Hillary Grigonis/Digital TrendsOlympus launched only a single digital camera final yr, the consumer-focused E-PL9, as three different manufacturers stole the present by venturing into full-frame mirrorless. Olympus now must show that the Micro Four Thirds system, a comparatively small-sensor format that usually falls wanting the decision and picture high quality of bigger APS-C and full-frame sensors, remains to be related in 2019. It goals to just do that with a brand new flagship digital camera, the OM-D E-M1X.
More Olympus Coverage
While the sensor inside it’s the identical as what’s used within the older OM-D E-M1 Mark II, the brand new $3,000 digital camera incorporates a number of firsts for Olympus, together with twin processors, a picture stabilization system rated for as much as 7.5 stops of shake discount, an autofocus system designed to rival DSLRs, and a built-in vertical battery grip. What it might lack in uncooked picture high quality, it makes up for in pace and efficiency.
The twin processors additionally energy one other main milestone within the business — utilizing deep studying algorithms to assist the autofocus system. This is primarily helpful when capturing motorsports, the place it would acknowledge and give attention to the motive force’s helmet as a substitute of the car. (It also can acknowledge trains and airplanes.)
So what went into making the OM-D E-M1X and the place’s Olympus headed subsequent? To discover out, Digital Trends sat down to speak with a number of the folks that truly make Olympus cameras: General Manager Elija Shirota, Team Leader and autofocus skilled Tetsuo Kikuchi, and Senior Supervisor and deep studying skilled Hisashi Yoneyama. The interview was translated by Akihito Murata, Vice President of Sales and Marketing, and the under transcript has been edited for readability.
Hillary Grigonis/Digital TrendsDigital Trends: Why did Olympus resolve to place twin processors on this digital camera?
Eija Shirota: The first place to begin is for reliability. We considered the best way to create the final word reliability and one of many solutions was to equip the digital camera with two engines.
Was the built-in grip crucial to slot in the twin processors? Why did you resolve to include the grip as a substitute of as an add-on accent?
Eija Shirota Olympus
Eija Shirota: [With] this mannequin, an enormous factor for us is final reliability. [If] we’ve a separate grip, the connectors — as soon as we’ve these sort of components within the center — we will’t obtain final reliability. We could make a dependable digital camera, however for that final reliability, this kind of factor ought to be built-in.
The different cause is that we’ve talked to photographers and we’ve truly noticed photographers and the way they use the cameras. Many of the photographers used the vertical place and would function the digital camera with out something.
To obtain that, we should always have the very same place of the buttons and shutter launch [when shooting vertically]. If we need to obtain that, the grip wanted to be built-in to maintain the whole lot in the identical place.
With this new autofocus system, we made positive that the main focus is stacked with the thing.
With a earlier digital camera, Olympus stated the stabilization couldn’t get any higher — nevertheless it simply did. Can the stabilization get any higher than 7.5 stops?
Eija Shirota: When we launched the E-M1 Mark II, we thought that we’d accomplished the whole lot apart from accounting for the rotation of the earth. But by eliminating all the opposite components, we’ve managed to attain as much as 7.5 [stops].
Now, lastly, with out eradicating the rotation of the earth, we shouldn’t be in a position to go any larger — not only for Olympus, however for every other model. So to reply your query: We managed to remove all these different components in addition to the rotation of the earth. We are assured that that is probably the most highly effective IS on this planet.
What have been a number of the challenges to designing the E-M1X?
Eija Shirota: First of all, designing the twin engines. This is the primary time for us to try this and that was a problem to place the 2 engines collectively. The second factor is that it’s not concerning the options, however to replicate all of the requests from professionals. It took us time to introduce a brand new mannequin. It’s been awhile since we’ve launched a brand new mannequin the final time. But, this time was used to take heed to skilled photographers and their necessities for a digital camera. The autofocus was a type of.
What different large options are within the E-M1X?
Eija Shirota: The handheld excessive decision shot is an enormous achievement for us. We even use hand shake to attain it. From a know-how viewpoint, this can be a large achievement. This will enable the customers to hold the smaller cameras with no large sensor. That’s an enormous achievement.
Hillary Grigonis/Digital TrendsWhat’s new on the autofocus system on the E-M1X?
Tetsuo Kikuchi: For film operate, we use part detection autofocus. We use a brand new manner of controlling the system that makes the autofocus system extra correct. To be very concrete, previously typically the autofocus simply goes to the background. With this new autofocus system, we made positive that the main focus is stacked with the thing. This is extra correct than earlier E-M1 sequence.
What is the AF system rated for in low mild?
Tetsuo Kikuchi: The sensor itself is identical, however we took a special strategy to have extra correct autofocus with new algorithms. As a end result, low mild circumstances are way more improved than the earlier mannequin. Tetsuo Kikuchi Olympus
This time, we’d prefer to stress the 9-point autofocus. The goal for us was to have the identical accuracy as a DSLR. We consider that we’ve achieved that with this new mannequin. The trick is utilizing the 9-point autofocus [mode]. Sometimes we see that when you use the continual autofocus, the main focus just isn’t all the time steady. Sometimes focus is right here and there, it’s a really small factor however this is essential for professionals.
We adjusted the algorithm to verify the autofocus is all the time within the heart. This is a really particular adjustment that we’ve made.
You will discover it while you use the nine-point AF and steady autofocus, you will notice a distinction from earlier fashions.
What know-how made it attainable to get that DSLR-like efficiency?
Tetsuo Kikuchi: We can’t clarify the small print, it’s confidential. But what we will say is that that is very a lot a brand new algorithm and the mixture with the sensor precedence autofocus makes this autofocus system very correct.
Hillary Grigonis/Digital TrendsHow did you practice the deep studying system?
Hisashi Yoneyama: It’s not accomplished inside the digital camera, we used a excessive specification laptop. We used 10,000 pictures per class.
For instance, when speaking about vehicles, there are completely different shapes of vehicles, like Formula 1 and NASCAR. Per sort, we give the system a pair thousand pictures to let the system run to acknowledge that automotive. This info is given on a excessive specification laptop computer, then transferred to the digital camera.
So you labeled these pictures by hand to acknowledge the completely different components of the automotive?
Hisashi Yoneyama: Yes.
We are contemplating making use of this know-how to extra cameras.
What have been a number of the challenges you confronted growing the monitoring system?
Hisashi Yoneyama: The greatest problem was how correct the system can detect the mannequin. For instance, there are a number of several types of backgrounds and several other several types of vehicles.
We needed to guarantee that the system acknowledges the automotive and focuses on the automotive and the [driver’s] helmet precisely. To obtain that, we wanted to offer numerous pictures, in order that was the largest problem.
Deep studying often takes lots of laptop energy. How did you match the whole lot within the digital camera?
Hisashi Yoneyama: You want an enormous quantity of information while you make the algorithms. This half is completed not within the digital camera however individually on a PC. Once we get the information to make the algorithm, and switch the algorithm to the digital camera, we don’t want that large quantity of information. The digital camera simply makes use of the algorithms.
Do you see utilizing extra deep studying algorithms on future cameras?
Hisashi Yoneyama Olympus
Hisashi Yoneyama: Yes, we’re contemplating making use of this know-how to extra cameras. But, the present problem is that this digital camera has two engines. We want large energy to run this algorithm, and this will’t be achieved by all of the fashions, so we’ve to contemplate which fashions will obtain this know-how. But the reply is sure.
Akihito Murata: I’d like so as to add that, to completely make the most of this know-how, you want a really highly effective engine. Without having two engines, it’s very troublesome to attain this. That’s why some manufacturers use a number of the deep studying applied sciences, however at present, it’s not attainable to completely make the most of that knowledge. That’s why Olympus is, in the meanwhile, the one one to make the most of deep studying know-how for vehicles, trains, and planes.