Sometimes you see a photo accessory and wonder, “where the hell were you all this time.” And the answer is, “it was too easy, so Sony canned it.” The GPS-CS3KA (“GPSman?”) is a smallish box, maybe two-thirds the size of a Metz 26AF flash. It only really does two things: (1) keeping a track log from GPS signals it receives and (2) writing them to the JPGs on your SD card.
A reasonable solution to a stupidly common problem?
Wait? What? Most GPS solutions for cameras have been pretty terrible. For reasons that are unclear (perhaps metal covers), high-end cameras have not had built-in GPS. In fact, few cameras period have it – aside from the ubiquitous iPhone or Android. This leaves you with some suboptimal options:
- Keep a tracklog with a separate device (GPS watch, tracklogger, battery-intensive phone app) and marry the coordinates to the files in Lightroom or Exiftool.
- Use a separate device with Bluetooth to feed coordinates into your camera’s remote port (a la Red Hen).
- Use a clunky GPS add-on that takes up both your remote terminal and hot shoe (looking at you, Canon and Nikon).
- Try to graft an NMEA cable to your DSLR’s accessory port.
- Use a clunky grip with GPS built-in (Leica Multifunction Grip M)
- Stick a GPS in some other accessory, like an EVF that you might otherwise not user (Leica EVF-3).
Sony quite possibly solved this problem by accident with the GPS-CS3KA, which takes a reading every 15 seconds into 128mb of memory – and when you insert an SD card will look for the closest matches and tag your JPGs in batches of 60. I say “by accident” because operation is far to simple for a Sony (at least compared to a Bravia TV). There are only three options:
- GPS: display GPS screen – hitting enter gives you different permutations of time and GPS coordinates.
- Match: automatically counts the number of files to be tagged and only lets you start or cancel. Matching stops the GPS reception.
- Tools: set the time zone, undo-ability, and erase internal memory.
How does it work?
- Stick a single AA battery in one slot.
- Set your correct GPS plus or minus time zone (as I write this, -400 for Eastern).
- Turn on the machine.
- Shoot a bunch of pictures.
- Put your SD card in the slot.
- Use the “matching” function to assign locations (use “undo” to clear all of the data you just wrote).
- Repeat as many times as necessary in batches of 60 files.
Note that when you initiate a card matching session, you may lose the GPS signal – but then again, you won’t be shooting pictures while your card is in the device.
GPS performance is actually quite good. Cold start will grab coordinates within about a minute; on a warm start, about 10 seconds. Your initial startup will be minutes as the device updates its GPS satellites. The device apparently can read a signal in many indoor settings, which is neat. Or scary.
My performance tests on accuracy landed this within about 15 feet of where I was standing. It does read out in minutes and seconds too. For most purposes, it suffices to see degrees to know that it’s locked on.
Observed battery life with alkaline was about 12 hours. Not terrible, considering how much power this probably draws.
I did not test the Sony software, but I did note that connecting the USB cable does not bring this up as a drive with an easy-to-access GPX log.
Where does it work and not work?
I tried this a Sony A7rii and with cards up to 64gb. The results were better than expected for a device this old.
Cards that work: up to 32gb only, the faster the card, the better (realistically, that’s a Sandisk 95mb/sec card).
To be safe, I would recommend using SDFormat and not opening cards with files on a Mac before encoding. Macs tend to throw indexing files on disks that are invisible to the user but can hang up particularly primitive embedded devices (of which you should assume this is one).
Cards that don’t: 64gb and up; WiFi-enabled cards. I suspect that 64gb is outside of the ability of the device to read cards (even devices that read FAT32 sometimes cannot address an entire card). You get “matching error” as your only clue. As to WiFi, my best guess is that since it works for a couple of frames and then blanks, that the card sees that x files have been read and that it’s time to turn on the WiFi. The problem is that one AA battery doesn’t have enough power to allow that. In my testing, there has been no way to shut off the FlashAir’s desire to start transmitting (unlike EyeFi, which could be set to transmit only images that were write-protected).
Files that get encoded: the spot of bad news is that the current ARW raw format doesn’t get location data with the Sony GPS. But since the device will record location data onto almost any JPG, it will work equally well (or poorly) with many types of cameras.
Within the limits of a certain card size, and therefore speed, the Sony GPS does allow a relatively automated geotagging process for JPGs. Like Lex Luthor’s henchmen, it has “one job.” But unlike those people who never succeded at killing Superman, the Sony performs that job well.
Notably, you can generate tracking data usable with multiple cameras, since you can insert SD card after SD card and use the same body of GPS data to code files shot in the same time period. This is a bit more flexible than solutions that would have to be transferred from camera to camera (or just duplicated with good old cash). It does require than your cameras’ clocks be synchronized reasonably closely.
It does not solve the problem of writing geolocation data to RAW files (Lightroom, for example, simply ignores this data if you import both tthe RAW and the JPG), and no one will likely ever solve the mystery of why cameras don’t have inbuilt GPS. But it’s a lot better than trying to marry track logs and files by manual labor.
Who says Lego bricks are only good for causing foot and back injuries?
Welcome to the world’s crudest 3D camera: four Duplo bricks, two DxO One cameras, and about half a meter of packing tape. With a stereo separation of about 120mm, forget about taking pictures of anything closer than 15 feet. But oh, the scary places you will go.
Surprisingly, with the OLED-frame-assist function on, the cameras don’t have much trouble focusing on exactly the same subject, which solves one weird technical hangup.
People understand why tilt lenses exist – making super-expensive Canon DSLRs produce pictures that look like they were taken with a toy camera (or making the subjects themselves look like toys). No one knows, though, why shift lenses were once a thing. It’s all a matter of perspective.
The truth, from a certain point of view
Photography always has (and always will) present this problem: needing to fit a large object into a frame that is constrained by lens focal length. Conceivably, with a superwide lens you could, but then you end up with a lot of extra dead space in the frame. Which defeats the purpose of using large film or sensors.
If you want to get the whole thing in frame with the minimum number of steps or expenditure of time and money, your choices are to use a really wide-angle lens, tilt a camera with a more moderate wide-angle up, learn to fly. All of these are sub-optimal. First, the really wide-angle lens is great in that you can capture the top of the object without tilting the camera. The problem is that making an engaging photo with a wideangle is actually extremely difficult – because it tends to shrink everything. Depending on how the sun is, it also stands a better chance of capturing the photographer’s shadow. Second, tilting up a camera with a more moderate wide-angle lens “up” turns rectangular buildings into trapezoids, which works for some pictures but definitely not others. Finally, learning to fly is difficult. But watch enough Pink Floyd concert films, toke up with the ghost of Tom Petty, or study Keith Moon’s hotel swims, and you might.
Do you skew too?
Assuming you are reasonably competent, you can correct perspective using software, by skewing the canvas. This is a take on the old practice of tilting the paper easel with an enlarger. This was a limited-use technique, generally practiced by people who could not use view cameras and tripods but still had to come up with a presentable representation of a tall object. There were (and substantially still are) three issues here: crop, depth of focus, and dis-proportion. First, the crop came from the fact that tilting an easel meant that the projected image was trapezoidal and not rectangular, meaning that from the get-go, it had to be enlarged until the paper was filed. This still happens with digital. Second, the depth of focus issue is related to the fact that enlarging lenses are designed to project to a surface that is a uniform distance from the enlarger (i.e., projecting one flat field onto another). You would have to stop down the lens severely, or use a bigger focal length, which in turn required a taller enlarger column to maintain the same magnification.
The digitization of perspective correction uses computation to project the flat image onto a skewed plane, using interpolation and unsharp masking. This solves the apparent sharpness issue, but it degrades quality. Finally, dis-proportion comes from the fact that straightening converging verticals starts from a place where certain details are already compressed via the original perspective. For example, looking up at a tall building from a short distance, the windows look shorter (top to bottom) than they would if you were looking straight at the window from its own level.
So even when you manage to re-skew the canvas/field/whatever, you now have an image that is too “fat.” On enlarging paper, you would be forced to make a cylindrical correction to the negative (which is not practical in real life). On digital, there are specific transformations that you can perform to correct (for example, the adjustable ratios on DxO Perspective and Lightroom.
So skewing is a useful technique, but it’s still better to skew less.
Shifting your thinking: the mirror years
View cameras have used the concept of shift and tilt to adjust for situations where the viewpoint was wrong (shift) or depth of field was insufficient (tilt). Raising the front standard of a bellows-type plate camera was always standard practice to improve photographs of tall objects, especially in an era where wideangle lenses were not super-wide by today’s standards. Lens board movements were easy to achieve because there was always some distance between the lens mount and film plane in which to insert a mechanism to raise the lens relative to the film. And because there is no control linkage between the lens/shutter and the rest of the camera, you’re not losing automation. You never had any!
But these cameras were not small. The smallest bellows-type camera with lens movement features was the Graflex Century Graphic, a delightful 6×9 press-style camera. On many bellows-type cameras, though, there was no real provision for using a shifting viewfinder. The press-style cameras had wire-frame finders that provided a rough guide, but nothing could tell you whether the lens was actually level outside a gridded ground glass. Later in the game, the Silvestri H would present as the first camera with automatic finder shift, as well as a visible bubble level. Linhof used a permanently-shifted lens assembly (and viewfinder) on the Technorama PC series, and Horseman provided shifted viewfinder masks for the SW612P, though these were available only as “all the way up/down” or “all the way left/right.”
The shift mechanism, though, could not be adapted to SLRs easily due to three constraints:
- Most SLRs lenses are retrofocal – meaning that the nodal point of the lens is more than the stated focal length from the imaging plane. It takes a ton of retrofocus to insert a shift mechanism into an interchangeable lens that has to focus past a mirror box. More retrofocus means bigger lenses So when perspective control lenses began to appear for SLRs (35mm and 6×6), they were huge. Maybe not huge by today’s standards, but a 72mm filter size is pretty big for a Nikon SLR whose normal filter size is 52mm.
- To achieve an image circle large enough to allow shift around what is normally a 24x36mm image circle, it is necessary to use a wide field lens and stop it down severely (illumination with almost any lens becomes more uniform as it is stopped down).
- Most cameras can only meter PC lenses correctly in their center position, wide-open. Where shift mechanisms eliminate direct aperture linkages to the camera, you’re back to the 1950s in metering and focusing – then shifting – then manually stopping down to shoot (now corrected by the use of electronic aperture units in $2K plus modern Nikon and Canon PC lenses).
Viewing is not a lot of fun with 35mm SLRs; when stopped down, PC lenses black out focusing aids (like split prisms and microprisms) and still require careful framing to keep parallel lines parallel. So you need a bright screen – plus a grid or electronic level. Suffice it to say, a lot of people regard perspective control to be a deliberative, on-tripod exercise when it comes to SLRs and DSLRs. Maybe it’s not.
A new perspective: full frame mirrorless?
So here come mirrorless cameras (well, they came a while ago). Now you can fit any lens ever made to any mirrorless body. The optical results may vary, but at least physically, they fit.
— Getting the lens in place
So I grabbed the nearest available PC lens I could find, which was a 28/3.5 PC-Nikkor. Not AI, not even from this century. Released in 1980, it is a beast. I plugged this into a Konica AR body to Nikon lens adapter, and from there into a Imagist Konica lens to Leica body adapter. Why all these kludgy adapters? The answer is actually pretty simple: the Imagist has the correct tolerance to make infinity infinity, and the Konica adapter does the same. This is not a small consideration where you might be zone focusing a lens.
Then I plugged this kludgefest into a Leica M typ 246 (the Monochrom). Because why not start with the OG of mirrorless camera platforms? Of course, you can’t use a rangefinder with a Nikon SLR lens, so I plugged in an Olympus EVF-2 (which is the ‘generic’ version of the Leica EVF-2.
— Getting it to work
The Nikkor has two aperture rings. One is the preset, where you set your target aperture. The other is the open/close ring, which goes from wide-open to where the preset ring is set.
I turned on focus peaking and set the preset for f/22 and the open/close for f/3.5. I was able to establish that infinity was correct.
Next, I stopped down the lens (both rings to f/22), expecting that just as on an SLR, the EVF would black out. Worked perfectly.
I hit the “info” button to get the digital level, and it was off to the races. The lens has a rotation and a shift.
— But how well does it actually work?
The functionality is actually surprisingly good. On a Leica, it’s just stick the camera in A, stop the lens down to f/16 and 22, and point and shoot.
The digital level obviates the use of a tripod or a grid focusing screen, and you really just frame, turn the shift knob until the perspective looks right, and there you go. There are a couple of limits
You can’t use maximum shift along the long side of the film, but the only penalty is a little bit of a tiny shadow in the corner. And that’s with a full-thickness 72mm B+W contrast filter. You get 11mm shift up and down (i.e., along the short dimension of the firm) and 8mm left and right (nominally; as I stated, you can get away with more under some circumstances).
Aside from that, there are some minor annoyances like making sure you haven’t knocked the aperture ring off the shooting aperture. Or knocking the focus out of position (it’s a very short throw…).
BUT THE DUST! And here is the rub – shooting at f/16 and f/22 brings out every dust spot on your lens. Normally, you would shoot a Leica M at f/5.6, f/8 max. But PC lenses – like their medium and large format cousins – are designed to max out their frame coverage at very small openings. So I had never cleaned the sensor on my M246 in four years, and I got to spend an evening working on a hateful task that included swabs and ethanol and bulbs and the Ricoh orange lollipop sensor cleaner.
— And how sharp?
Very. Diffraction is supposed to start becoming visible at f/11 on this combination at 1:1, with it showing up in prints at f/22.
Pictures stand up to the old 1:1 test, except in the corners where you have over-shifted along the long side. Recall that in lot of situations, two of the last bits of corner are usually sky, where a tiny amount of blur is not going to be of any moment.
How well this will work on a color-capable camera is a question, especially since lateral color would come out. But right now, this is posing the most acute threat to 6×4.5 cameras loaded with TMY.
My first DxO One (version 1, $125 new on clearance) bricked when I upgraded the firmware. Left with an inert toy while Amazon dug up another one to send me, I could not help but play with the dead one. I flew it up to the water/ice dispenser on the refrigerator. “Open the pod bay doors, HAL.” Nothing. The DxO One rotated 180 degrees so that it could eject the micro SD card into the…
“Dad, what are you doing?”
But seriously, the DxO One is one of strangest and most wonderful cameras to come out of France, or anywhere. Here’s why.
Sensor. The camera uses a 20Mp, 1″ Backside Illuminated (BSI) sensor (3x or so crop factor) made by Sony, the same one as on the RX100III. Two things make this a standout here: first, BSI sensors are quite good – meaning this returns results almost on par with the Sony a6300’s copper-wire conventional sensor. Second, almost all sensors perform equally at base ISO. In the software design, DxO biases the camera toward lower ISOs and wider apertures (which makes sense, since a 1″ sensor starts diffracting at f/5.6).
How does this compare to an iPhone XS sensor? Well, it’s almost 70% more resolution and 6.7x times the surface area (116mm² x 17.30mm²). Do the math. All the computations in the Apple world can’t make up for this type of difference in displacement. This does expose the genius of portrait mode, though – because not even a 1″ sensor is big enough to have easy-to-achieve subject isolation.
The sensor is used for contrast-detect AF (with face priority).
Lens. 32mm equivalent, f/1.8-11 aperture, six groups, six elements, with some of the weirdest aspherical shapes imaginable. It’s very tough to find a lens on a compact camera that approximates a 35/1.8. But here you are.
Far from being telecentric with an expected “folded optics” path, the DxO One uses the cellphone method with almost zero distance between the rearmost element and the sensor. The rearmost element looks like a brassiere. Like this:
The lens is happiest at larger apertures (f/2-f/4).
Storage. The DxO One accepts standard MicroSD cards. I was able to test up to 128Gb cards (Samsung EVO Plus), and it is able to read and write to them with no issues.
Power. Power comes from an internal battery but can also be fed directly from a micro USB cable. The battery takes about two hours to charge and does about 200 shots. Version 2 of the camera has a removable back door to accommodate an external battery pack DxO no longer sells. You also lose the free software (see below).
Viewfinder. Your choice of two. You can plug the camera into your iPhone, where you can use the DxO One application and the phone screen as a viewfinder. Alternatively, version 3.3 of the camera firmware turns the little OLED screen on the back into a square contour viewfinder, good enough at least to frame the middle square of the picture – and surprisingly good at estimating a level angle for the camera. You could also split the difference with a Lightning extension cord.
Connectivity. The camera was originally designed to connect via the Lightning port, but DxO enabled the onboard WiFi so that now you can use the application on the phone and control the camera (including view-finding) without a physical connection. The DxO One can also connect to your phone via your home wireless network. WiFi operation – no matter what the camera or phone – is not as much fun as it first sounds – which is why the DxO product is more flexible than Sony’s wireless-only solutions.
Software. In terms of the camera’s software, all the magic is under the hood. The camera switches on by sliding open the front cover (slide it all the way, and the Lightning connector will erect itself). There is a two-stage shutter button on the top and you can swipe up and down on the OLED to switch between controls and viewfinder and left and right to toggle photo and video. The camera stays on the exposure mode last selected from the DxO software on the iPhone.
The DxO One phone app is well-done and responsive. You can use it to frame, shoot the picture, and control what you want. Features include:
- JPG, Raw, and Super Raw (stacked) exposure modes.
- Single-shot, timer, and time-lapse settings
- Flash settings
- Subject modes and the usual PSAM modes.
- Program shift (between equivalent exposures with different shutter speeds or apertures).
- Single AF, Continuous AF, On-Demand AF, and Manual focus (manual includes an automatic hyperfocal calculation if desired).
- Matrix, centerweighted, or spot metering.
- Grid compositional overlay.
- “Lighting,” which is like a mini HDR compressor for JPGs.
You can also look through the exposures on the camera/card and move them to your phone as desired. As noted above, though, you do need to initiate wireless connections with the camera connected.
If you get a version 1 camera, new, it also comes with DxO Optics Pro 10 Elite (now Photo Lab 1 Elite) and DxO Filmpack Elite. But you have to be able to document that you are the original owner of the camera. Both of these can run as standalones or can be external editors for Lightroom. Photo Lab 1 is also capable of replacing Lightroom.
If you get version 2, you’re out of luck. But you do get a 4gb SD card and the detachable back door for that battery pack.
And either way, you do get DxO OpticsPro 10 for DxO One, which gives you a nice imaging/digital asset manager that can composite SuperRaw files. SuperRaw is a stack of four successive (and extremely rapid) exposures that cancel out high ISO noise.
And if you don’t like any of that, the DxO One outputs normal DNG files that you can simply edit to taste in Lightroom. There is a Lightroom profile for the camera’s minimal residual distortion.
Ergonomics. This is the one place where things are sketchy. It’s hard to hold onto a small ovoid object, especially one with a button on the top. I would highly recommend a wrist strap.
Upshot. Maybe not the most compelling camera at $700 plus when it came out, but now that it is a sixth of that and still a lot of fun to shoot, go for it!