure the focus is accurate. Sony is the main driver of this technology.
Low Light Focus
All autofocus methods need light to function without exception, when the scene is really dark cameras have some methods to achieve focus, this includes:
Using the lens widest aperture to focus
Bump the ISO and then adjust later
Auto focus illuminator and modelling lights
Generally low light is less than 1.25 Lux or candela per square meter representing a really dark scene.
Pro and Cons of Each System
If we look at the three systems each one has positive and negatives and depending on the subject this are more or less important.
AF comparison Table
Performance Requirement for Underwater Photography
Many underwater photographer think that they need a system that focus fast, can track moving objects and work well in the dark, this system of course does not exist.
In particular considering the availability of focus lights the performance in low light is definitely not a show stopper. More important are speed and accuracy. For the purpose of a comparison I have included here some models from Sony, Panasonic, Olympus, Nikon and Canon with a variety of formats representing some popular choices among underwater photographers.
I have included 3 performance metrics for comparison:
Low Light Low Contrast Ev
Low Light High Contrast Ev
The first measure tells you how quick the camera focuses in normal conditions, this is in my opinion the most important parameter as generally underwater photography is not below 1 Ev.
The second measure is the number of Ev of low light the camera can still focus with a low contrast subject, and finally the third is still a low light scenario with a high contrast subject. Let’s look at the results that are build using test data from imaging resource.
I have used conditional formatting so green is good amber is average and red is bad for each category.
First observation is that hybrid AF is very slow, second contrast AF as implemented by Panasonic is faster than most of DSLR peers in this table. If we consider 0.2 seconds as acceptable the full frame mirrorless Sony A7RIII has unacceptable performance. While the Nikon D850 AF is in another league both MFT Olympus and Panasonic are faster than other APSC and even the canon full frame.
Low Light Low Contrast AF
Mirrorless cameras dominate this category, the Panasonic GH5 can reach focus at -4.5 Ev that is practically dark on a low contrast subject, second is the Sony A7 RIII and third the Olympus OMD-EM1MKII.
In a low light scenario phase detection fails sooner so some of those cameras switch to contrast detection to achieve focus.
Low Light High Contrast AF
All cameras are able to work at least at -3 Ev so this is not a distinctive category, it is worth nothing that some phase detect system that failed in the low contrast target scenario perform well in this category but generally performance is pretty decent.
Why are your shot blurred?
Some people that have the camera in the table still struggle to get shots, why is that? I have found that for most users do not read instruction manuals and to make it worse modern camera have far too many AF settings. My GH5 for example has 6 options of AF area, 4 options for AF Mode, 3 parameters for tuning the AI (artificial intelligence) engine, plus additional custom modes to select the 225 focus points in any random shape you like. The average person will skip all of this and select one option and then fail the shots.
Surprisingly for some if we look overall at the camera that has green in all categories we find two mirrorless micro four thirds. Even more surprisingly both those cameras are faster to focus than APSC DSLR from Canon and Nikon although it is not really a great distance.
Typically when it comes to comparison between camera there is someone that says but camera X gets the shots blurred so speed does not matter. I talk by direct experience with outdoor and birds not just fish and I can tell you that each system will miss shots in burst mode but more importantly underwater photography is nowhere near requirements for birds in flight.
I have performed tests with a light meter at less than 1 candela per square meter with my GH5 with a 60mm macro lens and with my surprise it focuses just fine without the AF illuminator. I have to admit I do not really trust auto-focus so in most situation I use back button and peaking however based on my recent findings I need to trust autofocus a bit more it seems!
Following from my previous post I wanted to further investigate the implications of formats and megapixels on Macro Underwater Photography.
I also want to stress that my posts are not guides on which camera to choose. For Macro for example some people rely on autofocus so there is no point talking about sensors if your camera does not focus on the shot!
Macro underwater photography and fish portraits in general is easier than wide angle because is totally managed with artificial illumination, although some real masterpieces take advantage also of ambient light.
There are a number of misconceptions also here but probably on the opposite side of wide angle there is a school of thinking that smaller cameras are better for macro but is that really the case?
Myth 1: Wide angle lens -> More Depth of field than Macro
Depth of field depends on a number of factors you can find the full description on sites like Cambridge in Colour a good read is here.
A common misconception without even starting with sensor size is that depth of field is related to focal length and therefore a macro lens that is long has less depth of field than a wide angle lens.
If we look at a DOF formula we can see that the effect of focal length and aperture cancel themselvers
A long lens will have a smaller field of view of a wide lens so the distance u will increase and cancel the effect of the focal length f.
The other variables in this formula are the circle of confusion c and the F-number N. As we are looking at the same sensor the c number is invariant and therefore at equal magnification the depth of field depends only on F number.
Example: we have a macro lens 60mm and a wide angle lens 12mm, and a subject at 1 meter with the 60mm lens. In order to have the same size subject (magnification) we need to shoot at 20cm with the 12mm lens at that point the depth of field will be the same at the same f-number.
So a wide angle lens does not give more depth of field but it gets you closer to a subject. At some point this gets too close and that is why macro lenses are long focal so you can have good magnification and decent working distance.
Myth 2: Smaller Sensor has more depth of field
We have already seen that sensor size is not in the depth of field formula so clearly sensor size is not related to depth of field so why is there such misconception?
Primarily because people do not understand depth of field equivalence and they compare the same f-number on two different formats.
Due to crop factor f/8 on a 2x crop sensor is equivalent to f/16 on a full frame and therefore as long as the larger sensor camera has smaller possible aperture there is no benefit on a smaller sensor for macro until there are available apertures.
So typically the smaller sensor is an advantage only at f/22 on a 2x MFT body or f/32 on a APSC compared to a DSLR. At this small aperture diffraction becomes significant so in real life even in the extreme cases there is no benefit.
Myth 3: Larger Sensor Means I can crop more
The high level of magnification of macro photography create a strain on resolution due to the effects of diffraction this has a real impact on macro photography.
We have two cases first case is camera with same megapixel count and different pixel size.
In our example we can compare a 20.3 MFT 2x crop camera with a 20.8 APSC 1.5x crop and a 20.8 full frame Nikon D5.
Those cameras will have different diffraction limits as they have pixels of 3.33, 4.2 and 6.4 microns respectively those sensor will reach diffraction at f/6.3, f/7.1 and f/11 respectively so in practical terms the smaller camera format have no benefit on larger sensor as even if there is higher depth of field at same f-number the equivalent depth of field and diffraction soon destroy the resolution cancelling the apparent benefit and confirming that sensor size does not matter.
Finally we examine the case of same pixel size and different sensor size.
This is the case for example of Nikon D500 vs D850 the two cameras have the same pixel size and therefore similar circle of confusion. This means that they will be diffraction limited at the same f-number despite the larger sensor. So the 45.7 megapixels of the D850 will not look any different from the 20.7 megapixels of the D500 and none will actually resolve 20.8 megapixels.
So what is the actual real resolution we can resolve?
Using this calculator you can enter parameters in megapixels for the various sensor size.
In macro photography depth of field is essential otherwise the shot is not in focus, for this exercise I have assumed comparable aperture and calculated the number of megapixels until diffraction destroys resolution
Resolution in Megapixels at constrained DOF
Note that the apparent benefit of MFT does not actually exist as the aspect ratio is 4:3 so once this is normalised to 3:2 we are back to the same 6.3 megapixels of full frame. APSC that has the strong reputation for macro comes last in this comparison.
So although you can crop more with more megapixels the resolution that you can achieve is dropping because of diffraction and therefore your macro image will always look worse when you crop even on screen as now most screens are 4K or 8 megapixels.
For a macro image depth of field is of course essential to have a sharp shot however we have seen that sensor size is not actually a consideration and therefore everything is level.
Color depth is important in portrait work and provided we have the correct illumination full frame cameras are able to resolve more colours. We are probably not likely to see them anyway if we are diffraction limited but for mid size portraits there will be a difference between a full frame and any cropped format. In this graph you can see that there is nothing in between APSC and MFT but full frame has a benefit of 2.5 Ev and this will show.
Surprisingly for most the format that has an edge for macro is actually full frame because it can resolve more colours. The common belief that smaller formats are better is not actually true however some of those rigs will definitely be more portable and able to access awkward and narrow spaces to what extent this is an advantage we will have to wait and see. It may be worth noting that macro competitions are typically dominated by APSC shooters whose crop factor is actually the worst looking at diffraction figures.
The objective of this post is not to determine what is the best camera for underwater photography, as that is simply the best camera with the best housing and the best strobes and lenses. All needs to be seen as a system in order to take stunning images.
The purpose of this article is to provide some clarity and eliminate common misconceptions that seem to be hindering the decision making of a person wanting to take underwater photos. There is always a vested interested of camera manufacturers to drive sales as well as underwater photography equipment shops to push users to upgrade their gear as frequently as possible as that generates value to them, however this will not necessarily generate value to you the consumer, the only person injecting cash in this network.
I recently posted on WetPixel a discussion that to generate a debate about the gap between APSC and MFT cameras. This in turn made me do some more research on camera sensor and I found some information that is very insightful and confirms some of my suspicions I had years ago when I attended a workshop in the Red Sea with Alex Mustard. In that occasion I was the only user on the boat with a compact camera but managed to pull some decent shots and this made me realise that there are circumstances that equalise your equipment and make the gap in the image quality smaller to the point that a compact camera picture in some cases looks similar to a much larger sensor camera. Although I shoot micro four thirds underwater I have owned and shot DSLR full frame and cropped, film and digital, I have also had an array of compact cameras, so what you are going to read is not focused on one format being better than another.
Let’s discuss some of those misconceptions in more detail.
For those that do not understand optics of dome ports underwater the reason you need to stop down the aperture is NOT because you are looking for depth of field, in fact on land you would shoot a wide angle lens wide open and it would have plenty of depth of field. The reason to stop down the lens is the field of curvature of the dome which makes the areas off centre and on the edges soft this can only be fixed by stopping down the lens. So before you think I can shoot at f/4 on a APSC so what think that your pictures will be mostly blurry on the side and besides each format has got fast lenses so this is not a main consideration for what you are going to read.
Myth 1: Larger Sensor -> Better SNR
Signal to Noise ratio is an important factor in electronics as it allows to distinguish information from noise. Contrary to what most people think SNR is not related to sensor size.
The comprehension of some of the concept may be too hard for many so I will attempt a simplification. What R.J.Clark says is that you need to balance the amount of light hitting the sensor before drawing conclusion on SNR. For example assume a camera with a lens of 16mm on a full frame sensor and compare this with a camera with a lens of 8mm on micro four thirds, I am using MFT as crop factor is two and makes examples easier.
An exposure of f/8 on a 16mm lens on Full frame camera is equivalent to an exposure of f/4 on a 8mm lens on MFT. Those will send the same amount of light to the sensor at equivalent exposure. However the smaller sensor will have the same amount of light distributed on a surface that is 1/4 of the larger sensor and therefore if we equalise everything we have a situation whereby the exposure value are balanced and the SNR is pretty much identical because the gain or ISO value necessary was 1/4 of the larger sensor. This SNR 18% graph on DxOMark gives an idea. I have chosen 3 cameras with the same megapixel count to remove megapixels from the discussion.
Once exposure is equalised the larger sensor has no longer a benefit this is due to the fact that the components of noise shot noise and read noise do not depend on sensor size.
However an important consideration is that ISO 100 does not actually mean the same gain in all systems and in fact a larger camera will have more photons than a smaller one at the same ISO level, this means that at the so called base ISO the larger sensor camera will have an advantage as the smaller sensor can’t decrease the ISO anymore and will need to close aperture. It also means that ISO 100 does not mean the same SNR amongst different formats. So when we compare two shots at the same ISO larger sensors will have more signal than smaller ones. This is the reason sometimes you hear things like why is my shot on my compact camera so noisy at ISO 400 compared to a full frame that looks so clean at ISO 400 but those ISO are not actually the same thing and the smaller sensor has much less photons at that identical ISO number.
Another consequence of this is that as the camera in questions have the same megapixel size larger pixels do not yield better SNR.
However with larger pixels holding more signal it is possible to extend the range of an amplifier to higher value of gain therefore larger pixel camera (less megapixel on the same size) will be able to work at higher ISO levels. This is the reason why MFT camera have a lower maximum ISO than full frame at same megapixel count.
Underwater we use strobes to counter colour absorption and never reach those high ISO levels. If you were shooting at night on land without a flash you may easily reach high ISO value like ISO 25800 or 64000 with strobes however we rarely reach even values like 1600.
Myth 2: Larger Sensor -> Better Dynamic Range
The characteristic that drives dynamic range is not actually sensor size but pixel size however at some point DR no longer grows with very large pixels.
This graph shows that the Panasonic GH5 has a respectable DR at low ISO however it drops faster than the D500 and 1DX MkII. Surprisingly for some the D500 has more DR than the larger pixel 1Dx MKII.
If we look at the maximum possible DR and the ISO at which we would still have 7 bits colour and at least 10 stops of DR we have the following values:
Highest Usable ISO
Canon 1DX MKII
The larger pixel size makes usable DR go to higher ISO
Although the larger pixel camera does not hold the highest DR it is able to shoot at higher ISO and still keep a decent DR and color tone.
If we calculate the Ev between the ISO value we see that the MFT sensor is 2 Ev away and the APSC is 1.3 Ev away from full frame, this is pretty much in line with the crop factor and therefore once we equalise Depth of field there is no benefit between the various formats at same megapixel count, though the Nikon D500 is the camera that has the highest DR in absolute value. So if you have an extremely high amount of light the D500 would be able to product a high DR image. Underwater however this is rarely the case underwater so the conclusion is that if you are after a 20 Megapixel camera there is no material difference among the various formats in practical underwater use.
Myth 3: Larger Pixels are Better at equal sensor size
Although larger pixels are better at sustained dynamic range, for example in low light, evidence shows that as long as the camera is not limited by diffraction more megapixels are better.
I am comparing here 3 Nikon full frame cameras that have respectively 24, 36 and 47.8 Megapixels.
SNR is not impacted by the sensor resolution and this is due mostly to the fact that at similar size downsampling equalises the smaller pixels.
Looking at Dynamic range the situation is the same and actually the camera with more megapixels has an edge until ISO value become very high.
Finally the graph for color sensitivity, an important metric for shots with strobes and portrait work, confirms that more megapixels also bear better results.
Please note that this data is limited to sensor analysis and does not take into account the effect of very small pixels on diffraction and sharpness that is a topic on its own.
Choosing a Camera for Social Media
Today majority of people do not print their images and post them on social media or website. Those typically have a low resolution frequently less than 4 megapixels. Screens usually have low dynamic range, and JPEG images are generally limited to 12 Ev Dynamic Range this is a value that is at reach of any camera today starting from 1″ compact cameras but is unreachable to majority of computer screens or phones.
My suggestion for users that post on social media is to find the best camera that fits their budget and ergonomics and worry less about sensors, invest in optics either wet or lenses and port and strobes, as those will yield a higher return.
Today most cameras have a port system anyway so an advanced compact such as the Sony RX100 series or a Micro Four Third camera of small factor (Panasonic GX9 for example) are more than enough.
Choosing a Camera for Medium Size Print
I print my images typically on 16″x12″ or 18″x12″ paper or canvas.
Generally I want to have around 300 dpi so that means I need a 20 Megapixel camera as a minimum. This cuts out a large part of the smaller MFT cameras and also the compacts because the real life resolution is far from the declared pixels.
In my opinion, if you are a user that prints medium formats, a pro grade MFT or an APSC camera is all you need, besides the latest winner of UPY shoots an APSC with a Tokina lens and plenty of winners don’t use full frame.
For those who just want the Best
The best image quality today is produced by high megapixel full frame cameras there is no doubt about it. Full frame cameras however are subject to depth of field issues and as we have seen once you shoot at equal depth of field the benefit is for most eroded.
To get the best outcome of a high megapixel full frame camera you need to be able to shoot at the lowest possible ISO, this is almost impossible if you are shooting a fisheye lens behind a dome as your aperture of f/11 means very little light is hitting the sensor so your ISO will most likely hit 400 many times and at that point the benefit of full frame is gone.
I have looked at all technical details of Alex Mustard images on his book and nearly all shots taken with a full frame camera have at least ISO 400 or higher, with very few exceptions at 200 or lower.
So how to do you manage to shoot at the lowest possible ISO on full frame? You need to be able to shoot at wider aperture and this today means optics like the Nauticam WACP that have two stops benefit on a wide angle lens and three on a rectilinear lens behind a dome on full frame.
The WACP however has a field of view of 130 degrees and therefore is not as wide as a fisheye and unsuitable for close focus wide angle, recently Nauticam has released the WACP-2 that retails at $7,460 and can cover 140 degrees.
My consideration is that, if you are not prepared to spend money for a WACP like solution, then there is no point investing in a full frame system as the benefit goes away once you equalise depth of field.
Underwater photography is an expensive hobby and every time I am on a boat and see how much money goes into equipment to product average photos this saddens me. While improving technique is only a matter of practice and learning, making the right choice is something we can all do once we have the correct data and information at hand.
I hope this post is useful and helps your decision making going forward.