645D lost the competition
Posted 10/03/2010 - 20:58
Link
As i recall,It is about 2 or 3 years since Kodak announced the development of these new Sensors. Kodak developed the Bayer pattern over 30 years ago which is used in the majority of current Sensors, They designed this new generation of sensor from the ground up. The Photosites are claimed to be 50% smaller and instead of the usual 1 red, 1 blue, and 2 green sensitive pattern they replaced one of the greens with one that is sensitive to all colours. So as it not blocking out any wavelengths of light it makes it more sensitive. So ultimately you should get less noise, finer detail and improved low light performance. They said at the time that they intended to use them in Kodak Camera`s initially before being offered to other manufacturers.
Regards - Pete
Regards - Pete
Posted 10/03/2010 - 21:39
Link
As an ex Oly E-1 owner, I can also add that many Olympus owners think the original 4/3 cameras with Kodak sensors produced the nicest colours of any of the 4/3 bodies. Noise, sadly, was a different issue with the E-1. If Kodak have managed to work similar magic with this sensor, I'm wondering if it may be capable of astonishing results - in which case, pitching this camera at this price level when compared to the opposition may be quite an interesting move. I wouldn't be surprised if it was more successful than anyone might have predicted.
Posted 11/03/2010 - 00:19
Link
Mongoose wrote:
According to the spec sheet for the Kodak sensor which matches the specs of the 645D most closely (we don't know if that's the actual sensor to be used, but the specs will be similar) it has a saturation charge of ~41 thousand electrons.
Since you can't have part of an electron, there are 41000 possible levels of output from that sensor. 2^14 is ~16000 levels while 2^16 is ~65000 levels. So assuming that sensor (which may or may not be the one in use in either camera), 16bit should be better than 14 although not quite as much as might first appear.
...
So in short, yes, you're right, we're too close to quantum physics to pretend that true analogue signals exist in nature at this point, but the step up from 14 to 16 bit does appear to be justifiable from the sensor specifications.
I understand what you're saying but don't you have to take the read noise into account? I'd found MostlyHarmless wrote:
Neal
...Mongoose wrote:
the sensor is analogue, it produces a voltage signal which is measured by the A/D converter. The sensor itself doesn't have a "native" bit level.
Sorry, I wasn't clear. ...the sensor is analogue, it produces a voltage signal which is measured by the A/D converter. The sensor itself doesn't have a "native" bit level.
Neal
According to the spec sheet for the Kodak sensor which matches the specs of the 645D most closely (we don't know if that's the actual sensor to be used, but the specs will be similar) it has a saturation charge of ~41 thousand electrons.
Since you can't have part of an electron, there are 41000 possible levels of output from that sensor. 2^14 is ~16000 levels while 2^16 is ~65000 levels. So assuming that sensor (which may or may not be the one in use in either camera), 16bit should be better than 14 although not quite as much as might first appear.
...
So in short, yes, you're right, we're too close to quantum physics to pretend that true analogue signals exist in nature at this point, but the step up from 14 to 16 bit does appear to be justifiable from the sensor specifications.
this sensor, which quotes 42Ke at saturation - but also 13e read noise which (assuming my understanding of read noise is correct - effectively the error bar on how accurately you can read the no. of electrons in each well) means 14 is more than enough? In fact isn't 12 enough?
Posted 11/03/2010 - 08:12
Link
ahh I see what you're saying, good point.
42000/26 = ~1600 read noise size steps, which would make 12bit (4096 levels) more than enough.
It seems to me that it's still worthwhile to oversample somewhat beyond the size of the noise, after all when you're plotting a graph you don't quantise to the error bar size. There probably isn't a lot of benefit in going much beyond say 10 quanta per error bar size though (this is semi-educated guesswork, feel free to disagree) which just happens to be roughly 14bit = ~16000 steps.
So there's a theoretical benefit in going to 16 bit since you should be able to hold enough electrons per photosite to justify it, but the level of noise present in the sensor means there probably isn't much point.
I wonder if this exact discussion took place in the Pentax engineering dept a year or so ago....
42000/26 = ~1600 read noise size steps, which would make 12bit (4096 levels) more than enough.
It seems to me that it's still worthwhile to oversample somewhat beyond the size of the noise, after all when you're plotting a graph you don't quantise to the error bar size. There probably isn't a lot of benefit in going much beyond say 10 quanta per error bar size though (this is semi-educated guesswork, feel free to disagree) which just happens to be roughly 14bit = ~16000 steps.
So there's a theoretical benefit in going to 16 bit since you should be able to hold enough electrons per photosite to justify it, but the level of noise present in the sensor means there probably isn't much point.
I wonder if this exact discussion took place in the Pentax engineering dept a year or so ago....
you don't have to be mad to post here
but it does help
but it does help
Posted 11/03/2010 - 12:31
Link
Mongoose wrote:
... feel free to disagree....
No I think I'll agree - I'm on shaky ground when it comes to analogue and much happier with digital.... feel free to disagree....
Neal
Posted 11/03/2010 - 16:30
Link
Mongoose wrote:
ahh I see what you're saying, good point.
42000/26 = ~1600 read noise size steps, which would make 12bit (4096 levels) more than enough.
It seems to me that it's still worthwhile to oversample somewhat beyond the size of the noise, after all when you're plotting a graph you don't quantise to the error bar size. There probably isn't a lot of benefit in going much beyond say 10 quanta per error bar size though (this is semi-educated guesswork, feel free to disagree) which just happens to be roughly 14bit = ~16000 steps.
So there's a theoretical benefit in going to 16 bit since you should be able to hold enough electrons per photosite to justify it, but the level of noise present in the sensor means there probably isn't much point.
I wonder if this exact discussion took place in the Pentax engineering dept a year or so ago....
The discussion probably went along the lines of:ahh I see what you're saying, good point.
42000/26 = ~1600 read noise size steps, which would make 12bit (4096 levels) more than enough.
It seems to me that it's still worthwhile to oversample somewhat beyond the size of the noise, after all when you're plotting a graph you don't quantise to the error bar size. There probably isn't a lot of benefit in going much beyond say 10 quanta per error bar size though (this is semi-educated guesswork, feel free to disagree) which just happens to be roughly 14bit = ~16000 steps.
So there's a theoretical benefit in going to 16 bit since you should be able to hold enough electrons per photosite to justify it, but the level of noise present in the sensor means there probably isn't much point.
I wonder if this exact discussion took place in the Pentax engineering dept a year or so ago....
"Kodak have offered us a good price on a decent sensor, how many shall I say we want?"
If you can't say something nice about Pentax, you won't say anything at all.
Apparently.
Apparently.
Add Comment
To leave a comment - Log in to Pentax User or create a new account.
2891 posts
19 years
Wiltshire,
England
Einstein said something like "any inteligent fool can make something more complex, it takes bravery and a touch of genius to move in the oposite direction"
The technology you linked to has the apparent potential to make large format sensors much cheaper. That could really bring larger formats into the hands of whoever wants them.
Remember the days when an off brand SLR was £100 brand new body only? they could come again!
but it does help