HDR on MF digital backs

[Bob]

In a fruitless attempt to satisfy everyone oke:
I am starting a thread on HDR as it relates to Digital backs and how it may be best implemented if it is desired.
Please chime in.
thanks
-bob

 

[Stefan Steib]

The beauty of the Rambus Idea is NOT to do an HDR.
The same could be reached with timesliced control of overexposed pixels. But not for the full image just for the ones that get overexposed. An intermediate database that is saving several steps backwards (will need some ram though - the buffer ?) and get the last data before the pixel reaches full saturation after a pixel was overexposed. the file operation could be partial and would result in a complete control of overexposure no matter how bright the brightest spot is. There will be other drawbacks like less contrast if to much of the image will be overexposed, but to control highlights this would be perfect.

I am not an engineer and I suspect that blooming control today is already doing similar things in modern electronics and software of the backs, but there may be a way to implement the idea to have some kind of a chosable Anti Clipping automatic mode. This would also do a great deal for the people who are fans for expose to the right, reducing the danger of highlight clipping.

Regards
Stefan

 

[GrahamWelland]

I'm not sure how you'd do it but getting more DR in a single shot is much more interesting to me than more resolution.

Stefan's point about techniques for capturing intermediate exposures during a single shot sounds interesting. Wouldn't that require new technology compared to current MF chips though as its my understanding that they can only be read in darkness?

 

[Stefan Steib]

Hi Graham

Yes, you are right that rules out CCD´s............
I forgot about that. So again - a CMOS is needed.

Hmm. so far for the Firmware upgrade.....

But at least it would be a choice for Camera makers who are in control
of Chip and software, like Sony,Canon or Nikon.
Leica and Pentax if they go CMOS could implement this.
Sony has already filed patents for a global shutter capable chip which probably is also able to get the needed readout speed for intermediate exposure control.

Regards
Stefan

 

[scatesmd]

Hi Stefan,

There is another reason this technology applies to CMOS. Ultimately, the dynamic range could be increased exponentially.

It is because each pixel in a CMOS chip has its own electronics. In terms of what we have now, this is a questionable advantage, as CCD chips intrinsically have a higher DR and less noise than a CMOS chip. Color can be more uniform in a CCD device. Since each pixel of a CMOS chip has its own converter and amplifier, photons coming in are partially blocked by electronics and it is very difficult to match the output of every amplifier on a CMOS chip. The CCD amp and converter each are used for a bank of pixels, so each sees the same electronics, increasing uniformity and accuracy of output.

An advantage of CMOS, though, is that in theory every amp on every pixel can have its gain changed independently in real time, which is not possible with current CCD designs. Dynamic range can be adjusted all across the chip, dialing up gain for shadows and down for highlights on a pixel by pixel basis. Current CMOS chips adjust gain over the entire chip at once, but the opportunity is there to change gain as required for each pixel. Already we see how CMOS chips can have gain adjusted for high ISO use overall compared to CCD, but CMOS pixels allow for individual control as well.

I have read articles suggesting DR increases to over 100 using this technique in the future using CMOS chips.

Hope this made sense,

Steve

 

[Jan Brittenson]

If the clipping is due to the well saturating, then tweaking the readout design isn't going to make any difference. What you need is modify the photodetector so its sensitivity (here: fill rate) progressively drops as it fills up. It already does, but there may be ways to increase it. With CMOS it should be possible to add decay around the top 2-3 stops (by opening a leak from the well to the drain) to create a shouldering effect. The software has to be aware of this so it doesn't try to determine chroma, because it will be heavily desaturated. I can just let the chroma fade away.

 

[Stefan Steib]

Hi Stefan,

There is another reason this technology applies to CMOS. Ultimately, the dynamic range could be increased exponentially.

It is because each pixel in a CMOS chip has its own electronics. In terms of what we have now, this is a questionable advantage, as CCD chips intrinsically have a higher DR and less noise than a CMOS chip. Color can be more uniform in a CCD device. Since each pixel of a CMOS chip has its own converter and amplifier, photons coming in are partially blocked by electronics and it is very difficult to match the output of every amplifier on a CMOS chip. The CCD amp and converter each are used for a bank of pixels, so each sees the same electronics, increasing uniformity and accuracy of output.

An advantage of CMOS, though, is that in theory every amp on every pixel can have its gain changed independently in real time, which is not possible with current CCD designs. Dynamic range can be adjusted all across the chip, dialing up gain for shadows and down for highlights on a pixel by pixel basis. Current CMOS chips adjust gain over the entire chip at once, but the opportunity is there to change gain as required for each pixel. Already we see how CMOS chips can have gain adjusted for high ISO use overall compared to CCD, but CMOS pixels allow for individual control as well.

I have read articles suggesting DR increases to over 100 using this technique in the future using CMOS chips.

Hope this made sense,

Steve

That makes a lot of sense ! Thank you Steve !
this and the global shutter capability would make more sense to me than 100 Mpix or more. Probably 20-40 Mpix and a fully working 17/18 or maybe more stops DR would make many peoples days for Pro work.

Regards
Stefan

 

[Stefan Steib]

If the clipping is due to the well saturating, then tweaking the readout design isn't going to make any difference. What you need is modify the photodetector so its sensitivity (here: fill rate) progressively drops as it fills up. It already does, but there may be ways to increase it. With CMOS it should be possible to add decay around the top 2-3 stops (by opening a leak from the well to the drain) to create a shouldering effect. The software has to be aware of this so it doesn't try to determine chroma, because it will be heavily desaturated. I can just let the chroma fade away.

Thanks Jan ! Is that about similar as limiting lights (or shadows) at scanning in Software by value ? Guess so ?

we are still at the beginning I´d say. Digital Photography is a young technology and there are so many tricks not even thought about. I´m a bit frustrated though we do find these advantages in the smaller cameras and not in the so called Highend stuff nowadays.

Regards
Stefan