Luminous Landscape Forum
Equipment & Techniques => Digital Cameras & Shooting Techniques => Topic started by: thierrylegros396 on February 27, 2013, 11:46:24 am
-
Hi to all,
I've just found what seems to become a very interresting new Technology for HDR photo !
The idea is very clever and adaptable to CMOS sensors.
http://www.dpreview.com/news/2013/02/27/rambus-shows-binaryt-pixel-sensor-technology-for-expanded-dynamic-range (http://www.dpreview.com/news/2013/02/27/rambus-shows-binaryt-pixel-sensor-technology-for-expanded-dynamic-range)
http://www.rambus.com/us/technology/innovations/binary-pixel-imager/index.html (http://www.rambus.com/us/technology/innovations/binary-pixel-imager/index.html)
Have A Nice Day.
Thierry
-
Doesn't seem to be anything ground breaking. It's basically the same thing we have in many cameras now, except it does the 2nd exposure at a further level, at the pixels rather than just a whole 2nd exposure either with a sensor or normal shutter.
-
Doesn't seem to be anything ground breaking. It's basically the same thing we have in many cameras now, except it does the 2nd exposure at a further level, at the pixels rather than just a whole 2nd exposure either with a sensor or normal shutter.
How is that "the same thing" if it is done within the single exposure, i.e., not needing the second one?
-
It's not technically a single exposure, if you read it. The pixels 'reset for a second exposure'. It's just done electronically not involving the shutter.
-
It's not technically a single exposure, if you read it. The pixels 'reset for a second exposure'. It's just done electronically not involving the shutter.
The "second exposure" they refer to seems to be the second for the given - overloaded - pixel(s) only. They still refer to the whole process as "single-shot exposure."
-
Single-shot, yes. But again, nothing new, that's how phones do it now. The only difference is now they 2nd expose all the pixels, the new process does them as-needed. Same result to the end user.
-
What do you mean "that's how phones do it now"? That's how phones do what? HDR? That's how all cameras do HDR.
It doesn't appear to me that this is a second exposure. Looking at the sample scene they used, the 'proper' shutter speed for the shadowed water is going to be much longer than the 'proper' exposure for the sunlit mountains. It's entirely likely that, if their technology is that good, that they could instruct the device to shut off and 'restart' the exposure for the highlight pixels within the time of the exposure for the shadowed areas, then with some sort of remetering, turn the exposure for those pixels back on such that the overall exposure is still the same as for the shadowed area. That's my supposition as to what's happening. If my supposition is correct, it's nothing like anything any other cameras are doing currently.
-
The description of the technique is not very clear, but this is what I assume they are doing:
- Expose for the shadows -- more than normal exposure -- and let highlights saturate.
- Highlight pixels will receive more charge than the pixel wells can accommodate.
- Monitor every pixel with a threshold near full well saturation level.
- When threshold is crossed, reset the pixel (discard collected charge and continue collecting).
- Count how many times threshold is crossed, so that original signal can be recovered.
-
Doesn't sound the same. Sure there are HDR-esque features in smartphones now where it always takes two exposures and combine in-phone, but they often result in funky double exposures with moving subjects:
(http://farm8.staticflickr.com/7004/6407503785_7cb5ce8a30_o.jpg) (http://www.flickr.com/photos/rawhead/6407503785/)
Flock Together (http://www.flickr.com/photos/rawhead/6407503785/) by Dr. RawheaD (http://www.flickr.com/people/rawhead/), on Flickr
So let's say your camera meters the scene to be 1/125s. Conventionally, the phone might take a second exposure at 1/500s (for a total of 1/100s+) and the two will be combined into HDR.
With this technology, let's say your camera meters the scene to be 1/125s and commences the exposure. Now, let's say some areas of the sensor will be overexposed, because you have bright spots in the scene. What happens then is that, say, after 1/250s those area become saturated due to overexposure, ONLY those pixels are reset and thereby commence a second exposure @ 1/500s.
Note that the entire process will be completed after 1/125s, which is the "original" exposure. Furthermore, since you're not "combining" two layers into one, but rather merging the double exposed pixels with the single exposed pixels, it will reduce a lot (though not all) the funkiness that can happen.
I think this is a very cool idea, and I've had similar ideas in the past: basically, that overexposure can, theoretically, be completely avoided with digital technology, since the system/pixels would be aware when it receives too much light: it just needs to stop exposing at that point :-)
-
One big difference from doing HDR with multiple frames is timing: it all happens in the time of a single exposure for the darker parts of the scene, so that it works hand-held – and for video ...
... if and when it is actually developed into a working product!
-
let's say your camera meters the scene to be 1/125s and commences the exposure. Now, let's say some areas of the sensor will be overexposed, because you have bright spots in the scene. What happens then is that, say, after 1/250s those area become saturated due to overexposure, ONLY those pixels are reset and thereby commence a second exposure @ 1/500s.
Note that the entire process will be completed after 1/125s, which is the "original" exposure. Furthermore, since you're not "combining" two layers into one, but rather merging the double exposed pixels with the single exposed pixels, it will reduce a lot (though not all) the funkiness that can happen.
If it works like that, the strange effects that may take place are not possible at all. Saturated and reset pixels will represent the real information they would have achieved with a lower ISO gain during the entire exposure time. So they should not show ghosts or any other artifact, unless the reset time becomes comparable to the exposure time. Maybe chopped trails in moving spot lights.
The idea is not new, it is how the Fuji Super CCD worked. But this is much better since all pixels can work in 'multiexposure' mode while in Super CCD there were 2 kind of pixels, the 'regular' ones and a second set spatially overlapped with a sensitivity 3.6 stops lower that played the role of highlight saver.
Regards