While learning to take better photos with within the opportunities and limitations provided by whatever camera technology offers, it is also interesting now and then to stop to reflect on how things are evolving.
This weekend, I took some time to study rainy tones of Autumn, and also to hunt for the “perfect blues” of the Blue Hour – the time both some time before sunrise and after the sunset, when indirect sunlight coming from the sky is dominated by short, blue wavelenghts.
After a few attempts I think I got into the right spot at the right time (see the above photo, taken tonight at the beach of Hervantajärvi lake). At the time of this photo it was already so dark that I actually had trouble finding my gear and changing lenses.
I made the simple experiment of taking an evening, low-light photo with the same lens (Canon EF 50 mm f/1.8 STM) with two of my camera bodies – both the old, Canon EOS 550D (DSLR) and new EOS M50 (mirrorless). I tried to use the exact same settings for both photos, taking them only moments apart from the same spot, using a tripod. Below are two cropped details that I tried to frame into same area of the photos.
I am not an expert in signal processing or camera electronics, but it is interesting to see how much more detail there is in the lower, M50 version. I thought that the main differences might be in how much noise there is in the low-light photo, but the differences appear to go deeper.
The cameras are generations apart from each other: the processor of 550D is DIGIC 4, while M50 has the new DIGIC 8. That sure has a effect, but I think that the sensor might play even larger role in this experiment. There are some information available from the sensors of both cameras – see the links below:
While the physical sizes of the sensors are exactly the same (22.3 x 14.9 mm), the pixel counts are different (18 megapixels vs. 24.1 megapixels). Also, the pixel density differs: 5.43 MP/cm² vs. 7.27 MP/cm², which just verifies that these two cameras, launched almost a decade apart, have very different imaging technology under the hood.
I like using both of them, but it is important to understand their strengths and limitations. I like using the old DSLR in daylight and particularly when trying to photograph birds or other fast moving targets. The large grip and good-sized physical controls make a DSLR like EOS 550D very easy and comfortable to handle.
On the other hand, when really sharp images are needed, I now rely on the mirrorless M50. Since it is a mirrorless camera, it is easy to see the final outcome of applied settings directly from the electronic viewfinder. M50 also has an articulated, rotating LCD screen, which is really excellent feature when I need to reach very low, or very high, to get a nice shot. On the other hand, the buttons and the grip are just physically a bit too small to be comfortable. I never seem to hit the right switch when trying to react in a hurry, missing some nice opportunities. But when it is a still-life composition, I have good time to consult the tiny controls of M50.
To conclude: things are changing, good (and bad) photos can be taken, with all kinds of technology. And there is no one perfect camera, just different cameras that are best suited for slightly different uses and purposes.
Is it worth setting your smartphone camera to use RAW format, instead (or: alongside) of the standard JPG format?
I must say I am not sure. Above you should be able to see three versions of the same photo. The first one is one produced with the automatic settings of my Huawei Mate 20 Pro. It is a f/1.8 photo coming from the main camera module, processed with various algorithms to create a “nice”, tonally rather balanced JPG with 2736 x 3648 pixels.
The second one is direct/non-edited conversion of the original RAW (imported into desktop Lightroom, then directly turned into JPG), with 5456 x 7280 pixels and plenty of information that is potentially valuable for editing, yet it is also bit too dark and the lens quality is frankly probably not quite worth all those pixels, to start with (the depth of field is narrow here, and most of the photo is soft, when you look it 1:1 in a large screen).
The third version is the RAW-based and Lightroom-edited photo, where I have just accepted some “auto” corrections that the software has available for beginners. This time, we can see many of the details again better, since Lightroom has tweaked the exposure and contrast settings and tonal curves. Yet, the change of white balance setting into the automatic “daylight” version has made the cold, Autumn morning photo to appear a bit too warm in colours to my mind.
This could be of course fixed in further, more nuanced and sensible Lightroom editing, but the point perhaps is that the out-of-camera JPG that Huawei is capable of producing is a rather nice compromise in itself. It is optimised for what the small, fixed lenses are capable of achieving, and the file size is good for sharing in social media – which is what most smartphone photos are used for, in any case. Artificial intelligence does it best to produce what a typical “Autumn Leaves” shot should look like. That might then again be something that you like – or not.
It is surely possible to achieve more striking and artistically ambitious (“non-typical”) outcomes when the original photo is taken in RAW, even when it is coming from a smartphone camera. But I would say that the RAW based workflow probably really makes sense when you are using a SLR style camera with a lens that is sharp enough for you to really go deep into the details, do some more ambitious cropping or tonal adjustments, for example.
Or, what do you think?
There are various articles online that you can also have a look on this, e.g.
Two years ago, in August 2017, I installed a new operating system into my trusty old home server (HP Proliant ML110 Gen5). That was a rather new Linux distro called ElementaryOS, which looked nice, but the 0.4 Loki that was available at the time was not an optimal choice for a server, as it soon turned out afterwards. It was optimized for a laptop use, and while I could also set it up as a file & printer server, many things required patching and tweaking to start working. But since I install and maintain multiple operating systems in my device environment partly out of curiosity, keeping my brain alert, and for this particular kind of fun – of tweaking – I persisted, and lived with Elementary OS for two years.
Recently, there had been some interesting new versions that had come out from multiple other operating system versions. While I do most of my daily stuff in Windows 10 and in iOS (or ipadOS, as the iPad variant is now called), it is interesting to also try out e.g. different Linux versions, and I am also fan of ChomeOS, which usually does not provide surprises, but rather steadily improves, while staying very clear, simple and reliable in that it does.
In terms of the particular characteristic that I am here talking about – let’s call it “tweakability”– an iPad or Chromebook are pretty much from the opposite ends of spectrum, as compared to a personal computer or server system running some version of Linux. While the other OSs excel in presenting the user with an extremely fine-tuned, clear and simple working environment that is simultaneously rather limited in terms of personalisation and modification, the bare-bones, expert oriented Linux distributions in particular hardly ever are “ready” straight after the initial setup. The basic installation is in these cases rather just the starting point for the user to start building their own vision of an ideal system, complete with the tools, graphical shells, and/or command-line interpreters etc. that suit their ways of working. Some strongly prefer the other, some the opposite style of OS with their associated user experiences. I feel it is optimal to be able to move from one kind of system to another, on basis of what one is trying to do, and also how one wants to do it.
Tweakability is, in this sense, a measure of customisability and modifiability of the system that is particularly important for so-called “power users”, who have a very definite needs, high IT skill levels, and also clear (sometimes idiosyncratic) ideas of how computing should be done. I am personally not entirely comfortable in that style of operation, and often rather feel happy that someone else has set up an easy-to-use system for me, which is good enough for most things. Particularly in those days when it is email, some text editing, browser-based research in databases and publications (with some social media thrown in), a Chromebook, iPad Pro or a Windows machine with a nice keyboard and good enough screen & battery life are all that I need.
But, coming back to that home server and new operating system installation: as my current printer has network sharing, scanning, email and all kinds of apps built-in, and I do not want to run a web server from my home any more either, it is just the basic backup and file server needs that this server box needs to handle. And a modern NAS box with some decent-sized disks could very well do that job. Thus, the setup of this Proliant server is more of less a hobby project that is very much oriented towards optimal tweakability these days (though not quite as much as my experiments with various Raspberry Pi hobby computers, and their operating systems).
So, I finally ended up considering three options as the new OS for this machine: Ubuntu Server 18.04.3 LTS (which would have been a solid choice, but since I was already running Ubuntu in my Lenovo Yoga laptop, I wanted something a bit different). The second option would have been the new Debian 10 (Buster) Minimal Server (probably optimal for my old and small home server use – but I wanted to also experiment with the desktop side of operating system in this installation). So, finally I ended up with Linux Mint 19.2 Tina Cinnamon Edition. It seemed to have the optimal balance between reliable Debian elements, Ubuntu application ecosystem, combined with some nice tweaks that enhance ease of use and also aesthetic side of the OS.
I did a wipe-clean-style installation of Mint into my 120 GB SSD drive, but decided to try and keep all data in the WD Red 4 TB disk. I knew in principle that this could lead into some issues, as in most new operating system installations, the new OS will come with a new user account, and the file systems will keep the files registered into the original User, Group and Other specifications, from the old OS installation. It would have been better to have a separate archive media available with all the folder structures and files, and then format the data disk, copy all data under the new user account, and thereby have all file properties, ownership details etc. exactly right. But I had already accumulated something like 2,7 terabytes of data into this particular disk and there was no exact backup of it all – since this was the backup server itself, for several devices in our house. So, I just read a quick reminder on how chmod and chown commands work again, and proceeded to mount the old data disks within the new Mint installation, take ownership of all directories and data, and tweak the user, group and other permissions into some kind of working order.
Samba, the cross-platform file sharing system that I need for the mixed Windows-Linux local network to operate was the first really difficult part this time. It was just plain confusing to get the right disks, shares and folders to appear in our LAN for the Windows users, so that the backup and file sharing could work. Again, I ended up reading dozens of hobbyist discussions and info pages from different decades and from different forums, making tweak after tweak in users, groups, permissions and settings in the /etc/smb.conf settings file (followed every time to stop and restart the Samba service daemon, to see the effects of changes). After a few hours I got that running, but then the actual fun started, when I tried to install Dropbox, my main cloud archive, backup and sharing system on top of the (terabyte-size) data that I had in my old Dropbox folder. In principle you can achieve this transition by first renaming the old folder e.g. as “Dropbox-OLD”, then starting the new instance of service and letting it create a new folder named “Dropbox”, then killing the software, deleting the new folder and renaming the old folder back to its own default name. After which restarting the Dropbox software should find the old data directory where it expects one to be, and start re-indexing all that data, but not re-downloading all of that from the cloud – which could take several days over a slow home network.
This time, however, something went wrong (I think there was an error in how the “Selective sync” was switched on at certain point), leading into a situation where all the existing folders were renamed by the system as server’s “Conflicting Copy”, then copied into the Dropbox cloud (including c. 330 000 files), while exactly same files and folders were also downloaded back from the cloud into exact same folders, without the “Conflicting Copy” marking. And of course I was away from the machine at this point, so when I realised what was going on, I had to kill Dropbox, and start manually bringing back the Dropbox to the state it was before this mess. It should be noted that there was also a “Rewind Dropbox” feature in this Dropbox Plus account (which is exactly designed for rolling back in this kind of large situations). But I was no longer sure into which point in time I should rewind back to, so I ended up going through about 100 different cases of conflicting copies, and also trying to manually recover various shared project folders that had become dis-joined in this same process. (Btw, apologies to any of my colleagues who got some weird notifications from these project shares during this weekend.)
After spending most of one night doing this, I tried to set up my other old services into the new Mint server installation in the following day. I started from Plex, which is a media server and client software/service system that I use e.g. to stream our family video clips from the server into our smart television. There is an entire, 2600 word essay on Linux file and folder permissions at the Plex site (see: https://support.plex.tv/articles/200288596-linux-permissions-guide/). But in the end I just had to lift my hands up. There is something in the way system sees (or: doesn’t see) the data that is in the old 4 TB disk, and all my tricks with different users and permission settings that I tried, do not allow Plex to see any of that data from that disk. I tested that if I copy the files into that small system disk (the 120 GB SSD), then the server can see and stream them normally. Maybe I will at some point get another large hard drive, try setting up that one under the current OS and user, copy all data there, and then try to reinstall and run Plex again. Meanwhile, I just have to say that I have got my share of tweakability for some time now. I think that Linux Mint in itself is indeed perfectly nice and capable operating system. It is just that software such as Dropbox or Plex do not play so nicely and reliably together with it. Not at least with the tweaking skills that I possess. (While I am writing this, there are currently still over 283 500 files that Dropbox client should restore from the cloud into that problematic data drive. And the program keeps on crashing every few hours…)
My first book published in English was outcome of my PhD work conducted in late 1990s – The Demonic Texts and Textual Demons (Tampere University Press, 1999). As the subtitle hints (“The Demonic Tradition, the Self, and Popular Fiction”), this work was both a historically oriented inquiry into the demonic tradition across centuries, and an attempt to recast certain poststructuralist questions about textuality in terms of agency, or “Self”.
The methodological and theoretical subtext of this book was focused on politically-committed cultural studies on the one hand: I was reading texts like horror movies, classical tragedies, science fiction, The Bible, and Rushdie’s The Satanic Verses from perspectives opened up by our bodily and situated existence, suffering, and possibilities for empowerment. On the other hand, I was also interested in both participating and ‘deconstructing’ some of the theoretical contributions that the humanities – literary and art studies particularly – had made to scholarship during the 20th century. In a manner, I was turning “demonic possession” as a self-contradictory and polyphonic image of poststructuralism itself: the pursuit of overtly convoluted theoretical discourses (that both reveal and hide the actual intellectual contributions at the same time) particularly both fascinated and irritated me. The vampires, zombies and cyborgs were my tools for opening the black boxes in the charnel houses of twisted “high theory” (afflicted by a syndrome that I called ‘cognitocentrism’ – the desire to hide the desiring body and situatedness of the theorizing self from true commitment and responsibility in the actual world of people).
I have now produced a new version of this book online, as Open Access. After the recent merger of universities, the Tampere University Press (TUP) books are no longer available as physical copies, and all rights of the works have returned to the authors (see this notice). Since I also undertook considerable detective work at the time to secure the image rights (e.g. by writing to Vatican Libraries, and Warner Brothers), I have now also restored all images – or as close versions of the originals as I could find.
I hope that the new version will find a few new readers to this early work. Here are a couple of words from my Lectio Praecursoria, delivered in the doctoral defense at 29th March, 1999:
… It is my view, that the vast majority of contemporary demonic texts are created and consumed because of the anxiety evoked by such flattening and gradual loss of meaningful differences. When everything is the same, nothing really matters. Demons face us with visions which make indifference impossible. A cultural critic should also be able to make distinctions. The ability to distinguish different audiences is important as it makes us aware how radically polyphonic people’s interpretations really can be. We may live in the same world, but we do not necessarily share the same reality. As the demonic texts strain the most sensitive of cultural division lines, they highlight and emphasise such differences. Two extreme forms of reactions appear as particularly problematic in this context: the univocal and one-dimensional rejection or denial of the demonic mode of expression, and, on the other hand, the univocal and uncritical endorsement of this area. If a critical voice has a task to do here, it is in creating dialogue, in unlocking the black-and-white positions, and in pointing out that the demonic, if properly understood, is never any single thing, but a dynamic and polyphonic field of both destructive and creative impulses.
I made a significant upgrade to my main gaming and home workstation in Christmas 2015. That setup is soon thus four years old, and there are certainly some areas where the age is starting to show. The new generations of processors, system memory chips and particularly the graphics adapters are all significantly faster and more capable these days. For example, my GeForce GTX 970 card is now two generations behind the current state-of-the-art graphics adapters; NVIDIA’s current RTX cards are based on the new “Turing” architecture that is e.g. capable of much more advanced ray tracing calculations than the previous generations of consumer graphics cards. What this means in practice is that rather than just applying pre-generated textures to different objects and parts of the simulated scenery, the ray tracing graphics attempts to simulate how actual rays of light would bounce and create shadows and reflections in this virtual scene. Doing this kind of calculations in real-time for millions of light rays in an action-filled game scene is an extremely computationally intensive thing, and the new cards are packed with billions of transistors, in multiple specialised processor cores. You can have a closer look at this technology, with some video samples e.g. from here: https://www.digitaltrends.com/computing/what-is-ray-tracing/ .
I will probably update my graphics card, but only a little later. I am not a great fan of 3D action games to start with, and my home computing bottlenecks are increasingly in other areas. I have been actively pursuing my photography hobby, and with the new mirrorless camera (EOS M50) moving to using the full potentials of RAW file formats and Adobe Lightroom post-processing. With photo collection sizes growing into multiple hundreds of thousands, and the file size of each RAW photo (and it’s various-resolution previews) growing larger, it is the disk, memory and speed of reading and writing all that information that matters most now.
The small update that I made this summer was focused on speeding up the entire system, and the disk I/O in particular. I got Samsung 970 EVO Plus NVMe M.2 SSD (1 Tb size) as the new system disk (for more info, see here: https://www.samsung.com/semiconductor/minisite/ssd/product/consumer/970evoplus/). The interesting part here is that “NVMe” technology. That stands for “Non-Volatile Memory” express interface for solid stage memory devices like SSD disks. This new NVMe disk looks though nothing like my old hard drives: the entire terabyte-size disk is physically just a small add-on circuit board, which fits into the tiny M.2 connector in the motherboard (technically via a PCI Express 3.0 interface). The entire complex of physical and logical interface and connector standards involved here is frankly a pretty terrible mess to figure out, but I was just happy to notice that the ASUS motherboard (Z170-P) which I had bought in December 2015 was future-proof enough to come with a M.2 connector which supports “x4 PCI Express 3.0 bandwidth”, which is apparently another way of saying that it has NVMe support.
I was actually a bit nervous when I proceeded to install the Samsung 970 EVO Plus NVMe into the M.2 slot. At first I updated the motherboard firmware to the latest version, then unplugged and opened the PC. The physical installation of the tiny M.2 chip actually became one of the trickiest parts of the entire operation. The tiny slot is in an awkward, tight spot in the motherboard, so I had to remove some cables and the graphics card just to get my hands into it. And the single screw that is needed to fix the chip in place is not one of the regular screws that are used for computer case installations. Instead, this is a tiny “micro-screw” which is very hard to find. Luckily I finally located my original Z170-P sales box, and there it was: the small plastic pack with a tiny mounting bolt and the microscopic screw. I had kept the box in my storage shelves all these years, without even noticing the small plastic bag and tiny screws in the first place (I read from the Internet that there are plenty of others who have thrown the screw away with the packaging, and then later been forced to order a replacement from ASUS).
There are some settings that are needed to set up in BIOS to get the NVMe drive running. I’ll copy the steps that I followed below, in case they are useful for some others (please follow them only with your own risk – and, btw, you need to start by creating the Windows 10 installation USB media from the Microsoft site, and by pluggin that in before trying to reboot and enter the BIOS settings):
In your bios in Advanced Setup. Click the Advanced tab then, PCH Storage Configuration
Verify SATA controller is set to – Enabled
Set SATA Mode to – RAID
Go back one screen then, select Onboard Device Configuration.
Set SATA Mode Configuration to – SATA Express
Go back one screen. Click on the Boot tab then, scroll down the page to CSM. Click on it to go to next screen.
Set Launch CSM to – Disabled
Set Boot Device Control to – UEFI only
Boot from Network devices can be anything.
Set Boot from Storage Devices to – UEFI only
Set Boot from PCI-E PCI Expansion Devices to – UEFI only
Go back one screen. Click on Secure Boot to go to next screen.
Set Secure Boot state to – Disabled
Set OS Type to – Windows UEFI mode
Though, in my case if you put “Launch CSM” to “Disabled”, then the following settings in that section actually vanish from the BIOS interface. Your mileage may vary? I just backspaced at that point, made the next steps first, then made the “Launch CSM” disable step, and then proceeded further.
Another interesting part is how to partition and format the SSD and other disks in one’s system. There are plenty of websites and discussions around related to this. I noticed that Windows 10 will place some partitions to other (not so fast) disks if those are physically connected during the first installation round. So, it took me a few Windows re-installations to actually get the boot order, partitions and disks organised to my liking. But when everything was finally set up and running, the benchmark reported that my workstation speed had been upgraded the “UFO” level, so I suppose everything was worth it, in the end.
Part of the quiet and snappy, effective performance of my system after this installation can of course be just due to the clean Windows installation in itself. Four years of use with all kinds of software and driver installations can clutter the system so that it does not run reliably or smoothly, regardless of the underlying hardware. I also took the opportunity to physically clean the PC inside-out thoroughly, fix all loose and rattling components, organise cables neatly, etc. After closing the covers, setting the PC case back to its place, and plugging in a sharp, 4K monitor and a new keyboard (Logitech K470 this time), and installing just a few essential pieces of software, it was pleasure to notice how fast everything now starts and responds, and how cool the entire PC is running according to the system temperature sensor data.
I have been using the new Canon EOS M50 mirrorless system camera now for a month or so. The main experiences are pretty positive, but I have also some comments on what this camera is good and not so optimal for.
In terms of image quality and feature set, this is a pretty complete package. Canon can make good cameras. However, the small physical size of this camera is perhaps its most defining main characteristic. This means that M50 is excellent as a light and small travel companion, but also that it has too small grip to carry comfortably this body when there are some heavy “pro” lenses or telephoto lenses attached. One must carry the system from the lens instead.
I really like the touch screen interface of M50. The swiveling LCD is really functional, and it is easy to take that quick photo from extra low or high angles. The LCD touch interface Canon uses is perhaps the best in the market today: it is responsive, well designed and logically organised. This is particularly important for M50, since it has only few physical buttons, and a single rotating control. Photographer using M50 needs to use the touch UI for many key functions. This is perhaps something that many manual-settings oriented professional and enthusiast photographers do not like; it you like to set the aperture, exposure time and ISO from the physical controls, then M50 is not for you (one should consider e.g. Fujifilm X-T3 or T30 instead). But if one is comfortable working with electronic controls, then M50 provides multiple opportunities.
My old EOS camera had only few (nine) autofocus points (phase-detect), and only the single point in the middle was with the fast, cross-type AF. This M50 has 99 selectable AF points (143 with some lenses), covering 80 % of the sensor area (dual-pixel type). Coupled with the touch screen, this change has had an effect on my photography style. It is now possible to first compose the photo, look through the electronic viewfinder, and simultaneously use a thumb to drag the AF point/area (in a “computer mouse/touchpad style”) to the desired point in the screen. I am not completely fluent in this technique yet though, and my usual technique of center focusing first, then half-pressing to lock the focus, and then quickly making the final composition, and shooting, is perhaps in most situations quicker and simpler than moving the focus point around the screen. But since M50 remembers in Program mode (which I use most) where the AF point was left the last time, the center focusing method does not work properly any more. I just need to learn new tricks, and keep moving the AF points in the screen (or, let the camera do everything, in Full Auto mode, or go into Manual mode, and do focusing with the lens ring instead.
As a modern mirrorless camera, M50 is packed with sensors and comes with a powerful DIGIG8 processor, bright LCD screen and electronic viewfinder. All of this consumes electricity, and the battery life of M50 is nowhere near my old 550D (which, btw, also had an extra battery grip). A full day of shooting takes either two or three fully loaded LP-E12 batteries. Thus, this camera behaves like a smartphone with poor battery life. You need to be using that battery charger all the time. (The standard rating is 235 shots-per-charge, CIPA.)
When travelling, I have been using a lot the wireless capabilities of M50. It is really handy that one can move full resolution, or reduced resolution versions of photos into an iPhone, iPad or Android device while on the go. On the other hand, this is nowhere as easy as when shooting and sharing directly from a smartphone. Moving typical 200-300 photos from a shooting session into an iPad for editing and uploading is slow, and feels like it takes ages. (I have not yet cracked how to get the advertised real-time Bluetooth photo transfer to work.) The traditional workflow where the entire memory card is first read into a PC and processed with Lightroom makes still better sense, but it is nice to have the alternative, for mobile processing and sharing some individual photos at least.
Many reviewers of M50 have written a lot about the limitations of 4K video mode (high crop factor, no dual-pixel autofocus). I use video rarely, and then only full HD, so that is not an issue for me. There is an external microphone input, which might be handy, and the LCD screen can be turned to point forward, if I ever go into video blogging (not that I plan to do it).
The main plusses for me in M50 are the compact size, the excellent touch UI, and very nice image quality in still images. That I can use both the new, compact EF-M mount lenses, and (with adapters) also the traditional Canon EF lenses was a major factor when making the purchase decision, since the lens collection of a photographer is typically much more expensive part of the equipment, than the body only. Changing to Nikon, Fuji or Sony would have been a big investement.
The autofocus system in M50 is fast, and in burst mode the camera can shoot 10 fps for 30 jpg shots in a row to fill the buffer. I am not a sports or wildlife photographer as such, so this is good enough for me. A physically bigger body would make the camera easier to handle with large and heavy lenses, but shooting with a large lens is a two-hand operation in any case (and in some cases requires using a tripod), so that is not so critical. I still need to train more to use the controls and switch between camera modes faster, and touch interface is probably never going to be as fast as using a camera with several dedicated physical controls. But this is a compromise one can make, to get this feature set, image quality and lens compatibility in this small package, in this price.
It has been interesting to follow how since last year, there has been several articles published that discuss the “mirrorless camera hype”, and put forward various kinds of criticism of either this technology, or related camera industry strategies. One repeated criticism is rooted to the fact that many professional (and enthusiast) photographers still find a typical DSLR camera body to work better for their needs than a mirrorless one. There are at least three main differences: a mirrorless interchangeable camera body is typically smaller than a DSLR, the battery life is weaker, and the image from an electronic viewfinder and/or LCD back screen offers a less realistic image than a traditional optical viewfinder in a (D)SLR camera.
The industry critiques appear to be focused on worries that as the digital camera market as a whole is going down, the big companies like Canon and Nikon are directing their product development resources for putting out mirrorless camera bodies with new lens mounts, and new lenses for these systems, rather than evolving their existing product lines in DSLR markets. Many seem to think that this is bad business sense, since large populations of professionals and photography enthusiasts are deeply invested in these more traditional ecosystems, and lack of progress in them means that there is not enough incentive to upgrade and invest, for all of those who remain in those parts of the market.
There might be some truth in both lines of argumentation – yet, they are also not the whole story. It is true that Sony, with their α7, α7R and α7S lines of cameras have stolen much of the momentum that could had been strong for Canon and Nikon, if they would had invested into mirrorless technologies earlier. Currently, the full frame systems like Canon EOS R, or Nikon Z6 & Z7, are apparently not selling very strongly. In early May of this year, for example, it was publicised how Sony α7 III sold more units in Japan at least than the Canon and Nikon full frame mirrorless systems combined (see: https://www.dpreview.com/news/3587145682/sony-a7-iii-sales-beat-combined-efforts-of-canon-and-nikon-in-japan ). Some are ready to declare Canon and Nikon’s efforts as dead on arrival, but both companies have claimed to be strategically committed into their new mirrorless systems, developing and launching lenses that are necessary for their future growth. Overall though, both Canon and Nikon are producing and selling much more digital cameras than Sony, even while their sales numbers have been declining (in Japan at least, Fujifilm was interestingly the big winner in year-over-year analysis; see: https://www.canonrumors.com/latest-sales-data-shows-canon-maintains-big-marketshare-lead-in-japan-for-the-year/ ).
From a photographer perspective, the first mentioned concerns might be the more crucial than the business ones, though. Are mirrorless cameras actually worse than comparable DSLR cameras?
There is the curious quality when you move from a large (D)SLR body into using a typical mirrorless: the small camera can feel a bit like a toy, the handling is different, and using the electronic viewfinder and LCD screen can produce flashbacks of compact, point-and-shoot cameras of earlier years. In terms of pure image quality and feature sets, the mirrorless cameras are already equals to DSLRs, and in some areas have arguably moved already beyond most of them. There are multiple reasons for this, and the primary relates to the intimate link there is between the light sensor, image processor and viewfinder in mirrorless cameras. As a photographer you are not looking at a reflection of light coming from the lens through an alternative route into the optical viewfinder – you are looking at the image that is produced from the actual, real-time data that the sensor and image processor are “seeing”. The mechanical construction of mirrorless cameras can be made simpler, and when the mirror is removed, the entire lens system can be moved closer to the image sensor – something that is technically called shorter flange distance. This should allow engineers to design lenses for mirrorless systems that have a large aperture and fast focusing capabilities (you can check out a video, where a Nikon lens engineer explains how this works here: https://www.youtube.com/watch?v=LxT17A40d50 ). The physical dimensions of the camera body in itself can be made small or large, as desired. Nikon Z series cameras are rather sizable, with a conventional “pro camera” style grip (handle); my Canon EOS M50 is diminutive, from the other extreme.
I think that the development of cameras with ever more stronger processors and their machine learning and algorithm-based novel capabilities will push the general direction of photography technology towards various mirrorless systems. Said that, I completely understand the benefits of more traditional DSLRs and why they might feel superior for many photographers at the moment. There has been some rumours (in the Canon space at least, which I am personally mostly following) that new DSLR camera bodies will be released into the upper-enthusiast APS-C / semi-professional DSLR category (search e.g. for “EOS 90D” rumours), so I think that DSLR cameras are by no means dead. There are many ways in which the latest camera technologies can be implemented into mirror-bodies, as well as into the mirrorless ones. The big strategic question of course is that how many different mount and lens ecosystems can be maintained and developed simultaneously. If some of the current mounts will stop getting lenses in the near future, there is at least a market for adapter manufacturers.