Notes on OM System OM-1 digital camera (DRAFT)

Quite a lot has been already written in the internet about the OM System’s OM-1 camera, but I write down here some notes comparing it to the E-M1 Mk II and highlighting some use cases. I share this notes in case someone else finds them useful, although they are mostly intended for myself. Among the things frequently mentioned is a lack of improvement in image quality between old and new image sensors (e.g., OM system OM-1 vs OM-5 – The 10 Main Differences). This is one example of where my own experience differs drastically from the quick and dirty tests used by some blog-post writers. There are exceptions to such statements, of course, the most notable I have come accross being the thorough and very informative videos at the The Narrowband Channel  with the video OM1 Sensor, Everything You Need to Know being most relevant to the current post.

Image noise and high resolution modes

Image noise is difficult to objectively quantify in ways that are relevant to its “pictorial effect” or human observers’ reaction to it. Furthermore, the noise as seen in the final rendition of a photograph depends heavily on noise reduction during image processing. How effectively a noise reduction algorithm is depends on the characteristics of the noise and of the subject matter depicted by the image. I use mostly Capture One 23 and occasionally OM Workspace. My impressions below, are based on my experince with Capture One 23.

It is important to distinguish random noise (or thermal noise that affects each pixel at random in both time and space) and patterns like banding that affect pixels systematically in space and/or time. Noise in film is mostly random occasionaly affected by some clustering, while electronic sensors tend to suffer from unevenly distributed noise, either the result of spatially uneven sensor temperature or systematic variation among pixels, due to inherent light sensitivity or differences in the wiring within the sensor of pixels at different locations in the array or differences in the analogue to digital conversion circuitry of different pixels.

I think different features of image noise annoy different people differently. I find specially disturbing any recognizable pattern in the noise. In this respect the OM-1 performs much much better than the E-M1 (Mk I) or E-M1 Mk II. As a result, the upper ISO limits I am happy to use with the two cameras are 800 to 5000, respectively. This makes Auto ISO my most frequent setting in the OM-1, while I almost never used it in the E-M1 Mk II. In practice these limits allow significant cropping of images and printing to A3 based on my own standards. These self-imposed limits are without use of any AI-based noise reduction in post-processing, with which even higher ISO would become usable in some cases.

That noise is spatially uniform and mostly random in the OM-1 not only makes it less disturbing in the image but also easier to remove without removing features from the depicted object. In Capture One this means that one can use much higher settings for clarity and structure without making visible artifacts such as banding. It also means that any procedure that averages multiple readings from each pixel will decrease noise very effectively. High resolution 80MP images using sensor shift are noiseless or nearly noiseless even at high ISO settings, as the merging of 8 frames cancels most of the random noise. An alternative approach that is also very effective is to take multiple photograph using the same exposure settings and do an HDR merge of them in Capture One, as Capture One will in this case align and average the images. (Of course, averaging is also possible with other software.)

An intriguing predictable consequence of noice being predominantly random, is that it should affect short exposures the most. This could explain some of the contrasting assessments of the OM-1 sensor and is something worthwhile testing as it could be taken advantage of in everyday use of the camera.

Live ND in the OM-1 is based on image averaging, so it should also reduce noise. The usable shutter speeds are, however, limited. I need to test how effective Live ND Shooting is at cancelling out noise in camera.

It is of note that the high resolution mode by sensor shift is usable with any lens, even old manual lenses attached through purely mechanical adapters. Even 50MP handheld images are possible as focal length can be very precisely entered and saved allowing very effective image stabilization. Also Live ND Shooting mode can be used with any lens, including adapted ones.

Digitizing film negatives and positive transparencies using the 80 MPix high resolution mode works extremely well.

Sensor dynamic range

Do OM-1 ORF files converted using default settings have a wider dynamic range? My impression is that they do not, and there is no much reason for camera designers to alter this. The relevant question is if the ORF files from the OM-1 can whistand a wider range of adjustment when they are edited. This is currently only a subjective impression, but I see little change from the E-M1 Mk II in how much recovery is possible in the shades, but I see an improvement in the highlights. Of course whether the extra “room” is in highlights or shadows depends on how one sets the exposure.

Focusing in challenging situations

Birds’ focus tracking works most of the time well. The AI systems locates the area of the frame where the bird is, but focusing itself does not rely on AI, it just focuses on whatever high contrast object is in this area. Sometimes, this is a branch in front of a bird. In such cases it helps to manually pre-focus on the bird.

  • While using subject recognition in a crowded situation, such as a bird with twigs in front, with subject recognition engaged, while pressing the shutter half way down, manual focus until the object to be recognized is partly in focus. After this the recognition algorithm usually finds even small birds as long as the ocluding stuff remains out of focus. This works because if manual focus is enabled during automatic focusing (a setting in the AF menue), manual focus works not only as a follow up to single focus but also allows manual adjustment of focus during continuous focus with tracking with AI subject detection.

When there are many birds in the same image and they move so that recognition fails on the current target, a different bird can become the new target, I think, even with tracking enabled. This rarely happens with birds in flight, but occasionaly with birds on the ground and at small size in the frame.

Setting a small focus area does not limit the tracking of birds in the whole frame but it limits where the target is initially locked to: this is extremely usefull when one want to focus on a specific bird in a flock in flight or on the ground. (I am not sure, but this may be firmware version dependent.)

  • To quickly change the size of the focusing area without taking the eye from the viewfinder, press down the joystik and turn the front wheel.
  • To move the focus area use the joystik or the cursor pad.
  • If you have a button assigned to swithcing on-off the subject recognition, if you keep it pressed or do a long-press, you can change the subject type with the front wheel.

Focus tracking, obviously is not only useful for moving subjects. I usually find it faster to use re-framing after focusing than moving the focusing area with the joystick.

Autofocus in pouring rain to my amazement works. The E-M1 Mk II strugled with auto focus in heavy rain, even for steady subjects (I think the original E-M1 handled this better than the Mk II). The OM-1 seems to be minimally disturbed. Tracking birds in flight with the 300 mm F:4.0 objective (600 mm eq.) in heavy rain works reasonably well with about 1 in 3 or 4 frames with tack sharp focus on the eye.

Focus bracketing

With faster sequential shooting capability, focus bracketing is a lot faster in the OM-1 than in the E-M1 Mk II. With the camera on a tripod or copy stand it works perfectly with both slow and fast shutter speeds.

I have not yet used much in-camera focus stacking with close up or macro subjects. In tests it worked well, and has the advantage of getting quickly feedback on whether alignment succeeded or not.

I have not yet used much hand-held focus bracketing followed by off-camera focus-stacking. With Helicon Focus not always (only when shooting handheld at high magnification) alignment may fail at least for deep stacks. In situations like this a macro lens with image stabilization as the recently announced 90 mm 1:3.5 will most likely perform better than the 60 mm 1:2.8 macro.

Stabilization

Stabilization has improved once again, but it is now so effective for lenses like the 300 mm F:4.0 that shutter speed is most of the time constrained by subject movement. A tree shaking in the wind will obviously not be rendered without movement as different parts of the tree will not move together. Similarly if there is no steady reference with high-contrast edges at different angles it will be less effective, than for example with a building or book case in the frame. There are limitations imposed by the nature of subjects, but overall it is the strongest feature of the system: I regularly take photographs handheld with a 300 mm + 2x teleconverter, or 1200 mm-equivalent focal length, even ocasionally birds in flight. Still, the limiting factor for shutter speed is subject movement, not camera movement.

Good stabilization also helps a lot in tele-macro photography, for example of insects. One interesting aspect is that subject tracking works to some extent with un-supported subject types. Hopefully, in the future we will get insect recognition and tracking as a new option. Subject recognition and tracking could be also useful for “static” subjects like flowers, shaking in the wind.

When using adapted objectives, an additional improvement is how settings can be stored with precise focal length (to 0.1 mm). The lens name, focal length and maximum aperture are saved as metadata. This allows effective image stabilization with almost any lens. There is one limitation, when working at high resolution or focusing at close range: the camera will not receive information about focusing distance from an adapted objective. As this information is needed to compensate the stabilization requirement, stabilization becomes less effective. This limitation can to an extent be overcome by setting a longer focal length that the real focal length of the lens when working at close range.

The viewfinder

The higher resolution EVF of the OM-1 helps with manual focusing. However, the higher refresh rate of its EVF is what surprised me most: it makes tracking birds in flight so much easier compared to the E-M1 Mk II. Here I mean it makes a lot easier to keep the flying bird framed and anticipating changes in flight path.

The improved easy of manual focusing is of course hugely useful when using adapted manual lenses.

The new viewfinder makes also much easier to assess depth of field.

Sequential shooting

Fast sequential shooting is much more useful than what I had expected. At least, when photographing birds it is true that one ends with a heavy photo culling task, but one gets regularly one image out many where the position of the bird and the light reflection on its eyes are “perfect”, which, at least for me has been almost impossible to achieve more than by accident with earlier cameras.

Capture One HDR merge mode can be used to average a series of equally exposed images as a way of decreasing image noise effectively. Using the OM-1 sequential shooting mode with its excellent image stabilization one can get a series of frames within a fraction of a second. Even birds standing on the ground or sitting on a perch, frequently remain immobile long enough to get four, five or even 6 frames good for HDR merge within 1/5 s. Capture One aligns the images so a small framing error between images does not cause problems.

With bracketed exposure, Capture One does seem still to use to some extent, possibly weighted, averaging.

The sensor

Most of what I described above depends on the highly improved image sensor. There is still more to the sensor: it is a back-side illuminated (BSI) sensor. This has not only implications for the sensitivity to light but also for its optical properties. According to a video in The Narrowband Channel this makes a huge diffrence with vignetting, decreasing it markedly for wide angle objectives. MFT objectives contain memory and other electronics. The memory is used to store information describing the objective and required corrections, which allows MFT cameras to compensate for the vignetting, while, of course, adapted lenses lack such information. FT and MFT objectives were designed from the ground up to work well with front-side illuminated sensors by limiting the angle of incidence of the light on the sensor. The angle of incidence of light has a smaller effect on the film than on sensors. Additionaly, sensors have a filter stack and behave optically differently to film. A BSI sensor is optically more similar to film than earlier sensors, so  I would expect adapted wide angle lenses designed for film cameras to benefit the most in this respect.

I intend to test how much the BSI sensor improves the performance of adapted objectives with respect to vignetting and possibly also in relation to colour fringing.

Tethering and communication

Tethering through USB works reliably even with relatively long cables, it does not require USB-C port or anything faster than USB 3.0, even when using bracketing or the 80 MPix high resolution mode and trnaferring the raw files in real time to a PC. Charging/powering simultaneously with tethering does require a USB-C port with power delibery capable of supplying 9V at a minimum of 3 A.

Tethering through Wifi and a LAN connected PC.

Remote control by Bluetooth with a tablet or phone and the app remains similar as with E-M1 cameras. Or rather, the improved features of the app also work with ealier camera models.

Geotagging in real time with Bluetooth and  phone. The connection and settings are remembered both by the camera and phone app making activation fast and easy.

In-camera geotagging using a GPS log and connecting to a phone by Wifi. Not yet tested.

Touch screen

I haven’t tried this yet, but I learnt from a video that the touch screen can be very handy for waist-level shooting, and I also expect it to be useful for macro with the camera near the ground. When enabled, one can select the focus point and trigger the shutter by simply touching the screen, like in most smart phones. This feature is not new to the OM-1.

Leave a Reply