Pages: [1]   Go Down

Author Topic: Red and the end of analog ISO gain?  (Read 3775 times)

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Red and the end of analog ISO gain?
« on: April 27, 2018, 01:31:13 pm »

I just learnt something that might be old news to some of you: current Red video cameras do not vary gain in the analog domain with various settings for the ISO Exposure Index. That is, a given photo-electron count always goes to the same numerical level in the raw file, with the adjustment to suitable levels in the final displayed image all done in conversion from raw; See here:
http://www.red.com/learn/red-101/iso-speed-revisited

I cannot resist pointing out that as DXO Mark scores these things, the “true ISO” of a Red camera is the same at any “ISO” setting: it is the photo-site saturation based measure that the ISO 12232 standard describes as indicating the bottom of the sensor’s exposure latitude range.

P.S. That is, what the “ISO” setting does is:
1. Set the exposure index in the camera’s light meter.
2. Note that EI setting in the raw file.
3. Use that EI value if and when a default conversion from raw to a “display” output format is done.
« Last Edit: April 27, 2018, 01:39:48 pm by BJL »
Logged

Telecaster

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3686
Re: Red and the end of analog ISO gain?
« Reply #1 on: April 27, 2018, 03:28:37 pm »

My gain pedals are all analog.  ;D

Oops, wrong site!

-Dave-
Logged

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2370
    • The Last Word
Re: Red and the end of analog ISO gain?
« Reply #2 on: April 27, 2018, 03:47:04 pm »

I just learnt something that might be old news to some of you: current Red video cameras do not vary gain in the analog domain with various settings for the ISO Exposure Index. That is, a given photo-electron count always goes to the same numerical level in the raw file, with the adjustment to suitable levels in the final displayed image all done in conversion from raw;

Isn't that how P1 cameras worked 10 or 12 years ago?

That's how the GFX works above ISO 1600. But we're seeing more and more DR-Pix cameras, and I recently saw a patent for a sensor with three conversion gain settings.

Jim

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Re: Red and the end of analog ISO gain?
« Reply #3 on: April 27, 2018, 07:51:51 pm »

Isn't that how P1 cameras worked 10 or 12 years ago?

That's how the GFX works above ISO 1600. But we're seeing more and more DR-Pix cameras, and I recently saw a patent for a sensor with three conversion gain settings.

Jim
AFAIK, you are right on all points: and it fits a simple pattern: applying a greater amount of voltage gain at higher exposure index settings is only useful to the extent that it helps to raise the noise floor in that signal above noise sources that follow: during subsequent signal transmission or from the ADC itself. So:

- With CCD MF backs, the gain could only be applied after transfer off the CCD, so it was done right next to the ADC, and the ADC's were often very good, with better DR than the CCD itself, so variable gain was redundant.

- Other CCD DSLRs instead had less good ADCs (10 or 12 bit), so it did help to amplify a weak signal, at least for a few stops. But maybe not beyond about EI=800 or 1600, so often higher EI setting than that did not increase the gain further: no benefit in engineering more variability of gain than that, and potential disadvantages to doing so (the amps becoming noisier, or less linear?)

- Early CMOS sensors instead could apply gain very early (on-chip) and then had to transfer the signal to an off-board ADC. So there was more potential for noise to enter the signal after the gain, and plenty of evidence that the higher gain at higher EI was worthwhile—again up to about EI=1600, but not all the way to the extreme high EI values.

- Column parallel ADC again mostly eliminates noise entering between the gain and ADC, but the ADCs in some cases have less DR than the photosites (e.g. 12-bit ADC with 13 stop sensors?) so variable gain has some value. (As a further complication, some recent sensors can do 14-bit ADC at lower frame rates, dropping to 12-bit for higher frame rates, so maybe variable gain is of little or no value in the low frame-rate case, but is worth having in the hardware.) Yet again, the benefits seem to max out after a couple of stops, in the range about EI=400 to 1600.

- The latest twist is the new dual gain method (DR-Pix) which reduces the capacity (as in capacitance) of the photosites themselves at higher EI, and thus increases the voltage signal output from the photosites. That can reduce noise right in the photosites (less dark current?) so is worthwhile again!
Logged

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2370
    • The Last Word
Re: Red and the end of analog ISO gain?
« Reply #4 on: April 27, 2018, 11:58:45 pm »


- The latest twist is the new dual gain method (DR-Pix) which reduces the capacity (as in capacitance) of the photosites themselves at higher EI, and thus increases the voltage signal output from the photosites. That can reduce noise right in the photosites (less dark current?) so is worthwhile again!

Dr-Pix doesn't reduce dark current, AFAIK, but, by increasing the conversion gain, increases the voltage and thus reduces the proportional effect of voltage noise sources downstream from the photodiode.

Jim

Rob C

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 24074
Re: Red and the end of analog ISO gain?
« Reply #5 on: April 28, 2018, 06:04:38 am »

My gain pedals are all analog.  ;D

Oops, wrong site!

-Dave-

Yes, wrong site for you, but also wrong for me: were I to understand any of that stuff I would be terrified of pressing the button. In other words, how to cripple the freedom of the mind to roam where it pleases!

That said, just as well that some do understand it all, or I might be faced with the costs of Kodachrome again.

;-)

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Re: Red and the end of analog ISO gain?
« Reply #6 on: April 28, 2018, 09:00:22 am »

Yes, wrong site for you, but also wrong for me: were I to understand any of that stuff I would be terrified of pressing the button. In other words, how to cripple the freedom of the mind to roam where it pleases!

That said, just as well that some do understand it all, or I might be faced with the costs of Kodachrome again.

;-)
Surely you understood all the chemistry of Kodachrome exposure and development before you dared use it!? No, I am not worrying about this while actually doing photography.
Logged

Slobodan Blagojevic

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 18090
  • When everyone thinks the same, nobody thinks
    • My website
Re: Red and the end of analog ISO gain?
« Reply #7 on: April 28, 2018, 09:55:58 am »

Surely you understood all the chemistry of Kodachrome exposure and development before you dared use it!?...

Good point, BJL.

Then again, the two Kodachrome chemists were musicians, which explains why some of us swear Kodachrome could sing.  :)

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2370
    • The Last Word
Re: Red and the end of analog ISO gain?
« Reply #8 on: April 28, 2018, 10:10:24 am »

Then again, the two Kodachrome chemists were musicians, which explains why some of us swear Kodachrome could sing.  :)

There is a long-standing, and AFAIK, not-very-well-understood, nexus between photography and music. Ansel Adams, Paul Caponigro, Don Worth, Charles Cramer, and Huntington Witherill are a few that come to mind. I'm sure the LuLa folks can add more.

Jim

Rob C

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 24074
Re: Red and the end of analog ISO gain?
« Reply #9 on: April 28, 2018, 10:24:46 am »

Surely you understood all the chemistry of Kodachrome exposure and development before you dared use it!? No, I am not worrying about this while actually doing photography.

Only that it was different to Ektachrome, which I had to process as one of my duties in the industrial photo-unit where I was a trainee.

Later, as a working snapper, all that mattered was that I knew it's speed, could use an incident light meter and understood why it mattered in order for me to get pleasing skin tones. Its chemistry had nothing to do with me, and held even less interest for my mind. It's like driving a car: no need to know how to make one in order to drive one. I don't think of KWETG (knowing what exposure to give) as any sort of chemistry, at a stretch, it might slide - just - under the door of the Physics lab.

With B/W it was slightly different, because those films lent themselves to interpretation, much as I see the use of RAW today.

Of course one had to know about exposure, contrast, development and their combined effects on granularity and so on, but it didn't mean I had to know how to make the damned stuff from scratch, albeit the theory is pretty simple compared with what little I do know about colour materials beyond processing them and replenishing tanks.

To me, none of that is about photography, neither then nor today.
« Last Edit: April 28, 2018, 10:41:27 am by Rob C »
Logged

UlfKrentz

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 530
    • http://www.shoots.de
Re: Red and the end of analog ISO gain?
« Reply #10 on: April 28, 2018, 01:38:42 pm »

Yes, wrong site for you, but also wrong for me: were I to understand any of that stuff I would be terrified of pressing the button. In other words, how to cripple the freedom of the mind to roam where it pleases!

That said, just as well that some do understand it all, or I might be faced with the costs of Kodachrome again.

;-)

Rob,

It´s probably extremely easy for you :-) REDs ISO setting is simply dialing in an intended push / pull processing. (Actually with very similar effects like more grain etc) AFAIK it has always been this way, at least it already worked that way when I started shooting RED with a Scarlet X quite a long time ago. It has been confusing for first time RED shooters and often resulted in heavily underexposed clips when unexperienced users gained up the ISO (which practically has no effect at all, just meta data).
Cheers!

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Re: Red and the end of analog ISO gain?
« Reply #11 on: April 28, 2018, 07:08:01 pm »

... as a working snapper, all that mattered was that I knew it's speed, could use an incident light meter and understood why it mattered in order for me to get pleasing skin tones. Its chemistry had nothing to do with me, and held even less interest for my mind.
I quite agree, even when it comes to the digital equivalents, though for me it is a more modest knowing the chosen speed and how to use a reflected light meter (the camera's). The only added wrinkle is the extra choices offered by the Exposure index dial, which are a bit like a choice of which speed of film to use, or as some would say, how much push/pull processing to use.
It is mostly out of idle curiosity (perhaps an occupational hazard of being a mathematical scientist) that I want to understand the electronic, physical and mathematical basis of why the various camera settings and sensor designs have the effects that they do on the final displayed image. (That and a desire to have a scientific basis for disputing out-dated or flat-out wrong claims and predictions about how cameras are likely to perform under various conditions.)
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Re: Red and the end of analog ISO gain?
« Reply #12 on: April 28, 2018, 09:43:52 pm »

Dr-Pix doesn't reduce dark current, AFAIK, but, by increasing the conversion gain, increases the voltage and thus reduces the proportional effect of voltage noise sources downstream from the photodiode.

Jim
Could be: I was only speculating that the extra capacitance used in low EI mode is a source of dark current, but I suppose that is a small part of total read noise in typical “low light” scenarios anyway.

If anyone is interested, here is Aptina's white paper on DR-Pix
http://www.photonstophotos.net/Aptina/DR-Pix_WhitePaper.pdf

At the naive level of circuit diagrams, it looks easy enough in principle to add a second switchable capacitor to get three levels.
Logged

bcooter

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1520
Re: Red and the end of analog ISO gain?
« Reply #13 on: May 27, 2018, 05:48:38 pm »

Rob,

It´s probably extremely easy for you :-) REDs ISO setting is simply dialing in an intended push / pull processing. (Actually with very similar effects like more grain etc) AFAIK it has always been this way, at least it already worked that way when I started shooting RED with a Scarlet X quite a long time ago. It has been confusing for first time RED shooters and often resulted in heavily underexposed clips when unexperienced users gained up the ISO (which practically has no effect at all, just meta data).
Cheers!



Ulf, I agree with you about moving the ISO around.   Especially early on with the MX sensors, as they kept upping the firmware which seemed like every month.

Once they hit the last MX firmware upgrade, I think two years ago and combined with Resolve 10, then 12.5 it was like a new camera with a much different look in color and grain.

Early on with my MXs 640 iso was tops, set to the best compression, now they go to 800 easy, not that  I routinely go to 1000, but I can with very light noise reduction.

I think a lot of this is positive change is resolve and am now testing Resolve 14 which is much faster in loading, viewing, grading and debayering than the previous versions.

A lot faster.

I love(d) RED but they’re very strange.    Now they’ve changed their lineup with one camera body and three different sensors, obviously not interchangeable sensors.  The change to one body style wouldn’t bother me as long as they continue to service their past cameras, but we all know they dropped service on the MX cameras, so what’s next?   

Two weeks ago I filled out the order form for the Scarlet Dragon, as everyone raves about the Dragon Sensor, then a few days later saw that RED made a change and discontinued all but the three cameras mentioned.  I think the total order was $32,000 but lightly used if you search,  a fairly complete kit is half that, so I’ve held off.   I’ll call RED and see how long they’ll service this camera, but who knows.   What I do know is the electronic world drives me crazy.

On this shoot, one of the young camera assistants looked at my MXs and said wow, old school.   I kind of smiled and thought man, today there are movies shot around the world with film cameras that are 25 years old, but yea 8 or 9 year old electronic movie cameras are old school? 



I guess it’s just a matter of personal preference.   

IMO

BC
Logged

rdonson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3263
Re: Red and the end of analog ISO gain?
« Reply #14 on: May 27, 2018, 06:28:36 pm »

There is a long-standing, and AFAIK, not-very-well-understood, nexus between photography and music. Ansel Adams, Paul Caponigro, Don Worth, Charles Cramer, and Huntington Witherill are a few that come to mind. I'm sure the LuLa folks can add more.

Jim

I'd like to add a stellar musician, photographer and printer, Graham Nash!!!
Logged
Regards,
Ron

eronald

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6642
    • My gallery on Instagram
Re: Red and the end of analog ISO gain?
« Reply #15 on: May 28, 2018, 05:42:02 am »

I'd like to add a stellar musician, photographer and printer, Graham Nash!!!

People who just want results without any understanding of the process are rarely called Masters of their craft, they get called MBAs :)

Edmund
Logged
If you appreciate my blog posts help me by following on https://instagram.com/edmundronald

Rob C

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 24074
Re: Red and the end of analog ISO gain?
« Reply #16 on: May 28, 2018, 11:40:05 am »

Gosh! I didn't know true masters went to university and wore titles! I always thought they had a gift. Silly me; explains a lot.

eronald

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6642
    • My gallery on Instagram
Re: Red and the end of analog ISO gain?
« Reply #17 on: May 28, 2018, 03:41:50 pm »

Gosh! I didn't know true masters went to university and wore titles! I always thought they had a gift. Silly me; explains a lot.

Rather than indulge in sarcasm, why don't you just ask the OP or Jim whether this is relevant to you as a practical photographer?

Mostly everybody understands the general relevance of lens tests even if most of us cannot understand the details; why should sensor low ISO  characteristics be unimportant?

Do you really think that cine shooters have infinite lighting budgets and truckloads of lights and generators to throw at any project? The low ISO behaviour of RED cameras will define the lighting budget needs of a bunch of cine projects. *How* the low-light behavior is achieved is of interest to the geeks; *what* the low-light behaviour is will be of interest to every DP. 

Edmund
« Last Edit: May 28, 2018, 03:51:00 pm by eronald »
Logged
If you appreciate my blog posts help me by following on https://instagram.com/edmundronald

capital

  • Full Member
  • ***
  • Offline Offline
  • Posts: 222
    • Website
Re: Red and the end of analog ISO gain?
« Reply #18 on: May 28, 2018, 05:13:29 pm »

Rather than indulge in sarcasm, why don't you just ask the OP or Jim whether this is relevant to you as a practical photographer?

Mostly everybody understands the general relevance of lens tests even if most of us cannot understand the details; why should sensor low ISO  characteristics be unimportant?

Do you really think that cine shooters have infinite lighting budgets and truckloads of lights and generators to throw at any project? The low ISO behaviour of RED cameras will define the lighting budget needs of a bunch of cine projects. *How* the low-light behavior is achieved is of interest to the geeks; *what* the low-light behaviour is will be of interest to every DP. 

Edmund

In the behind the scenes film for the making of the Hobbit, one of the videographers refers to the Red cameras as 'color eaters', so when they shot the set-made forest scenes they applied excessive amounts of saturated paints. I think the point being, is they overcame the sensor limitation by artificially saturating the scene, instead of doing so digitally where the color signal would have been weaker.
Logged

Rob C

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 24074
Re: Red and the end of analog ISO gain?
« Reply #19 on: May 28, 2018, 05:29:10 pm »

Rather than indulge in sarcasm, why don't you just ask the OP or Jim whether this is relevant to you as a practical photographer?

Mostly everybody understands the general relevance of lens tests even if most of us cannot understand the details; why should sensor low ISO  characteristics be unimportant?

Do you really think that cine shooters have infinite lighting budgets and truckloads of lights and generators to throw at any project? The low ISO behaviour of RED cameras will define the lighting budget needs of a bunch of cine projects. *How* the low-light behavior is achieved is of interest to the geeks; *what* the low-light behaviour is will be of interest to every DP. 

Edmund


Goodness, what did I touch?

:-)
Pages: [1]   Go Up