Ambarella announced the availability of the A5 SoC platform for hybrid cameras. The A5 family consists of three products: A530, A550 and A570. Each is available with a full reference design. The A530 integrated circuit can capture 10 MP still pictures while generating and compressing high-definition video in both 720p60 and 1080p30 formats. The A550 in addition supports the standard HD broadcast format 1080i60. The A570 captures as many as 5.3 MP at 60 frames per second. This allows full HD video recording while concurrently storing 5.3 MP still images (16:9 aspect ratio) without interrupting the video recording.
The A5 platform is available now in sampling quantities.
Friday, January 09, 2009
Thursday, January 08, 2009
Omnivision Cuts TSMC Orders
EETimes: FBR Capital Markets predicts that utilization rates at TSMC could decline to below 40 percent in 1Q '09. Omnivision is named among the companies making the biggest order cuts.
What Faster Sensors Can Do
The fast CMOS sensors are supposed to be a big differentiator in the near future. Casio just announced its fast cameras line-up showing a part of future high-speed camera capabilities:
High-Speed Anti-Shake and High-Speed Night Scene functions
These functions work because the cameras can combine a number of images taken in burst mode, and then align the position of the subject to form a single shot. This means that, even without a tripod, zoom photography and night shots are beautifully rendered, virtually making photo blur a thing of the past.
30 shot-per-second high-speed burst shooting ensures that every crucial moments is captured
Both of these cameras can take up to 30 shots per second of high-resolution 6.0-megapixel images. What’s more, they can save up to 25 frames prior to the instant that the shutter button is actually pressed, so even if the user hits the shutter a bit late, that special moment will not be missed. The number of shots per second can be set at 30, 15, 10, 5, 3 or auto, and the user can set the total number of shots in one burst at 30, 20, 10 or 5.
Lag Correction
These cameras make it possible to set in advance the time lag between the moment that the user wants to capture and the actual time of taking the picture. This way, it is possible to capture that precious instant at any time, with minimal lag. This works because the cameras select from the images they temporarily store in their burst mode buffers the one image that was taken at the specified interval prior to the shutter being released, and just that image is saved.
Slow Motion View
By pressing a dedicated button during shooting, it is possible to view the movement of the subject in slow motion on the LCD monitor. The user can press the shutter while viewing the LCD monitor, selecting the perfect moment to capture from a scene that looks as if time has been slowed down.
I wonder how long the battery lasts with all these nice features activated.
High-Speed Anti-Shake and High-Speed Night Scene functions
These functions work because the cameras can combine a number of images taken in burst mode, and then align the position of the subject to form a single shot. This means that, even without a tripod, zoom photography and night shots are beautifully rendered, virtually making photo blur a thing of the past.
30 shot-per-second high-speed burst shooting ensures that every crucial moments is captured
Both of these cameras can take up to 30 shots per second of high-resolution 6.0-megapixel images. What’s more, they can save up to 25 frames prior to the instant that the shutter button is actually pressed, so even if the user hits the shutter a bit late, that special moment will not be missed. The number of shots per second can be set at 30, 15, 10, 5, 3 or auto, and the user can set the total number of shots in one burst at 30, 20, 10 or 5.
Lag Correction
These cameras make it possible to set in advance the time lag between the moment that the user wants to capture and the actual time of taking the picture. This way, it is possible to capture that precious instant at any time, with minimal lag. This works because the cameras select from the images they temporarily store in their burst mode buffers the one image that was taken at the specified interval prior to the shutter being released, and just that image is saved.
Slow Motion View
By pressing a dedicated button during shooting, it is possible to view the movement of the subject in slow motion on the LCD monitor. The user can press the shutter while viewing the LCD monitor, selecting the perfect moment to capture from a scene that looks as if time has been slowed down.
I wonder how long the battery lasts with all these nice features activated.
Tuesday, January 06, 2009
Two Patents Came Late
Two patent came late for the last week patent review. One of them is Forza's US7471231 differential dual slope ADC - easily understandable for the picture below:
The differential approach is said to enable a more consistent switching point, allowing the amplifier to be linear over a larger part of its range. It's probably true, but at the expense of added area and power consumption. Also, the PREAMP block should have an wide common mode range and good high-frequency common mode rejection, which might increase power even more.
Altasens' application US20080316342 looks quite obvious to me. It talks about gain control with fine steps. The gain control is split between analog and digital. The analog gain provides coarse gain steps and the digital gain provide finer gain steps between adjacent coarse analog gain values. Since analog gain can suffer from low precision, on-chip calibration is implemented to calibrate the analog and digital gain.
If granted, it would become quite broad and strong patent, but to me it seems obvious and I think there would be some prior art, even though I can not point to it specifically.
The differential approach is said to enable a more consistent switching point, allowing the amplifier to be linear over a larger part of its range. It's probably true, but at the expense of added area and power consumption. Also, the PREAMP block should have an wide common mode range and good high-frequency common mode rejection, which might increase power even more.
Altasens' application US20080316342 looks quite obvious to me. It talks about gain control with fine steps. The gain control is split between analog and digital. The analog gain provides coarse gain steps and the digital gain provide finer gain steps between adjacent coarse analog gain values. Since analog gain can suffer from low precision, on-chip calibration is implemented to calibrate the analog and digital gain.
If granted, it would become quite broad and strong patent, but to me it seems obvious and I think there would be some prior art, even though I can not point to it specifically.
Monday, January 05, 2009
Aptina Developed Image Quality Measurement Ruler
SPIE: Aptina has developed a softcopy ruler method to measure image quality. It enables the calibrated visual assessment of image quality. Because the assessments are measured in just noticeable difference (JND) units, it is possible to combine measurements of different attributes and predict overall quality. The method was developed during phase 2 of the Camera Phone Image Quality Initiative sponsored by the International Imaging Industry Association (I3A).
Aptina Acquires Handset Software Company
Trading Markets: Shanghai, China-headquartered fabless semiconductor company Chipnuts has sold its mobile phone software platform business to Aptina Imaging. How much Aptina paid for the newly-acquired unit is unknown, according to an employee for the Shanghai company, adding that the buyer will probably get the 30-40 Chipnuts employees at the unit.
Samsung Expands in California
Samsung Semiconductor has opened two R&D centers in California for advanced image sensor technology to expand the R&D activity of its Semiconductor R&D Center and Systems LSI Business Unit in South Korea. One center, associated with Systems LSI (Image Sensor Products Group) will be located in San Jose and focus on next-next generation technology. The second, associated with the Semiconductor R&D Center will be located in Pasadena, not far from Caltech and JPL, and will be looking at imaging technologies for more future application.
Dr. Eric R. Fossum will be directing the R&D activities in the 3 locations and is now looking to hire from the best and brightest at all levels to staff the R&D activities. Disciplines of interest include advanced imaging algorithms and ISP design, deep sub-micron device physics, optics, device simulation, pixel design and layout, and device and test chip electro-optical characterization.
Resumes can be sent to Fossum dot eric at partner dot Samsung dot com
Congratulations to Eric with the promotion and good luck with this new assignment! My only concern is that Eric might not have a time to organize the sensor community events now.
Dr. Eric R. Fossum will be directing the R&D activities in the 3 locations and is now looking to hire from the best and brightest at all levels to staff the R&D activities. Disciplines of interest include advanced imaging algorithms and ISP design, deep sub-micron device physics, optics, device simulation, pixel design and layout, and device and test chip electro-optical characterization.
Resumes can be sent to Fossum dot eric at partner dot Samsung dot com
Congratulations to Eric with the promotion and good luck with this new assignment! My only concern is that Eric might not have a time to organize the sensor community events now.
Sunday, January 04, 2009
Awaiba Comes with Its Own Branded Products
Awaiba, the custom sensor design house, has come with its own branded high speed line sensors. The sensors are oriented to machine vision market with pixel sizes ranging from 3.5um to 500um.
The company's main business remains the development of custom specific CMOS image sensors which are not available on the market, while the standard products are just a byproduct of its core activity.
The company's main business remains the development of custom specific CMOS image sensors which are not available on the market, while the standard products are just a byproduct of its core activity.
Friday, January 02, 2009
Tensilica Talks about Video Processing Pipeline
Embedded.com published Tensilica article describing HD video processing pipeline and benchmarks - a marketing-level educational article.
University of Glasgow, Oxford University and Sharp Laboratories Europe Cooperate on Image Sensor Improvements
BBC News reports that UK Engineering & Physical Sciences Research Council granted £489,234 to the team led by University of Glasgow for work on image sensor improvement. The article is mostly based on University of Glasgow news published just over a month ago.
It's hard to understand what exactly is going to be developed. The project will take advantage of a phenomenon called plasmon resonance in their efforts to create an imager that will produce sharper, more colourful images. Plasmon resonance refers to an interaction produced when light waves fall on a metal surface, or in this case, the thin metal film used on CMOS image sensors.
When light shines on the metal film, electrons on the surface absorb the energy of the light waves and begin oscillating, or shaking, in groups. The resultant combined waves are called plasmons and they modify the light distribution around the metal. The metal nanostructure is claimed to increase the sensitivity of the CMOS image sensor and result in higher-quality images.
The structures will enable the plasmon resonators to be ‘tuned’ into the same frequency as various colours of light, thereby improving color discrimination in images. This could offer a cheaper way of filtering different colors of light, reducing the current number of processes currently used to distinguish between different colors.
To me this work sounds like a search for a substitute for color filter and, may be, microlens.
Update: The Herald also runs an article clearly stating that the technology is intended to substitute or improve color filters.
It's hard to understand what exactly is going to be developed. The project will take advantage of a phenomenon called plasmon resonance in their efforts to create an imager that will produce sharper, more colourful images. Plasmon resonance refers to an interaction produced when light waves fall on a metal surface, or in this case, the thin metal film used on CMOS image sensors.
When light shines on the metal film, electrons on the surface absorb the energy of the light waves and begin oscillating, or shaking, in groups. The resultant combined waves are called plasmons and they modify the light distribution around the metal. The metal nanostructure is claimed to increase the sensitivity of the CMOS image sensor and result in higher-quality images.
The structures will enable the plasmon resonators to be ‘tuned’ into the same frequency as various colours of light, thereby improving color discrimination in images. This could offer a cheaper way of filtering different colors of light, reducing the current number of processes currently used to distinguish between different colors.
To me this work sounds like a search for a substitute for color filter and, may be, microlens.
Update: The Herald also runs an article clearly stating that the technology is intended to substitute or improve color filters.
Subscribe to:
Posts (Atom)