The ClearVID CMOS, Sony's Chip of Choice

An increasing number of consumer and prosumer camcorders are turning away from CCDs in favor of CMOS technology. Sony, more than any other company, has pushed ahead in implementing CMOS technology.  Their decision seems to be paying off.  Since the introduction of the HDR-HC1, in 2005, Sony camcorders using CMOS imagers have had a near monopoly over the top-end of CamcorderInfo.com’s video performance scoring scale.

Starting with the standard definition DCR-DVD505 almost a year ago, Sony introduced their second generation CMOS, the ClearVID sensor. A month later they dropped the same chip into their second consumer HDV camcorder, the HDR-HC3. Both new AVCHD camcorders, the HDR-UX1 and the HDR-SR1, are also equipped with the ClearVID CMOS.  Sony certainly seemed to have found an imager they liked.

To learn more about Sony’s new ClearVID technology, Camcorderinfo.com sat down with Harry Haruna, Manager of Camcorder Product Planning for Sony USA, and Kevin Brinkman, Product Information Manager for Sony Camcorders.  The result was a wide-ranging discussion about the differences between CCD and CMOS technology, what differentiates the ClearVID CMOS from competing CMOS designs, and insights into the future of imaging technology.

CMOS versus CCD

CamcorderInfo.com: I want to take a chronological route, starting with how and why this chip came into development.  Let’s start with the fundamental differences between a CCD and a CMOS chip.

Harry Haruna: Okay, so, the big difference between CCD and CMOS is both of them use very old technology, but Sony has tried to improve these technologies. To make a CCD, a manufacturer needs more knowledge of the technology whereas CMOS requires greater manufacturing expertise.  Because we have had many advantages in CCD technology for a long time, we’re driving CCD development forward. Today, [CMOS] manufacturing technology is getting better, and CMOS has some good points vs. CCDs – for example, lower-power consumption.  So we’ve invented the ClearVID CMOS to improve our camcorders’ features.

CCI: Noise was a common complaint early on – can explain you why that was an issue with early CMOS designs?

HH: Even though CMOS itself has a huge performance for detail, it’s a middle-aged technology.  Everybody is developing CMOS technology, and everybody still has a lot of space to improve CMOS sensors.  We have definitely improved this noise issue [for our] CMOS.  In my personal opinion – everyone needs time to improve CMOS technology because we [have been focused on] CCD for years...

The Sony HDR-SR1, one among four camcorders currently using the ClearVID CMOS chip.

Kevin Brinkman:  As Harry was just explaining, more development went into CCD technology – increasing the picture quality and decreasing noise, versus the technology advances in CMOS.  Previously, fewer resources were applied to CMOS development versus CCD.  And now a transition is happening to CMOS from CCD.  More [of our] technology and resources are being pushed toward CMOS development – increasing picture quality, adding features, and decreasing noise.

CCI:  Can you talk about your first standard CMOS camcorders, and the first camcorder with the ClearVID CMOS?

KB: The DCR-DVD505 was the first Sony camcorder to use the ClearVID CMOS sensor, and in 2005, we had two products with regular CMOS sensors in them.  The first was the DCR-PC1000, which was a three-chip CMOS, and then the single chip DCR-HC1, the high-definition model.

CCI: Can you talk about the one-CMOS versus three-CMOS design?
 
KB: Right now, we don’t have any three-chip CMOS models in the [consumer] camcorder lineup.  As we were just discussing, the three-sensor model we had last year was the PC1000, and that was developed so we could deliver three-CMOS performance in a more affordably priced camcorder, instead of a $2000 three-CCD camcorder that we had released previously, like the DCR-HC1000.  We developed the DCR-PC1000, added some features and a Handycam Station, with three-CMOS performance for approximately $1300.

CCI:  So are CCDs on the way out, and CMOS is on the way in?  

HH: They both have their benefits.  Even though CCD is getting older as a technology, it still shows good performance, and lower development costs.  We still have a bunch of CCD technology appropriate for lower-cost products.  

KB: CCD continues to be a very good sensor technology.  I wouldn’t say that any changes to CCD, as far as implementation, are on the horizon.  Harry just mentioned, we look not only at feature sets but also what’s best for the customer at each individual target. In doing so it’s a business decision to switch between CCD to CMOS, just to enhance feature sets and the quality of video.

HH: CCD is still good for compact camcorders because there are ten types of CCD chips, but they all must be located near the lens. The other features are located elsewhere in the camcorder.   So CCD lends itself to making more compact products [due to this design attribute]. CMOS is great but the sensor size is bigger than CCDs, so they work best in larger camcorders.

KB:  That means the processor doesn’t have to be next to the CCD [as opposed to CMOS]; it can be located somewhere else inside the camcorder.  That allows us to make the camcorders smaller because we can relocate that processor to open space inside, decreasing the size of the overall body.  Part of the inherent design of a CMOS is the on-board processor – and its lower power consumption.

CCI: So does that mean that there’s a minimum size for CMOS camcorders?

KB: That would be almost talking about future technology, but what I can tell you is when CCDs first came to market they were huge in size, you know, quarter-inch, half-inch, third-inch, and now the technology is advanced so the CCDs can be actually smaller.  Right now we’re using one third-inch CMOS chips, so I can’t really speak about future technology, but if it goes the way CCDs did, I’m sure something will end up being smaller.

CCI: Were the camcorders that use the ClearVID now intentionally designed as small as possible?

KB: As small and compact and lightweight as possible based on the feature set.  And the reason I say that is, compared to the DVD405, which is using CCD, the DVD505, which uses the ClearVID CMOS, the body size is essentially the same.  However, the LCD screen size is the huge visual difference that you can see from a 2.7 to a 3.5 inch, so we do try to keep the camcorders as small as possible based on the feature set.

The ClearVID CMOS Design and Features

CCI: What was the difference between the regular CMOS sensor and the ClearVID? What changed from one version to the next?

KB: The ClearVID CMOS sensor is a complete redesign.  The previous sensor used a Bayer [pixel distribution] pattern versus the newer design on the ClearVID sensor.  Basically we increased the size of each pixel on the sensor, and then rotated it by 45 degrees. We increased the amount of green pixels as well because green is the color human eyes are most sensitive to – [they] enhance color quality and give you higher resolution.  The distance between the center points of each pixel did not change, meaning you get the same light sensitivity, the same depth of field, the same imagery, with the enhancement of the green pixels providing better resolution.

 

Pixel arrangement in the traditional Bayer pattern		Pixel arrangement on the ClearVID CMOS

CCI: Why tilt them 45 degrees?  What is that doing?

KB: For one, it allows us to use larger pixels over a standard horizontal pixel layout.  If we rotate the position into a diamond [shape], it allows us to increase the size of individual pixels but keep the same spacing between the center points of each pixel, so you don’t lose any depth of field or degrade color.  Earlier you asked about advances in CMOS technology, and that’s one of them.

CCI: So it follows that pixels are going to be shifted back to a square pattern to display …?

KB: The ClearVID pixels are still in a square pattern.  However, the layout on the sensor itself is rotated.  The nature of a square pixel shape [versus the octagonal array used by some manufacturers] allows for points to be in closer contact with each other so that there’s no spacing in between each individual pixel.

CCI: You talked about the benefit of more green pixels.  Is that actually increasing resolution or is that just increasing what the human eye perceives as more resolution?

HH: Basically, human eyes have good sensitivity to green color, so the green pixel’s pitch* is the same as current CCDs even though it’s rotated by 45 degrees. That means the actual green pixel pitch is the same as a CCD pitch, but larger green pixels [sharpen the image] to the eye.  In the case of the ClearVID sensor, the ratio of green to red and blue pixels is 6:1:1.  In the Bayer pattern, it’s 2:1:1.

CCI: At first glance that seems to be a pretty big ratio increase of green pixels in the ClearVID over the Bayer pattern. Does that require significantly different signal processing?

KB: The ClearVID CMOS models have what we’re calling an Enhanced Imaging Processor (EIP), which by processing increased green pixels still give the customer a proper color rendition, with enhanced resolution.  It’s not an increase of resolution as far as lines go, but in terms of the quality of imagery that the customer perceives.  It’s not overly green; it’s just more vivid, I guess you could say.

 

CCI: Is Sony offering the ClearVID design to any third-party manufacturers right now?

KB: No, we are not.

CCI: And how many camcorders are using the ClearVID design right now?

KB: Currently in the market are the DVD505, the HC3 HDV camcorder, and the two new AVCHD high-definition models: the HDR-UX1 and HDR-SR1.

CCI: You’ve clearly found a chip design that you like.  Why move all your top models towards the same sensor and what are the benefits?

HH: Well, CMOS has a faster data transfer capability [than CCDs], allowing us to add features like Smooth Slow Recording on high-end camcorders.  That faster data rate and high resolution make this and other new features possible.

 

The typical spacing between pixels on a CCD		The tighter spacing between pixels on the ClearVID CMOS

KB: Essentially, we look at the market and we try to develop each model for a specific target audience, and by doing so we have to decide on what features that would work best for each particular target and then choose the best sensor for that target.  It just so happens that the ClearVID CMOS sensor can give the user more features.

CCI: What’s the performance ceiling of a chip like this? Can you can speak to how and why you’re using the same CMOS sensor on both standard def and high def models?

KB: We did the same thing with last year’s DCR-PC1000 and the HDR-HC1.  Both models used the same CMOS sensor, and one was standard definition and one was high definition; it’s a function of the processing ability of the CMOS sensor.

HH: Also, the resolution depends on the data format rather than the sensor.  The CMOS sensor does capture more information [than standard definition DVDs can resolve], so we have to reduce the amount of data before burning a DVD.  But CMOS sensors have enough capability for high-definition camcorders using the HDV or AVCHD format.  We change the data based on recording format, so many pixels for [DVD] and so many pixels for HDV or AVCHD.

A Tour through a ClearVID Camcorder

CCI: For those who may not know how a sensor works, can you, in a very basic way, walk us through how camcorders with the ClearVID CMOS capture light and interpret it into a signal?

KB: Okay.  The basic processing hasn’t changed between the standard CMOS and the ClearVID CMOS, in that once the light goes through the lens it hits the pixel.  The pixel then processes that light – white, black, and so forth.  But by processing it on the board, it uses less power than a CCD, which has to transfer the data off of the board to a separate processor. 

CCI: Okay, so light hits the ClearVID CMOS sensor, the initial processing happens and the data is passed on. What next?

KB: Once it leaves the chip it goes through the EIP, which processes the data into an image.  The EIP also has different controls such as the Smooth Slow Recording feature.  If you are in Smooth Slow Recording mode, the EIP allows the imagery to be slowed down.  Video is 60i, or thirty frames per second so the EIP slows it down by [a factor of] four, so it captures in 60i, but plays back at 240 frames per second.

CCI: So if you’re using smooth flow record, is the processing speed actually increasing?

KB: Processing speed is increasing due to the fact that it’s slowing down live video.

CCI: Is that why it can only be used for short bursts at a time, changing three seconds of realtime to a twelve-second playback?

KB: That’s correct.  The EIP is the workhorse behind everything. It was found on the HC1, the high-definition piece but not on the PC1000.

CCI: So this is the same EIP as on the HC1?

KB: A similar processor – the HC1 didn’t have the Smooth Slow Recording feature. I guess you could say it’s version two – we’ve just added features to it. Smooth Slow Recording is a consumer-friendly feature, meaning that it allows for immediate playback of a fast moving object such a golf swing.  It allows golf pros or golf students or anybody to be able to analyze their swing and then play it right away.  Recording in the higher frame rate or higher shutter speed and slowing it down in post would take away from that immediate gratification.

CCI: And the EIP compresses the data as well?

KB: It is where the compression happens, as well as applying color quality, manual control settings, everything pertaining to the image. And then the camcorder’s EIP compresses the data into high-definition or standard definition video.  If you’re talking about the DVD505, it would discard unneeded information, and record an MPEG-2 standard definition file on the disc.

CCI: The DVD405 and 505 make a really interesting comparison because so much of their technologies are similar, and the core difference seems to be the type of chip.  The 405 has a CCD, a one third-inch CCD, and the 505 has a one third-inch CMOS.  What sort of visual differences could we expect to see in them, if lighting and everything else was the same for both camcorders?

KB: I would say, for one, on the DVD505 it would appear to have better resolution to the average consumer due to the enhanced green pixel count.  But using the ClearVID CMOS sensor gives the possibility to add greater functionality to the camcorder.  I think everybody goes looking for a camcorder from Sony with good picture quality, but do they want good picture quality and the ability to do slow motion video and you get a higher resolution still photo, as well as a dual record function?  And of course less power, which means longer battery life.

 

CCI: Thank you very much for talking with us.

 

---

*Dot or pixel pitch describes the distance between sub-pixels.  A smaller dot pitch measurement generally results in a sharper image.