2 Questions about Dynamic Range and Pixel Design

[John Nouel]

Hi,

(please excuse my English)

This DP from this Film-school teach this (rather good initiative) Digital Video (Cinema!) Seminar. Although I did not take it, I've heard he has made the following statement (as fact) in class.

(Paraphrased) "Because of Pixel design, digital cameras can't have more than 5 stops of dynamic range."

I found this from the first time I heard it FALSE. Not because I'm a camera engineer but because of common sense. Not only for being in this forums or following RED, but for paying the least of the attention to the digital world. I've come to see cameras like the Red ONE, Epic, the Alexa, and others to achieve the obvious more than 5 stops in dynamic range. You don't have to know the exact theory of the earth orbiting around the sun to know it does and it is there (in case someone tells it's not and impossible).

It's funny the seminar have cameras like the Red ONE and Sony F3, and the professor has experience with them (That's why he is teaching it). I really think he is burning himself and it's just a matter of time with the Epic and such cameras taking such a big impact on the industry that people will just come to realize that he was wrong. But nevertheless I find it so hard to believe that still on this day he is capable of making (To a bunch of filmschool students and other DP Teachers who are trusting him their learning) that statement which is so obviously wrong and going away with it.

Whats gets me more than anything is that other DPs, teachers, and fellow students repeats him as if this was true

"No no, its impossible, that's why those cameras don't really have a high dynamic range. Is all Post Productions and tricks. You should see this class, he gives a great explanation"

-> So I would show a picture of a chart with the stops from the Alexa and Epic test and they would say

"Oh boy John, you fanboy fool. Well that's a trick, there are the 5 stops and in between its just corrected to look like more dynamic range"

(And I think to myself. Really? Disbelief has come to this childish thinking that everything and everyone that's not film is trying to cheat and lie to us? It's plain ignorant, or not thinking with fair logic. They're all tools, we can use it or not. But taking this position baffles me )


So here's the thing.

As there is very intelligent capable people here I'm asking the following 2 questions (And It'll be amazing if Mr. Nattress read this and give one of those great technical explanations):

First Question: "Where do this PIXEL DESIGN 5 STOPS line of thought comes from? It must be from a misunderstanding of fundamentals or actual fact of the DV age. I refuse to believe it's just invented" (This would be great to know)

Second Question: "How and why is this just not true" (Feel me in with all the technical engineer babble. So I can read it and read it and have a base to google more and learn).

Bonus: What variables like pixel design, size of pixel in a sensor, pixel count, filters, etc. affects how a sensor captures light.

In advance, Thank you for you time!

 

[Martin Stevens]

5 stops of DR comes from old poor NTSC video cameras.

All of the Red cameras have REAL DR......no tricks needed.

 

[Graeme Nattress]

It sounds like the confusion between the image that the sensor sees (and records) and the image you view on a monitor. Generally on a normal monitor, you don't want to show the viewer too much dynamic range as the image would be of very low contrast, look yucky, be aesthetically boring. THink what viewing a REDlogFilm (or LogC) image would look like.

I have no idea where "pixel design limits you to 5 stops DR" comes from as it's just plain wrong....

Why is it not true - because you can shoot a scene with a measured 12 stops of DR and see detail in the deepest shadows and the brightest highlights. Yes the shadows are noisy, but yes, you can still make out the detail in them. If you tolerate less noise, you can reduce that figure.... That's what DR is - what is the difference in stops between the brightest and darkest object you can shoot in the same scene and still see un-clipped detail at both ends.

If you think about it, a simple macbeth chart has 5 stops of DR between the bright and dark chip. To say Pixel design limits you to 5 stops is to say you can't capture a scene with more range than a macbeth chart. You can easily prove this out by taking a macbeth out on a sunny dark, capturing detail in the clouds, and some shadows that are beyond the macbeth range.

Graeme

 

[Nick Shaw]

I have no idea where "pixel design limits you to 5 stops DR" comes from as it's just plain wrong....

Maybe he read an article such as this one by Yuri Neyman, which says:

Observable “Flat” Dynamic Range of the most cameras is (including RED MX) is 5.7 Fstops

Personally I believe Yuri's statements such as these are based on some fairly fundamental misunderstandings.

 

[Graeme Nattress]

No, not fundamental mis-understandings. He's just plain wrong. Go read my CML discussion with him on it.

Graeme

 

[Nick Shaw]

No, not fundamental mis-understandings. He's just plain wrong. Go read my CML discussion with him on it.

Graeme

I was being polite! I participated in that CML discussion (and others) and gave up trying to debate with him!

His concept of Observable "Flat" Dynamic Range appears to be based on the somewhat circular argument that if you display an image using a curve which only shows 5.7 stops range on your monitor then you can only see 5.7 stops of dynamic range!

 

[Martin Stevens]

If Monitors were perfectly linear then an image of the Sun could blind you!

I think low contrast displays are much healthier.

 

[John Nouel]

Thank you all for your answers! Great input as usual