With the premium interchangeable lens camera FUJIFILM X-Pro1, we developed the X-Trans CMOS Sensor which has an image quality surpassing those of 35mm full frame sensor single lens reflex cameras, despite being an APS-C size. The impressive image resolution and imaging ability achieved by the optical low pass filterless sensor was highly praised by many customers.
The next development we took was to achieve the high image quality of X-Trans CMOS not only with interchangeable lens cameras but also with compact cameras as well. Furthermore, believing that photographers are satisfied with camera for not only its image quality but also its response function and AF speed, we focused our efforts into faster response and increased precision. As part of those efforts, we decided to work toward including phase detection pixels within the image sensor for detecting subject mismatching. Inclusion of these phase detection pixels would vastly increase AF speed.
X-Trans CMOS pixle array
Generally, compact cameras and mirrorless cameras use contrast AF. Contrast AF uses a mechanism that detects the focus location using the contrast of the subject. The benefit of this mechanism is that the focus precision is higher than external phase detection sensors due to its use of the image loaded within the sensor. However, a drawback is that the focus is taken while moving the lens back and forth, which requires quite some time to implement.
Meanwhile, phase detection AF splits light received by the lens into two directions, detecting the distance between the focus location based on the difference between those images. With this method, no movement by the lens is required to calculate the distance to the focus location, and saving such time allows for speedy focusing. Standard single lens reflex cameras have their phase detection AF sensors installed in a separate location from the subject image sensor. The phase detection method which we developed is the “image phase detection method,” where pixels which can detect phase detection are placed within the subject image sensor. This is what helps to achieve both speed and precision.
Fujifilm made strides early on toward installing phase detection AF within its compact cameras. This led to the development of Super CCD EXR, the first camera in the world to contain phase detection pixels within its sensors. FinePix F300EXR went on sale in 2010. It realized an AF speed of 0.158 seconds, among the fastest in the world at the time.
By installing phase detection pixels within the sensor and having the phase detection pixel on the left and right sides detect blurring (mismatching) between their optical images, the distance from the subject can be instantaneously and correctly calculated to put the image in focus. This is the mechanism used in the phase detection sensor. Phase detection pixels are distributed on nearly 40% of the total area of X-Trans CMOS II Sensor and the center of its screen, to achieve both high speed AF and high image quality via our unique calculation and image processing technologies. The maximum speed realized for FUJIFILM X100S is 0.08 seconds(*1), while the maximum speed for X20 is 0.06 seconds(*1).
The phase detection AF operates during video photography as well as during still image photography. The hybrid AF technologies allow free use and switching between two focus methods, such as switching to contrast AF in dark areas, for high speed and high precision AF in any situation imaginable.
*1: Fujifilm research based on CIPA guidelines and conducted in “High Performance” mode.
Generally speaking, placing phase detection pixels on the sensor surface changes its collection characteristics and affects other characteristics such as sensitivity and color mixing. Therefore, we decided to re-examine and re-design the sensor structure, so as to realize both high image quality and increased AF speed for X-Trans CMOS.
As stated earlier, installing phase detection pixels on the sensor surface changes its collection characteristics, affecting other characteristics such as sensitivity and color mixing. To prevent this, the collection efficiency of the sensors had to be improved. Therefore, we decided to optimize the structure of the collection optical system, including the distribution of phase detection pixels and microlenses, to greatly increase acceptance function. This allowed us to realize both increased sensitivity and reduced interference, leading to the achievement of high image quality.
X-Trans CMOS Sensor II is much faster as well. Thanks to the high speed circuit technology, it can load at approximately double (*2) the speed of its predecessors. Combined with the EXR Processor II image processing engine, this shortens photographing time intervals, increases the number of consecutive photographs which can be taken and speeds up video frame rates (from 24fps to 60fps).
It must be noted that FUJIFILM X100S is also compatible with 14bit formats. When RAW development is output in the TIFF (14 bit) format, its rich gradation expression allows for smoother photographic images.
The 2/3-inch sensor efficiently receives light and utilizes a back-illuminated type sensor to increase the freedom of design. By increasing collection efficiency, sensitivity is also increased. Optimizing the separation of the photodiodes nearby means color mixing can be avoided.
Also, efforts made toward optimizing the microlens resulted in more than 30% noise reduction(*3) and nearly double(*2) the speed of previous models, as well as 130% to 140% expansion of dynamic range, compared to its predecessors.
FUJIFILM X20 combines face detection and scene recognition functions. This, combined with intelligent switching to the AF method based on the subject scene, results in even faster and simpler photography with high image quality.
*2: Compared with the EXR Processor Pro.
*3: Comrared with the FUJIFILM X10
Phase detection pixel data is actively used in image formation as well, in order to realize distance calculation and high image quality. Focusing via digital split image included in FUJIFILM X100S is an innovative new idea which was realized by visualizing the image data received from phase detection pixels.
During image sensor development, we strive to resolve three major issues: 1. reducing interference; 2. increasing sensitivity; and 3. increasing dynamic range. The optimal method for sensor size was used for X-Trans CMOS Sensor II. By re-examinig and optimizing its design, we worked toward increasing each and every relevant value, as well as to increase image quality. Our goals remained unchanged in the development of the phase detection pixel sensor.
We are confident that the high image quality of the interchangeable lens premium camera FUIJFILM X-Pro1 was realized in a compact camera, and that this was combined with the increased speed in the design of the X100S and X20, resulting in the creation of a camera which lets the photographer easily take photos that come out exactly as envisioned. We hope you will personally experience its high functionality.
Image sensor development project leader: Kazuya Oda, Senior Engineer
Image phase detection AF optical system development leader: Takashi Aoki
X-Trans CMOS Sensor (2/3 type) development leader: Tomohiro Sakamoto, Senior Engineer
X-Trans CMOS Sensor (APS-C) development leader: Noriko Kawamura