snubbr.com

CMOS dynamic range
I was wondering if anyone could answer this question for me. What is the effective sensitivity of a CMOS sensor?.

I know they use 12-bit or 14-bit A/D converters, but I have heard that the sensors are really only effective to about 8-bit..

It seems to me that there has never been a time personally where RAW has yielded different results than JPEG, despite the exponential increase in dynamic range. If I mess up and blow out my highlights, they've always been blown out (or close enough to it), even if I saved as RAW and try to fix them.Thanks...

Comments (16)

Little Bear wrote:.

It seems to me that there has never been a time personally where RAWhas yielded different results than JPEG, despite the exponentialincrease in dynamic range..

"exponential increase" ??? Seems to me I usually seen describe as RAW providing a other stop or so of headroom..

For example:.

"Experience has told us that there is typically around 1 EV (one stop) of extra information available at the highlight end in RAW files and that a negative digital exposure compensation when converting such files can recover detail lost to over-exposure.".

Quote fromhttp://www.dpreview.com/reviews/canoneos40d/page20.asp.

If I mess up and blow out my highlights,they've always been blown out (or close enough to it), even if Isaved as RAW and try to fix them..

So you compared a JPEG shot to a RAW shot, same scene, same settings taken moments apart? And thus you're not seeing any "advantage" to highlight recovery with RAW versus JPEG?.

Good Day,Roonal.

'Money doesn't buy happiness, but it makes for an extravagant depression' by golf tournament sportscaster..

Comment #1

Roonal wrote:.

"exponential increase" ??? Seems to me I usually seen describe as RAWproviding a other stop or so of headroom..

A RAW file is 12-bit, meaning it's colors are defined on a scale of 0-4095, which is exponentially more than the 0-255 scale of 8-bit. That is all I meant..

For example:"Experience has told us that there is typically around 1 EV (onestop) of extra information available at the highlight end in RAWfiles and that a negative digital exposure compensation whenconverting such files can recover detail lost to over-exposure.".

This is good to know. I do not have that much experience with RAW, and I didn't know dpreview wrote this..

So you compared a JPEG shot to a RAW shot, same scene, same settingstaken moments apart? And thus you're not seeing any "advantage" tohighlight recovery with RAW versus JPEG?.

No, I haven't seen first hand that RAW has any real advantages over JPEG, because if something is blown out or close to it, it still looks quite crappy even if I tweak the RAW. I haven't, however, done extensive testing, so I am afraid I will take a life defining shot that is slightly blown out, and only taking it as RAW would save it..

I am still wondering how sensitive the CMOS sensor actually is, and if this 12-bit or 14-bit A/D and RAW files are actually justified. For example, is each photodiode in a Canon CMOS actually able to store an amount of photons that it can accurately measure on a scale of 0-4095?..

Comment #2

Little Bear wrote:.

Roonal wrote:.

"exponential increase" ??? Seems to me I usually seen describe as RAWproviding a other stop or so of headroom..

A RAW file is 12-bit, meaning it's colors are defined on a scale of0-4095, which is exponentially more than the 0-255 scale of 8-bit.That is all I meant..

Yes, but you have to realise that the information in jpg files are stored against the gamma curve. So, it's basically the same range of information..

I am still wondering how sensitive the CMOS sensor actually is, andif this 12-bit or 14-bit A/D and RAW files are actually justified.For example, is each photodiode in a Canon CMOS actually able tostore an amount of photons that it can accurately measure on a scaleof 0-4095?.

I don't know  They say there is a increase in dynamic range - but more bits might mean better tonality gradation, as well..

/d/n..

Comment #3

Devnull wrote:.

I am still wondering how sensitive the CMOS sensor actually is, andif this 12-bit or 14-bit A/D and RAW files are actually justified.For example, is each photodiode in a Canon CMOS actually able tostore an amount of photons that it can accurately measure on a scaleof 0-4095?.

Yes, the full well capacity of a good quality, large pixel is in the tens of thousands of charges..

I don't know  They say there is a increase in dynamic range - butmore bits might mean better tonality gradation, as well..

A sensor has a analog dynamic range that is limited by the full well capacity on the top and noise terms on the bottom. While the number of bits can not change the analog dynamic range of the sensor, the number of bits can limit the measureable and useable dynamic range after the analog to digital converter. For example, you can not measure more than 12 stops (usually a bit less) stops after a 12 bit converter even though there may be more stops of dynamic range in the analog signal. Going to 14 bits seems to allow a bit more of the analog dynamic range to be seen after the analog to digital conversion. How useable the extra measurable dynamic range is another question. I can envision situations where the finer tonal gradations would be useful..

/d/n.

Leonhttp://homepage.mac.com/leonwittwer/landscapes.htm..

Comment #4

Thanks everyone, that helped me understand it much better...

Comment #5

Little Bear wrote:.

I know they use 12-bit or 14-bit A/D converters, but I have heardthat the sensors are really only effective to about 8-bit..

The Canon flagship cameras use CMOS sensors as does the Sony R1. I know some of Nikon's DSLR's are also CMOS, just not sure which ones..

There really is no advantage or disadvantage of CMOS vs CCD. It depends a lot on the whole package (sensor, support electronics and firmware). So I would not worry about the sensor type so much as how the whole package performs..

It seems to me that there has never been a time personally where RAWhas yielded different results than JPEG, despite the exponentialincrease in dynamic range. If I mess up and blow out my highlights,they've always been blown out (or close enough to it), even if Isaved as RAW and try to fix them..

If it's blown, it's blown and raw cannot save you. Getting the exposure right is the most important thing in photography..

The greatest of mankind's criminals are those who delude themselves into thinking they have done 'the right thing.'- Rayna Butler..

Comment #6

Little Bear wrote:.

A RAW file is 12-bit, meaning it's colors are defined on a scale of0-4095, which is exponentially more than the 0-255 scale of 8-bit.That is all I meant..

Black is still black and white is still white. The number of bits simply defines how many discrete steps there are between black and white..

8 bits = 256 steps10 bits = 1024 steps12 bits = 4096 steps14 bits = 16384 steps16 bits = 65536 steps.

Dynamic range is something else. It is the number of stops of light between black and white. The number of bits used simply defines how many slices are used to represent that dynamic range..

The greatest of mankind's criminals are those who delude themselves into thinking they have done 'the right thing.'- Rayna Butler..

Comment #7

Early exponents of CMOS sensors claimed massive dynamic range etc. etc. but when the first practical ones were produced for consuner cameras they were no better than CCD sensors and that still remains the case..

On the subject of blown highlights, I will as a direct question without trying to be be rude, why can't you get your exposure right without blowing out the highlights? What metering scheme are you using? Try using the histogram to help balance the exposure..

Cheers..

Comment #8

Thank you for the input..

Just to clarify, I was not wondering what is superior: CMOS or CCD. I just wrote CMOS as an example because Canon's highest end cameras use a CMOS sensor..

Anyway, the problem is not frequent overexposure. When you shoot literally hundreds of shots a day, especially when on a time constraint such as travel, it is difficult to examine each image. Sometimes little things like the reflections on leaves are blown out. Although it does not detract from the image, it is something that I thought RAW might provide a solution to. In my experience, however, if something is blown out to the point of being white, then JPEG or RAW, it is white. This led me to believe that sensors can't really detect slight gradations of light, and despite their A/D converters, are really constricted to a scale that is as rough as 8-bit.This was just a curiosity question, rather than a question of technique...

Comment #9

Both CMOS and CCD sensors provide good images. The main issue is cost. CMOS has significant cost advantages and likely will dominate the large sensor market particularly for 35 mm and cropped versions..

Recent CMOS sensors have on sensor electronics that do a good job in reducing electronic and such noise. I'm not aware that CCD sensors have similar capabilities so CMOS sensors may be somewhat better.Leonhttp://homepage.mac.com/leonwittwer/landscapes.htm..

Comment #10

GodSpeaks wrote:.

Little Bear wrote:.

I know they use 12-bit or 14-bit A/D converters, but I have heardthat the sensors are really only effective to about 8-bit..

The Canon flagship cameras use CMOS sensors as does the Sony R1. Iknow some of Nikon's DSLR's are also CMOS, just not sure which ones..

D2x (which has almos the same sensor as the R1), the new D300 and D3..

There really is no advantage or disadvantage of CMOS vs CCD..

Well, there are both advantages and disadvantages to both CMOS and CCD, but I think you are saying that neither has an advantage where DR is concerned?.

Charlie DavisNikon 5700 & Sony R1HomePage: http://www.1derful.infoBridge Blog: http://www.here-ugo.com/BridgeBlog/..

Comment #11

Little Bear wrote:.

Sometimes little things like the reflections on leaves are blown out.Although it does not detract from the image, it is something that Ithought RAW might provide a solution to. In my experience, however,if something is blown out to the point of being white, then JPEG orRAW, it is white. This led me to believe that sensors can't reallydetect slight gradations of light, and despite their A/D converters,are really constricted to a scale that is as rough as 8-bit.This was just a curiosity question, rather than a question of technique..

I've read your several replies to your advisers and I conclude that you still fail to understand their comments. I'll probably fail too, but let me try anyway:.

1. There is NO relationship to the number of bits of quantification and dynamic range..

2. Image elements that are outside the quantification range of the A/D converter can not be "fixed"...period..

3. There is no requirement, nor should there be an expectation, that a JPEG image must quantify a narrower range of light values than a RAW image..

Many people don't have a clue what all this means. They grasp trivial facts, like that a specific camera has 12-bit A/D conversion, but produces 8-bit JPEG files and jump to entirely wrong conclusions. Or, on reading that one camera does 14-bit conversions, they decide that means it has better DR than all those other cameras with only 12-bit conversion..

Let me state that DR is not measured or stated in "bits". Doing so is like saying that the weight of a box of eggs is measure in "dozens". "Bits" and "dozens" are just numbers...they have no "units"! To describe DR, we need to reference some physical parameter...Lux, watts/unit area, Volts, optical density, etc. DR is very similar to "signal-to-noise ratio", which can be and often is expressed as db. But one "bit" can not be translated into ANY of these units. The relationship of a "bit" to any measured parameter is completely variable; a relationship that the design engineer chooses..

For example (a silly one), let's take two light values. The first is the reflected light from a white paper under the illiumination of a full moon. The second is the reflected light from the same white paper under the illumination of the sun. Just to be consistent, our location on the earth and the date will be the same. We'll also not allow clouds. We have two designers that are asked to create A/D converters to quantify these two brightness values.

The second designer, stupidly uses a 14-bit A/D converter. In both these designs, the minimum outout value (either 0 or 00000000000000) represents the brightness of the paper under moonlight and the maximum output value (either 1 or 11111111111111) represents the brightness of the paper under sunlight. Which design has the greatest dynamic range? Please ignore noise considerations and any limitations of the sensor (if there is one?)..

Charlie DavisNikon 5700 & Sony R1HomePage: http://www.1derful.infoBridge Blog: http://www.here-ugo.com/BridgeBlog/..

Comment #12

Clearly I mistitled this "CMOS dynamic range." My question was with regards to how accurately a sensor can quantify the tonal range of an image. I had thought that by using RAW, the leaves on the tree might have very fine gradations in them, but that JPEG would simply round them up to 255. Even though a 12-bit A/D converter quantifies things on a scale of 0-4095, I questioned the accuracy of such chips...

Comment #13

Chuxter wrote:.

1. There is NO relationship to the number of bits of quantificationand dynamic range..

I assume you mean the analog dynamic range of the sensor. The number of bits can limit the useable dynamic range after the A/D converter. If you have a 12 bit converter, you can not get more than 12 stops (or a bit less) of of useable dynamic range even though there might have been more than 12 stops of dynamic range in the analog sensor itself..

2. Image elements that are outside the quantification range of theA/D converter can not be "fixed"...period..

Yes..

3. There is no requirement, nor should there be an expectation, thata JPEG image must quantify a narrower range of light values than aRAW image..

In the three Canon DSLRs that I have used, the white point for jpg images is about a stop below the max that can be recorded and digitized. Thus I have had about a stop of useable dynamic range available with doing raw that I did not have when shooting jpg..

As the sensors get sufficiently small, jpg images must use all that is available from the sensor to get a good dynamic range..

Many people don't have a clue what all this means. They grasp trivialfacts, like that a specific camera has 12-bit A/D conversion, butproduces 8-bit JPEG files and jump to entirely wrong conclusions. Or,on reading that one camera does 14-bit conversions, they decide thatmeans it has better DR than all those other cameras with only 12-bitconversion..

It might be able to show more measureable DR after A/D conversion assuming the sensor had enough DR to begin with. Obviously, the A/D can not make DR if it is not already there in the analog sensor..

Leonhttp://homepage.mac.com/leonwittwer/landscapes.htm..

Comment #14

Thank you for the recent and former explanation Leon. You answered the questions I had in my mind and helped me understand these things, even if I didn't ask for them clearly...

Comment #15

Leon Wittwer wrote:.

Chuxter wrote:.

1. There is NO relationship to the number of bits of quantificationand dynamic range..

I assume you mean the analog dynamic range of the sensor. The numberof bits can limit the useable dynamic range after the A/D converter.If you have a 12 bit converter, you can not get more than 12 stops(or a bit less) of of useable dynamic range even though there mighthave been more than 12 stops of dynamic range in the analog sensoritself..

I think we agree...sorta. The key word in your last sentence is "useable". My statement was that it's possible to map ANY range of values to any A/D converter. It's done with amplifiers and scaling resistor networks. The results can be poor or worthless. For example, I can take your 12-bit A/D and digitize 1% of that 12-stops of DR.

Or I can use the 12-bits to digitize 20-stops of DR, even though the sensor has only an 8-bit range..

2. Image elements that are outside the quantification range of theA/D converter can not be "fixed"...period..

Yes..

3. There is no requirement, nor should there be an expectation, thata JPEG image must quantify a narrower range of light values than aRAW image..

In the three Canon DSLRs that I have used, the white point for jpgimages is about a stop below the max that can be recorded anddigitized. Thus I have had about a stop of useable dynamic rangeavailable with doing raw that I did not have when shooting jpg.As the sensors get sufficiently small, jpg images must use all thatis available from the sensor to get a good dynamic range..

Different cameras and different settings of the parameters on those cameras result in different amounts of "headroom". Your Canon experience is probably typical, but it can't be extrapolated accurately..

Many people don't have a clue what all this means. They grasp trivialfacts, like that a specific camera has 12-bit A/D conversion, butproduces 8-bit JPEG files and jump to entirely wrong conclusions. Or,on reading that one camera does 14-bit conversions, they decide thatmeans it has better DR than all those other cameras with only 12-bitconversion..

It might be able to show more measureable DR after A/D conversionassuming the sensor had enough DR to begin with. Obviously, the A/Dcan not make DR if it is not already there in the analog sensor..

One problem with these "word" experiments is that your lab and my lab are quite different. .

Charlie DavisNikon 5700 & Sony R1HomePage: http://www.1derful.infoBridge Blog: http://www.here-ugo.com/BridgeBlog/..

Comment #16

Click Here to View All...

Sponsored Amazon Deals:

1. Get big savings on Amazon warehouse deals.
2. Save up to 70% on Amazon Products.


This question was taken from a support group/message board and re-posted here so others can learn from it.

 

Categories: Home | Diet & Weight Management | Vitamins & Supplements | Herbs & Cleansing |

Sexual Health | Medifast Support | Nutrisystem Support | Medifast Questions |

Web Hosting | Web Hosts | Website Hosting | Hosting |

Web Hosting | GoDaddy | Digital Cameras | Best WebHosts |

Web Hosting FAQ | Web Hosts FAQ | Hosting FAQ | Hosting Group |

Hosting Questions | Camera Tips | Best Cameras To Buy | Best Cameras This Year |

Camera Q-A | Digital Cameras Q-A | Camera Forum | Nov 2010 - Cameras |

Oct 2010 - Cameras | Oct 2010 - DSLRs | Oct 2010 - Camera Tips | Sep 2010 - Cameras |

Sep 2010 - DSLRS | Sep 2010 - Camera Tips | Aug 2010 - Cameras | Aug 2010 - DSLR Tips |

Aug 2010 - Camera Tips | July 2010 - Cameras | July 2010 - Nikon Cameras | July 2010 - Canon Cameras |

July 2010 - Pentax Cameras | Medifast Recipes | Medifast Recipes Tips | Medifast Recipes Strategies |

Medifast Recipes Experiences | Medifast Recipes Group | Medifast Recipes Forum | Medifast Support Strategies |

Medifast Support Experiences |

 

(C) Copyright 2010 All rights reserved.