Slyv

Has a website at sylvainphoto.com
Joined on Jan 14, 2013

Comments

Total: 5, showing: 1 – 5
On article What you need to know about Sony's a7R III (622 comments in total)
In reply to:

Slyv: Wonder how they supposedly achieve 15 stops of dynamic range when you have 14-bits memory ?
The only way I see is a non-linear digitalization, but I cannot see this info anywhere.
Or they are lying, the detector can be 15 stops of DR, but the memory can't record it.

Okay. On another forum somebody answer my question partially and this is what I thought, they are using a non linear digitalization :

“I asked that very question in an exec interview there (stay tuned, hope to post the transcript of it sometime this week), and as you and I both suspected, the answer is that there's a gamma curve applied, so the 14 bits of A/D are non-linear and spread over a larger range of light levels.

I had a longish discussion with Richard Butler from DPR at the Sony briefing event, and he pointed out that you can throw away some effective bits on the highlight end, because they're noise anyway.” ....

Link | Posted on Oct 30, 2017 at 18:51 UTC
On article What you need to know about Sony's a7R III (622 comments in total)
In reply to:

Slyv: Wonder how they supposedly achieve 15 stops of dynamic range when you have 14-bits memory ?
The only way I see is a non-linear digitalization, but I cannot see this info anywhere.
Or they are lying, the detector can be 15 stops of DR, but the memory can't record it.

@Azathothh

14 bits -> 2^14 = 16384 tones per color channel = 14 stops of DR (max)

Link | Posted on Oct 30, 2017 at 15:11 UTC
On article What you need to know about Sony's a7R III (622 comments in total)
In reply to:

Slyv: Wonder how they supposedly achieve 15 stops of dynamic range when you have 14-bits memory ?
The only way I see is a non-linear digitalization, but I cannot see this info anywhere.
Or they are lying, the detector can be 15 stops of DR, but the memory can't record it.

The DR at the pixel level (before amplification and numeration) is the ratio between the full-well and the read out noise (or 1 if read out noise is lower than 1). In other word the ratio between the total electron capacity and the smallest electron quantity you can measure.
This can be higher than 15 stop for modern detector at low read out noise.
So, the dynamic range at pixel level gives the ability to disentangle, for instance, 2^15 level of grey, or tones but you need the memory to go with:

For instance imagine a (dummy, stupid) pixel of a full well of 16 electrons and a read out noise of 1e-, the dynamic range is then 16 -> 4 stops.
But if you have a 2 bits memory, for instance, you will be able to record only 2^2 = 4 tones out of it, not the 16.

It is however feasible if the AD convertor is not linear. For high flux, most of the bits are waisted to record the photon noise. One can make a AD convertor adaptative function to flux. But not without a huge cost.

Link | Posted on Oct 30, 2017 at 14:58 UTC
On article What you need to know about Sony's a7R III (622 comments in total)

Wonder how they supposedly achieve 15 stops of dynamic range when you have 14-bits memory ?
The only way I see is a non-linear digitalization, but I cannot see this info anywhere.
Or they are lying, the detector can be 15 stops of DR, but the memory can't record it.

Link | Posted on Oct 28, 2017 at 14:31 UTC as 14th comment | 9 replies
On article Capture One Fujifilm X-Trans Raw support tested (211 comments in total)
In reply to:

Digitall: Fuji please, put the option of DNG in your cameras. ;)

Not if Fuji provide a pre-processed DNG file. Apparently they deal well with there JPEG so one can imagine a DNG file demosaiced from a x-Trans to a normal Bayer CMOS like sensor to allow other third party to use the file correctly without having to re-invent the wheel.
At least if they do not want to bring that inside the camera, Fujy can propose a software to do it on computer.

Link | Posted on Jan 14, 2013 at 21:18 UTC
Total: 5, showing: 1 – 5