[reh321]

Not sure I follow what you are saying... The highlight clipping can be a function that can increase dynamic range even further without adding to the file size needed to store all the extra bits. The Prime engine does a good job of that now, even among the competitors, as many still try to circumvent by using raw.

Dynamic Range requires that you store dark details and light details in the same file. That can be accomplished only by increasing number of bits used to represent a single word so it can either represent a very dark pixel {0000 0000 0000 0001] or a very bright pixel {1111 1111 1111 1110}. As a computer guy, I can assure you that there is no other way.

[ggascay]

Dynamic Range requires that you store dark details and light details in the same file. That can be accomplished only by increasing number of bits used to represent a single word so it can either represent a very dark pixel {0000 0000 0000 0001] or a very bright pixel {1111 1111 1111 1110}. As a computer guy, I can assure you that there is no other way.

What I am saying is that the 16 bits would not need to all be stored across the image, only used to increase dynamic range in the areas needed, such as in highlight protection mode. The K3 has an 86k RGB sensor and can apply clipping and highlight protection and metering for each of the 86k segments. By expanding the clipping and highlight protection to 16 bits - for the areas that need it - and not the rest of the scene, it would dramatically save on file size while increasing DR. This would work similar to the way multi-auto white balance works, but highlight clipping works before the output to Raw. As an engineer this is the way I would use the 16 bit sensor data.

[MikeyBugs95]

Ahem!...

Twitter

I have to admit, it is a very handsome camera. Definitely one of the reasons why I want to get one... But not the top reason of course!

[reh321]

What I am saying is that the 16 bits would not need to all be stored across the image, only used to increase dynamic range in the areas needed, such as in highlight protection mode. The K3 has an 86k RGB sensor and can apply clipping and highlight protection and metering for each of the 86k segments. By expanding the clipping and highlight protection to 16 bits - for the areas that need it - and not the rest of the scene, it would dramatically save on file size while increasing DR. This would work similar to the way multi-auto white balance works, but highlight clipping works before the output to Raw. As an engineer this is the way I would use the 16 bit sensor data.

As far as I am aware, methods like Dolby Sound are not applied to 'raw' data - 'raw' data saves actual sensor output, not interpreted output. You increase DR by increasing number of bits.

[phoebus]

Not sure I follow what you are saying... The highlight clipping can be a function that can increase dynamic range even further without adding to the file size needed to store all the extra bits. The Prime engine does a good job of that now, even among the competitors, as many still try to circumvent by using raw.

I think you are correct that the Pentax/Ricoh image processing does an exceptionally good job of squeezing as much DR as possible within the current 14 bit RAW image, though I think the conversation has been about what extra would might be achieved with 16 bit architecture. Clipping the highlights does not increase the DR, but preserves the shadows.

I think the aim is to prevent highlight clipping, which the latest sensor may achieve;

- if the analog/digital converter has 2 more bits of resolution, and
- the photosites can hold a sufficiently large range of electrons, and
- the the gain of the analog output from the photosite can accurately match the input range of the adc so that black is 0000000000000000 and white is 1111111111111111
..(or more accurately, each of the red, green and blue components of white)

then you can theoretically have 4 times as many steps between black and white, while not clipping the highlights or losing the shadow details.

In practice I suspect that signal noise will prevent all 16 bits being meaningful for a single exposure.

---------- Post added 31-12-19 at 17:01 ----------

Ahem!...

Twitter

Thanks @Mistral75 - Google translate doesn't give a good result of the tweet:

PENTAX "APS-C size best model" to be released next year Not even if not. The appearance of "Wolf Shonen PENTAX", which doesn't come out easily, saying "I'm going to come out"? Do I have to wait patiently?

Any Japanese speaker make better sense out of :

来年発売予定のPENTAXの"APS-C判最高機種"ですが、夏前に発売されればいいなあと期待していたものの、"その予定"より少しズレ込んで、ひょっとすると夏過ぎになる予感もしないでもない。
出るぞ出すぞ、と言って、なかなか出てこない「狼少年PENTAX」の面目躍如？　我慢して待つしかないか。

[ggascay]

As far as I am aware, methods like Dolby Sound are not applied to 'raw' data - 'raw' data saves actual sensor output, not interpreted output. You increase DR by increasing number of bits.

All of the modern Pentax cameras go through the Prime Processor and the Accelerator when available, to adjust the data BEFORE writing to the RAW output. You do not get direct sensor output data with the Raw format. Highlight correction and High ISO noise reduction are some of those things that are applied to the sensor data before being written to a raw file.

[Belnan]

I think you are correct that the Pentax/Ricoh image processing does an exceptionally good job of squeezing as much DR as possible within the current 14 bit RAW image, though I think the conversation has been about what extra would might be achieved with 16 bit architecture. Clipping the highlights does not increase the DR, but preserves the shadows.

I think the aim is to prevent highlight clipping, which the latest sensor may achieve;

- if the analog/digital converter has 2 more bits of resolution, and
- the photosites can hold a sufficiently large range of electrons, and
- the the gain of the analog output from the photosite can accurately match the input range of the adc so that black is 0000000000000000 and white is 1111111111111111
..(or more accurately, each of the red, green and blue components of white)

then you can theoretically have 4 times as many steps between black and white, while not clipping the highlights or losing the shadow details.

In practice I suspect that signal noise will prevent all 16 bits being meaningful for a single exposure.

---------- Post added 31-12-19 at 17:01 ----------



Thanks @Mistral75 - Google translate doesn't give a good result of the tweet:

PENTAX "APS-C size best model" to be released next year Not even if not. The appearance of "Wolf Shonen PENTAX", which doesn't come out easily, saying "I'm going to come out"? Do I have to wait patiently?

Any Japanese speaker make better sense out of :

来年発売予定のPENTAXの"APS-C判最高機種"ですが、夏前に発売されればいいなあと期待していたものの、"その予定"より少しズレ込んで、ひょっとすると夏過ぎになる予感もしないでもない。
出るぞ出すぞ、と言って、なかなか出てこない「狼少年PENTAX」の面目躍如？　我慢して待つしかないか。

I think the tweet suggests they believe that it could come out before summer 2020.

[reh321]

All of the modern Pentax cameras go through the Prime Processor and the Accelerator when available, to adjust the data BEFORE writing to the RAW output. You do not get direct sensor output data with the Raw format. Highlight correction and High ISO noise reduction are some of those things that are applied to the sensor data before being written to a raw file.

Yes, I am aware of the data path. My point is that people expect that a 'raw' file will resemble sensor output, not with some parts scaled up or down - and 16 bits give another factor of four in range of the image stored. It was sensible for them to increase that range in storage when they increased range in detection.

[clickclick]

Much ado about nothing, indeed. And how many responses can you make to “we still don’t know anything, but my random guess is...” posts?

-Eric

If you filter just five posters you cut down 80% of the nonsense. And I am aware I could be one of the five (but I haven’t hit the contest monthly total yet).

Yes, and yes, and almost there aka me too!

[Ian Stuart Forsyth]

Highlight correction and High ISO noise reduction are some of those things that are applied to the sensor data before being written to a raw file.

If you have clipped highlights there is no way to correct for them. think of the sensor as a 1m ruler and if try and go outside of that ruler to measure 1.05 or 2.0m how can that sensor ( the measuring devise) know how much outside of that 1m to apply corrections for.
1.05 or 2.00 the sensor will not know the difference.

The highlight correction headroom we normally see with raw files in pentax cameras has not really change from the first found in the ds model to what we have today. There is a highlight headroom developers have design into how they use the sensor and how they write the data that is collected and stored within the raw file and this has not really changed with pentax DLSR's from the start.

You can increase the headroom for highlight protection you would need to decrease the exposure and then apply a correction to bring the lightness of the photo back into the tonal range, but this has a negative affect on how much DR your are going to capture and store. DR is closely tied to how much light you can store so is you are going to decrease that amount light then you raise the bottom floor and your DR.

For Pentax the highlight headroom in the raw file is around 1 stop from full saturation and this is varies over the 3 color channels.

[Belnan]

May 2020 be the even better than 2019 for all of us Pentax fans. Now what is on the wishlist?

[mlt]

2020 is here and hopefully an auspicious (in a good way) year for Pentax and its fans. Looking forward to the announcements to come.

[monochrome]

2020 is here and hopefully an auspicious (in a good way) year for Pentax and its fans. Looking forward to the announcements to come.

K-New will be released this year !!!

[reh321]

K-New will be released this year !!!

and it will work correctly, a detail sometimes lost in all these words.

[TedH42]

and it will work correctly, a detail sometimes lost in all these words.

So we needn't fear being an "early adopter"?