Very Very Very new, newbie

[Colin Southern]

What does this mean?

Hi Ash,

Dynamic Range is the range of the brightest to the darkest detail that a camera can capture - and it changes depending on the ISO used. At their lowest ("base") ISO most modern DSLR cameras can capture around 12 or 13 stops.

A purely reflective scene (eg bride in a white dress standing next to a groom in a black suit -- both in the sun) requires about 4 stops - so "4 fits into 12 or 13 really easily" and a camera would have no trouble capturing that scene faithfully). If we expand on that theme - if the bride & groom were in the sun but behind them was the groom's father also in a black suit, but he's in the shade, then that may typically require another 3 stops to capture suit details ... so then the scene would have a dynamic range of 7 (still easily doable). If we have a different type of scene - say we're photographing a house for sale; we're photographing inside and the sun is shining outside then inside you may still need 7 stops to capture reflective and shadow detail, but the sun shining on objects outside (viewable through the window) may require another 3 stops (or more) - so we're up to around 10 stops. Still "doable", but that shadow detail inside in now only 2 or 3 stops above the sensor noise floor (so not very clean data). And as luck wouldn't have it, we've forgotten our tripod and have to shoot hand-held - and at ISO 100 the shutterspeed it too low inside - so we use ISO 800 to raise the shutterspeed, and this drops the camera's dynamic range to 9 or 10. Houston, we have a problem (an 11 stop DR scene isn't going to be captured by a camera with an effective 9 or 10 stop capability).

Why does the sensor dynamic range decrease with increasing ISO I hear you ask (as per the graph)? Basically it all comes down to the physical/electrical capabilities of the camera sensor; there is a maximum amount of light that they can record without additional light being too much to handle (called the saturation point) and because all sensors have inherent noise at low levels (think of it as "mud at the bottom of the well") there's a point at the other end where the signal and the noise are indistinguishable. At base ISO (typically 100 for Canon and 200 for many Nikon cameras) we get the full range of signals possible, but when we increase the ISO (say to 200 for Canon), the physical characteristics of the sensor doesn't change (because they're part of it's physical makeup); all that happens is that when you double the ISO the shutter is open for 1/2 the time - letting in 1/2 as much light - and the camera is thus working 1 stop closer to the noise floor. After the shot is taken the signal (including the noise) is simply amplified up to the point where it would have been had it have been shot at base ISO. So if the camera started with a base ISO of 100 and had a dynamic range of 12 stops at that ISO then at - say - ISO 400 it's dynamic range (commonly called "DR") will have dropped to 10 stops; not a problem if it's a 4 stop DR scene; big problem if it's a 13 stop DR scene.

Sooooooo ...

In practice, the higher the ISO that you select, the lower the DR you're able to capture. Photography is all about managing limitations, and this one is a classic. You need to be able to evaluate the DR of the scene you want to capture. If it's a snatch and grab of a sportsman where you need a high shutterspeed to freeze the motion then you'll probably be OK at a higher ISO because that type of scene will mostly be reflective and have a low DR (although if you're capturing sky it may well increase it). Conversely, capturing a sunset where you're shooting into the light but still want to retain foreground shadow detail often requires a HUGE DR; so scenes like that are always shot at base ISO (well they are if you want to have a hope of capturing them properly!).

That's the 3 minute intro to DR - in practice it's pretty much like I've described, but with a few speed-balls:

- In practice we sometimes use GND (Graduated Neutral Density) Filters to compress the dynamic range of a scene into something the camera can better handle (eg it knocks back the brighter (usually sky) portions). Another technique is to take a bracket of shots covering different portions of the scenes brightness and combine them on our computers (HDRI) (High Dynamic Range Imagery)

- In practice, most camera exposure metering adds a safety margin above the highlights; for low DR scenes at base ISO this is fine, but for higher DR scenes and higher ISO modes we need to add what's called EC (exposure compensation) to eliminate this safety margin (it's a luxury we can't afford) (Called ETTR - Expose to the Right) (referring to the histogram) (not normally needed, but some people like to do it anyway because on paper it gives a signal-to-noise ratio advantage, but in practice it can introduce other issues).

- ISO noise is very small and is mostly noticeable when viewing an image at 100% magnification; easy answer is "don't view your images at 100% magnification" (do you check your carpets are properly vacuumed by going over every inch with a microsope?) (me neither!)

- Noise in dark tones usually isn't noticeable anyway - and often the tones can be easily clipped (eg by raising the black clipping point in a night scene to practically eliminate noise).

- Many cameras have extreme high-ISO performance that doesn't appear to follow the linear relationship between ISO and DR that I've tried to explain. Usually this is due to noise reduction that they've applied to JPEG images. It's effective, but it involves "trickery" as noise reduction always involves loss of detail (which may or may not be significant or obvious).

All in all it's a can of worms, but any photographer who knows his stuff will understand this to the point where it helps a lot in making the best compromise, as photography is always a compromise. Generally ISO noise is by far the lesser evil when the alternatives are camera shake / subject motion from too low a shutterspeed or insufficient DoF from too wide an aperture; so if you NEED a higher ISO (with it's reducing effect on DR then use it), but remember that it'll "take something away" somewhere else. Whether that "something" is significant or not is the $6,000,000 question!

Hope this helps!

(Hope even more that it answers the question!)

[Iceash]

Hi Ash,

Dynamic Range is the range of the brightest to the darkest detail that a camera can capture - and it changes depending on the ISO used. At their lowest ("base") ISO most modern DSLR cameras can capture around 12 or 13 stops.

A purely reflective scene (eg bride in a white dress standing next to a groom in a black suit -- both in the sun) requires about 4 stops - so "4 fits into 12 or 13 really easily" and a camera would have no trouble capturing that scene faithfully). If we expand on that theme - if the bride & groom were in the sun but behind them was the groom's father also in a black suit, but he's in the shade, then that may typically require another 3 stops to capture suit details ... so then the scene would have a dynamic range of 7 (still easily doable). If we have a different type of scene - say we're photographing a house for sale; we're photographing inside and the sun is shining outside then inside you may still need 7 stops to capture reflective and shadow detail, but the sun shining on objects outside (viewable through the window) may require another 3 stops (or more) - so we're up to around 10 stops. Still "doable", but that shadow detail inside in now only 2 or 3 stops above the sensor noise floor (so not very clean data). And as luck wouldn't have it, we've forgotten our tripod and have to shoot hand-held - and at ISO 100 the shutterspeed it too low inside - so we use ISO 800 to raise the shutterspeed, and this drops the camera's dynamic range to 9 or 10. Houston, we have a problem (an 11 stop DR scene isn't going to be captured by a camera with an effective 9 or 10 stop capability).

Why does the sensor dynamic range decrease with increasing ISO I hear you ask (as per the graph)? Basically it all comes down to the physical/electrical capabilities of the camera sensor; there is a maximum amount of light that they can record without additional light being too much to handle (called the saturation point) and because all sensors have inherent noise at low levels (think of it as "mud at the bottom of the well") there's a point at the other end where the signal and the noise are indistinguishable. At base ISO (typically 100 for Canon and 200 for many Nikon cameras) we get the full range of signals possible, but when we increase the ISO (say to 200 for Canon), the physical characteristics of the sensor doesn't change (because they're part of it's physical makeup); all that happens is that when you double the ISO the shutter is open for 1/2 the time - letting in 1/2 as much light - and the camera is thus working 1 stop closer to the noise floor. After the shot is taken the signal (including the noise) is simply amplified up to the point where it would have been had it have been shot at base ISO. So if the camera started with a base ISO of 100 and had a dynamic range of 12 stops at that ISO then at - say - ISO 400 it's dynamic range (commonly called "DR") will have dropped to 10 stops; not a problem if it's a 4 stop DR scene; big problem if it's a 13 stop DR scene.

Sooooooo ...

In practice, the higher the ISO that you select, the lower the DR you're able to capture. Photography is all about managing limitations, and this one is a classic. You need to be able to evaluate the DR of the scene you want to capture. If it's a snatch and grab of a sportsman where you need a high shutterspeed to freeze the motion then you'll probably be OK at a higher ISO because that type of scene will mostly be reflective and have a low DR (although if you're capturing sky it may well increase it). Conversely, capturing a sunset where you're shooting into the light but still want to retain foreground shadow detail often requires a HUGE DR; so scenes like that are always shot at base ISO (well they are if you want to have a hope of capturing them properly!).

That's the 3 minute intro to DR - in practice it's pretty much like I've described, but with a few speed-balls:

- In practice we sometimes use GND (Graduated Neutral Density) Filters to compress the dynamic range of a scene into something the camera can better handle (eg it knocks back the brighter (usually sky) portions). Another technique is to take a bracket of shots covering different portions of the scenes brightness and combine them on our computers (HDRI) (High Dynamic Range Imagery)

- In practice, most camera exposure metering adds a safety margin above the highlights; for low DR scenes at base ISO this is fine, but for higher DR scenes and higher ISO modes we need to add what's called EC (exposure compensation) to eliminate this safety margin (it's a luxury we can't afford) (Called ETTR - Expose to the Right) (referring to the histogram) (not normally needed, but some people like to do it anyway because on paper it gives a signal-to-noise ratio advantage, but in practice it can introduce other issues).

- ISO noise is very small and is mostly noticeable when viewing an image at 100% magnification; easy answer is "don't view your images at 100% magnification" (do you check your carpets are properly vacuumed by going over every inch with a microsope?) (me neither!)

- Noise in dark tones usually isn't noticeable anyway - and often the tones can be easily clipped (eg by raising the black clipping point in a night scene to practically eliminate noise).

- Many cameras have extreme high-ISO performance that doesn't appear to follow the linear relationship between ISO and DR that I've tried to explain. Usually this is due to noise reduction that they've applied to JPEG images. It's effective, but it involves "trickery" as noise reduction always involves loss of detail (which may or may not be significant or obvious).

All in all it's a can of worms, but any photographer who knows his stuff will understand this to the point where it helps a lot in making the best compromise, as photography is always a compromise. Generally ISO noise is by far the lesser evil when the alternatives are camera shake / subject motion from too low a shutterspeed or insufficient DoF from too wide an aperture; so if you NEED a higher ISO (with it's reducing effect on DR then use it), but remember that it'll "take something away" somewhere else. Whether that "something" is significant or not is the $6,000,000 question!

Hope this helps!

(Hope even more that it answers the question!)

OMG! Now that's a write up! Fantastic,blown away. Didn't even know such a thing even existed. So the gist is my d3200 nearly has 13 stops at ISO 100?

[Donald]

OMG! Now that's a write up! Fantastic,blown away.

Yep, you've got to admit you get good value for your membership fee here on CiC!

[Colin Southern]

OMG! Now that's a write up! Fantastic,blown away. Didn't even know such a thing even existed. So the gist is my d3200 nearly has 13 stops at ISO 100?

In short (for once!) ...

... Yes

[Colin Southern]

Yep, you've got to admit you get good value for your membership fee here on CiC!

In the interests of full disclosure I guess we should probably reveal that we get paid double for longer posts like that.

That reminds me - I'd better send in my bill for this week. What's double nothing?

[xpatUSA]

So ted, what's your typical shot you go for?

Thanks for your interest, Ash. My primary hobby is watch repair and collecting, so I'd have to say close-ups of those and other table-top stuff. I live in the country and try to shoot flora and fauna, more snapshots than great works of art. In photography, it's the technical part that fascinates. Just started IR - a whole new ball game!

cheers,

[xpatUSA]

I make a million statements a day - you'd have to give me a bit of a hint before I know which one you're referring to.

Sure, Colin,

Your post, #31 in this thread:

"It's largely a reflective scene (thus around 4 stops DR required) and at ISO 400 a Nikon D3200 would have a DR of around 11 stops."

cheers,

[Colin Southern]

Sure, Colin,

Your post, #31 in this thread:

"It's largely a reflective scene (thus around 4 stops DR required) and at ISO 400 a Nikon D3200 would have a DR of around 11 stops."

cheers,

Well I'm not sure what your issue is with that statement Ted, surely you can't be arguing that reflective scenes aren't around 4 stops (because just about the entire world knows that), and surely you can't be arguing about the DR of a Nikon D3200 at ISO 400 because I posted you a link and included a screen shot showing that I was EXACTLY right.

[xpatUSA]

Well I'm not sure what your issue is with that statement Ted, surely you can't be arguing that reflective scenes aren't around 4 stops (because just about the entire world knows that) . . .

Surely you haven't missed my response in post #33 to that statement:

You're right, Colin, image DR was indeed about 3-1/2 stops.

"You're right, Colin" meant that I had already agreed with you!

And surely you didn't miss the clue that, by bolding part of the quote in my post #47, I was responding to your Nikon D3200 DR claim, not the DR of scene.

. . . and surely you can't be arguing about the DR of a Nikon D3200 at ISO 400 because I posted you a link and included a screen shot showing that I was EXACTLY right.

Your premise is that DXOMark's tests are the only valid ones in the genre, to the exclusion of such as DPR. That is your narrow view and you are, of course, entitled to it. I suggest that we let the subject of Nikon DRs lay - since nobody else is coming in to provide a concensus one way or the other.

The sad part is that I'm going to be thought of as the bad guy in this exchange because I have been forced to descend to the petty level of quoting post numbers and trying to keep the sequence of who said what and when on the record. So, continuing in that role:

1) You said the scene DR was about 4 stops.

2) I agreed.

3) You said the Nikon DR was about 11 stops but provided no reference.

4) I disagreed and provided a reference.

5) You came back with a reference to back up your claim of 11 stops.

6) Then you became not sure what my issue was with the scene DR, in spite of 1) and 2) above!

I know it's hard for you but I think you should just let it drop, to be honest. However, I have absolutely no doubt that the last word will be yours.

[Colin Southern]

Your premise is that DXOMark's tests are the only valid ones in the genre, to the exclusion of such as DPR.

If you'd studied your DPR review more thoroughly, you'd have noted that the figure you quote is for JPEGs where the DR of the resultant file depends on the mode that the camera is in. The DxO measure what the camera is capable of - thus, RAW files.

Going back to the beginning, the point of my original response to your post was to debunk your rather silly notion that the camera may have been DR limited capturing that scene at 400 ISO -- so be it 9 stops (JPEG) or 11 stops (RAW), clearly, a reflective scene is going to fit -- with ease. Could you not appreciate that that was a low DR reflective scene without measuring it?

That is your narrow view and you are, of course, entitled to it. I suggest that we let the subject of Nikon DRs lay - since nobody else is coming in to provide a concensus one way or the other.

A consensus isn't going to change the facts regardless, but it's obvious to me that you have absolutely no idea of how much of a "biggie" DxO Labs are in standardised camera performance testing. It wouldn't even surprise me if DPReview used DxO equipment to conduct the tests. Trying to blow-off DxO's results for camera testing is much like trying to blow-off the international standard definition of a metre because it you get a different measurement with a ruler you bought at the supermarket for $2. It's just absurd.

1) You said the scene DR was about 4 stops.

Yes.

2) I agreed.

Yes, which makes me wonder why you made the comment about the possibility of the camera being DR limited for that scene in the first place.

3) You said the Nikon DR was about 11 stops but provided no reference.

Yes

4) I disagreed and provided a reference.

Yes - one based on JPEGs that none-the-less reinforced that you were wrong about the possibility of the camera being DR limited in that scene.

5) You came back with a reference to back up your claim of 11 stops.

Yes - detailed information from the industry "gold standard" tester.

6) Then you became not sure what my issue was with the scene DR, in spite of 1) and 2) above!

As far as I'm concerned BOTH of my assertions were more than proven, but frankly Ted, I've come to learn that where you're involved, it's possible to get an argument if I assert that 1+1=2. Need I bring up past examples where you've tried to argue: That light doesn't fall off according to the inverse square law? That images don't need sharpening after down-sampling? Argue over techniques for photographing watches? (until I had to do one to prove the point) ... the list goes on.

I know it's hard for you but I think you should just let it drop, to be honest. However, I have absolutely no doubt that the last word will be yours.

I've got a better idea Ted - how about YOU let it drop, before I do something I'm probably not going to regret. You've already caused enough problems in this thread being rude to the OP with your invalid rants concerning large images that YOU chose to download.