Q&A session with Mr. Takahiro Tosaka,
General Manager Strategy
Brand Development & Marketing
Consumer Products Division
This Q&A session was actually held in writing, and in English. So Mr. Tosaka’s answers are taken literally word by word from his letter. This should help to avoid ambiguities, that sometimes happen after several consecuent translations.
Dimitri Kroupski: – Thank you for the great new camera! And for the willingness to answer to my questions about it.
The new camera Olympus OM-D E-M1X uses two TruePic VIII processors, so it has twice the processing power of the previous model E-M1 Mark II. How is this power utilized, what is it mainly used for?
T.T. – Mainly for following functions:
- Handheld High Res Shot
- Live ND
- Intelligent subject detection AF
- Faster start-up time and processing speed
- Dual UHS-II SD card slot
D.K. – A few years ago Olympus leaders told me, that one of the main points in developing new camera always is – to allocate computing power between image quality and shooting speed, to find the optimum balance. Is it the case now?
T.T. – The basic concept is the same. We aim high image quality without sacrificing the mobility and the high speed performance of mirrorless system. E-M1X achieved evolution of image quality with hand-held high resolution shot, thanks to dual processors.
D.K. – Has the image processing algorithms (on pixel level) changed from the previous model? If yes, in which terms? Better high ISO performance? More precise colors? Has the algorithm itself changed, OR the processor just applies more power within the same algorithm?
T.T. – E-M1X has improvements on overall image quality. Particularly, color reproduction, image resolution at low sensitivity and noise & resolution at high sensitivity (ISO6400) have been improved by modified tuning based on feedback from professional photographers.
Note: The improvements on overall image quality are not thanks to dual processors.
D.K. – Let’s imagine, if you had even more processing power – would it further improve image quality? Or do you have already achieved some kind of maximum quality?
T.T. – In principal, once image processing engine and image sensor evolve, overall image quality at high sensitivity can be improved and new shooting function can be developed. Of course current image quality is not maximum, we will pursue high image quality by focusing on utilising our high quality lens performance as much as possible.
D.K. – We see that the formal speed characteristics remain the same (burst speed in H and L modes, with mechanical and electronic shutter). Does it mean that THIS part of functionality has not required additional computing power? (This is again about the same – WHERE do the extra power go to? 🙂
T.T. – Although the continuous shooting speed is the same as on the OM-D E-M1 Mark II, the maximum recordable frames of sequential shots and reduced buffer clearing time has resulted in significant improvements in practical high-speed sequential shooting usage.
D.K. – In the camera menu (page E1) there is an item: — Low ISO Processing — Detail Priority / Drive Priority (“At low ISO speeds, maximum recordable sequential shooting frames take priority in image processing for still photography”). Could you please comment on that? What is meant by “Low” ISO here? Does this relate to my above questions regarding the allocation of the power between speed and quality? Why does this relate to LOW ISOs only, isn’t it relevant to ANY ISO value?
T.T. – This is a menu to select image processing at sensitivity below 2000, the default setting «Drive Priority» allows you to have enough image quality in normal use. «Detail Priority» is a special mode for fulfilling the demands from specific number of professional photographers, that they prefer to prioritise on image processing which allows to have better image resolution even by reducing the speed of sequential shooting.
D.K. – Probably, extra processing power mostly go to the improved autofocus system?
T.T. – Although we cannot answer details, dual processors have beneficial effects widely including image processing, not only speeding up AF.
D.K. – How does the “objects tracking” work? (for trains, planes, cars…) – Does it use some database of patterns? (similar to ESP, can we say so?) — Does some extra AI, some learning process being added?
T.T. – It is merely using an algorithm which was developed by utilizing deep learning, a type of AI. Not self-learning.
D.K. – The objects list seems a bit limited (trains, planes…). Do you plan to add some extra types of objects? Is that possible?
T.T. – Although we cannot give details, there are plans to add more subjects via a firmware upgrade.
We cannot give any information on the candidates for specific subjects.
D.K. – Regarding the Handheld Hi-Res. Firstly, thank you very much for adding this long awaited function!
We hoped since the very introduction of Hi-Res, that some time it would become handheld. Great.
T.T. – Thanks a lot for your feedback, we will input it to our R&D department.
D.K. – Frankly, I had imagined that it would be implemented differently. Let me explain. The Image Stabilization System knows, how the camera is being shaken (and ISS in normal modes compensate these shakes). In Hi-Res (old style, “tripod mode”) the ISS shifts the sensor by tiny increments. So it seems possible to superimpose these two processes – to shift the sensor in a 8-shots-circle, AND taking into consideration the shakes. Why not? You have implemented the Handheld Hi-Res differently – why?
The Handheld Hi-Res uses natural hand shaking as a source to get shifted images. So, it is BAD if the photographer holds the camera very steadily?
T.T. – Tripod High Res Shot shifts the sensor in 0.5 pixel increments while the camera is hold on tripod still, capturing 8 shots which are then composited to form a single shot. On hand-held high res shot, it is not possible to shoot at aimed imager position by shifting it 0.5 pixel because position of the camera itself moves due to hand shake of the photographer. Therefore we changed our mindset and developed the method to composite a single shot from several images by utilising the handshake.
Difference of the aimed imager position should be very small so even professional photographers cannot avoid it (there is almost nobody who can hold the camera still without shake level of imager position).
D.K. – When the Hi-Res mode was originally introduced (in E-M5 II), it was not possible to add it to functionality of the older models, since it required more precise sensor movements. What about new Handheld Hi-Res function? Can it be added to other cameras with firmware updates? Is it purely a computational feature, or again, requires more precise sensor movements?
The effectiveness of ISS is improved in E-M1X (and it’s great! The fantastic Olympus’s feature!). How is it achieved? Better sensors? More precise sensor movements? Faster processing?
T.T. – Olympus is the worlds’ first brand which introduced 5 axis IS and has been taking actions carefully to the feedback from the market, we have been offering unrivaled IS performance in the industry.
We have adopted a gyro sensor developed jointly with Epson and an improved algorithm. The new gyro sensor is nearly five times more accurate at detecting camera shake than the previous model.
D.K. – When ISS achieved 5.5 EV, the Olympus told us that further improvements are somewhat limited by the Earth rotation. How was this problem solved? What have you done to the Earth rotation? J
T.T. – Although it is not possible to cancel the effects of the earth’s rotation, 7.5 stops of compensation is achieved by a more precise gyro sensor and minimizing other errors.
D.K. – By the way, regarding the 25Mp Hi-Res mode (actually, not in E-M1X only, but in general). How does this mode work? I guess that the sensor makes bigger increments (not half-between-the-pixels steps, but the whole pixel step), and collects the full color information (usually stolen by the Bayer filter and interpolated)? Similar to the Pentax’s hi-res mode. Is it correct? Or if no, then what is the meaning of 25Mp Hi-Res?
T.T. – 25MP high res mode is available for demands for who need high resolution image but with minimum data size. It’s not similar to Pentax’s high res mode.
D.K. – What about compatibility of the Image Stabilization systems between Olympus and Panasonic cameras/lenses? Do you work together on this issue?
T.T. – We have compatibility for both brands’ bodies and lenses, however it is unfortunately difficult to support the image stabilization mutually, because our sync IS uses different algorithm from Panasonic’s one. Compatibility about aperture ring has not been realized, however we understand the market demands.
D.K. – Is the image sensor in E-M1X the same as in E-M1 II? Does it have the same readout speed?
T.T. – Yes, it is same including readout speed (compare to E-M1 Mark II).
D.K. – As far as I understand, the E-M1X is not replacing the E-M1 II, it just starts another line of professional cameras? And the “ordinary” E-M1 line will continue, correct?
T.T. – Both E-M1 Mark II and E-M1X are positioned as professional models. We will continue with the current line-Up strategy with successor models.
D.K. – What can we expect earlier – E-M5 III or E-M1 III?
T.T. – Although we cannot clearly tell which model comes first, please look forward to it.
D.K. – Thank you very much!
1. Общий обзор и органы управления
2. Меню, новые функции и параметры
3. Качество изображения, стабилизатор, режим High Res с рук
4. Примеры снимков
5. Режим Live ND
6. Интервью с Такахиро Тосака, Olympus (на русском)
7. Q&A session — Takahiro Tosaka, Olympus (ENGLISH)
8. Тест скорострельности при серийной съёмке
9. Система автофокусировки
10. Дополнения и ответы на вопросы