Sony View-DR HDR Technology

Steve Castle

Well-known member
HDR technology already used in Sony security cameras, hopefully we'll see this in Sony's other EXMOR CMOS cameras in the near future. Basically, the sensor simultaneously takes 4 different exposures within the same exposure time and combines them for a wide dynamic range image.

Via Image Sensors World:
http://image-sensors-world.blogspot.com/2011/06/sony-view-dr-hdr-technology.html

I seem to have missed this news from almost a year ago that Sony has announced View-DR HDR technology for security cameras. A fast sensor simultaneously (!) captures four exposures combined in a single image:


Sonyview-dr-image.jpg
 
should not, since the additional info is taken from shorter periods taken inside the longest one.
the only drawback of this is it's like shooting with a slow shutter speed to get a maximum of light.
 
Slow shutter speed shooting does affect motion! Another issue is that if the sensor readout is ultra fast to be read four times in the space of one, the image would look like a shutter speed that is four times faster as well. No problem for security cameras, but could be a mess for everybody else.

I wondered if a sensor could be made with three times the photosites needed for the chip size, (sort of like the 5D) and all of the photosites would be active at once. But instead of line skipping, the three sets of photosites would grab a high, mid and low exposure then be combined to form one HDR image. I don't know if the tiny offset of the three groups would create detail issues but the images could have a lot of DR. Not that it needs to go overboard, but it would be a way to pull more out of a scene.
 
Slow shutter speed shooting does affect motion! Another issue is that if the sensor readout is ultra fast to be read four times in the space of one, the image would look like a shutter speed that is four times faster as well. No problem for security cameras, but could be a mess for everybody else.

I wondered if a sensor could be made with three times the photosites needed for the chip size, (sort of like the 5D) and all of the photosites would be active at once. But instead of line skipping, the three sets of photosites would grab a high, mid and low exposure then be combined to form one HDR image. I don't know if the tiny offset of the three groups would create detail issues but the images could have a lot of DR. Not that it needs to go overboard, but it would be a way to pull more out of a scene.


I think there is some misunderstanding about how this technology operates...

This Sony HDR technology takes 1 long frame, and DURING that one frame takes three short frames while that longer frame is being taken. This is possible due to a double mid-point shutter.

Just as importantly, the frames are integrated at the end of the long frame and are then passed to the ADC. So you are not dealing with 4 different exposures and overlapping ghosting type artifacts that you would usually get from that, it'll be more like a filmic blur. But, as discussed in the 2007 paper which this technology was first introduced, you will get false colors at the boundaries of fast moving objects that are set against high-contrast backgrounds, but there is error-correction built in to reduce this artifact.

What's interesting about this HDR technique is that there is no sensitivity loss and actually suppresses FPN (fixed pattern noise). The reason for this is that it actually takes the full-exposure frame rather then 4 really short frames. The downside of this is technique is that it requires additional logic and requires a very fast sensor. This HDR technique has to be integrated from the sensor level, its not something you can add via software update.
 
Last edited:
Back
Top