Here is a good example of the Exposure Triangle.

[AlanKlein]

Raising ISO number adds noise because the electronic circuits are amplifying the signal that the sensor receives and that causes electrical noise the higher the amplification. The better sensors handle amplification better some there is less noise at higher ISO's. It's like turning up the volume on loudspeakers. At some point the speakers start to distort. Sure the sound is louder, but there are distortions in the sound that are not there at lower volume levels. You can't make a silk purse from a pig's ear.

 

[Ysarex]

So if I'm forced to raise the ISO to maintain shutter speed, I should attempt to overexpose a bit to help reduce read noise?

...to reduce shot noise.

Is there a real world solution to higher ISO = more noise?

I'm trying to apply the information in this thread, but it seems like when I go and shoot, I will still set my camera based on the exposure triangle, and higher ISO will equal more noise...

Less exposure (less light) will equal more noise and there's no getting around that. So real world -- the reason we're raising the ISO is because we don't have enough light. The noise is a function of the total light exposure. The only way to beat it is to get more light. This photo: socks at ISO 12K was taken to demonstrate that the noise source is insufficient exposure. The camera electronics that brighten the image to compensate (ISO) in fact help to suppress noise, they're not the cause of it.

Joe

EDIT: I encounter this with my students all the time. They think the noise is coming from something involved with the camera's ISO setting -- the exposure triangle nonsense that ISO causes noise. So they're reluctant to raise the ISO. So much so that what they'll do is keep the ISO lower than needed and in fact underexpose for that ISO. In other words they really should raise the ISO to 3200 and hopefully the camera meter is indicating +.3. But instead they'll set the ISO to 1600 with the camera meter indicating -1 and think they're doing the best they can to keep the noise down because 1600 is less noisy than 3200. That's just digging the hole deeper. When forced into a low light condition, expose as much as you possibly can and set the ISO to match.

 

[zulu42]

Thanks Joe.

 

[BananaRepublic]

Whilst I found his voice hypnotic, overall the clip reminded me of this,

 

[zulu42]

Okay, the exposure triangle works as us noobs were taught, just remove the misconception that ISO causes noise. Low light causes noise, and underexposing makes it worse.

 

[TreeofLifeStairs]

Raising ISO number adds noise because the electronic circuits are amplifying the signal that the sensor receives and that causes electrical noise the higher the amplification. The better sensors handle amplification better some there is less noise at higher ISO's. It's like turning up the volume on loudspeakers. At some point the speakers start to distort. Sure the sound is louder, but there are distortions in the sound that are not there at lower volume levels. You can't make a silk purse from a pig's ear.

This sounds counter to what Joe is saying though. In your analogy the speaker is producing the noise (the camera in our situation), but from what I’m understanding, Joe is saying it’s the actual signal that has the noise in it. I don’t think the analogy is the same. Even with no music, if I turn my stereo all the way up, I can hear the noise. Joe made a great example by turning up his ISO to the max and still was able to show no apparent noise.


Sent from my iPhone using ThePhotoForum.com mobile app

 

[Designer]

Okay, the exposure triangle works as us noobs were taught, just remove the misconception that ISO causes noise. Low light causes noise, and underexposing makes it worse.

ISO has nothing to do with exposure, but other than that, you're right.

The term applied gain from your ISO setting implies that the gain is applied to the data after capture. Ergo; the ISO setting is not involved with exposure.

As I have written on here before; the ISO setting (gain) is applied to the data in order to generate a JPEG file and present it as a viewable image.

 

[zulu42]

Thanks designer. I feel like I understand that the sensor sees the exposure dictated solely by aperture and shutter speed. ISO setting doesn't alter the data as recorded by the sensor during the exposure.

The point is entirely academic, though, right? Even though the gain is applied after the capture, you still have to set the ISO prior to exposing, and it affects the brightness of the image we see in 1/3 stop increments just like ss and f stop.

 

[AlanKlein]

Raising ISO number adds noise because the electronic circuits are amplifying the signal that the sensor receives and that causes electrical noise the higher the amplification. The better sensors handle amplification better some there is less noise at higher ISO's. It's like turning up the volume on loudspeakers. At some point the speakers start to distort. Sure the sound is louder, but there are distortions in the sound that are not there at lower volume levels. You can't make a silk purse from a pig's ear.

This sounds counter to what Joe is saying though. In your analogy the speaker is producing the noise (the camera in our situation), but from what I’m understanding, Joe is saying it’s the actual signal that has the noise in it. I don’t think the analogy is the same. Even with no music, if I turn my stereo all the way up, I can hear the noise. Joe made a great example by turning up his ISO to the max and still was able to show no apparent noise.


Sent from my iPhone using ThePhotoForum.com mobile app

Well an analogy is not the same. If it was the same, it wouldn't be an analogy. OK, lowering the exposure shutter time or making the aperture smaller allows less photons in. So you compensate by increasing the gain of the amps by changing the ISO to a higher number to "normalize" the exposure. Some noise may be coming from the exposure (less photons) and some from the amps. Regardless, the end result is more noise in the final picture file. That's why you want to use the lowest ISO that's practical.

By the way, maybe a better analogy would be the shadow slider in post. You raise the exposure in that area but you start getting artifacts, noise, etc.

 

[AlanKlein]

Thanks designer. I feel like I understand that the sensor sees the exposure dictated solely by aperture and shutter speed. ISO setting doesn't alter the data as recorded by the sensor during the exposure.

The point is entirely academic, though, right? Even though the gain is applied after the capture, you still have to set the ISO prior to exposing, and it affects the brightness of the image we see in 1/3 stop increments just like ss and f stop.

+1

 

[AlanKlein]

Here's a good explanation on where noise comes from.
Image noise - Wikipedia

 

[Ysarex]

Raising ISO number adds noise because the electronic circuits are amplifying the signal that the sensor receives and that causes electrical noise the higher the amplification. The better sensors handle amplification better some there is less noise at higher ISO's. It's like turning up the volume on loudspeakers. At some point the speakers start to distort. Sure the sound is louder, but there are distortions in the sound that are not there at lower volume levels. You can't make a silk purse from a pig's ear.

This sounds counter to what Joe is saying though. In your analogy the speaker is producing the noise (the camera in our situation), but from what I’m understanding, Joe is saying it’s the actual signal that has the noise in it.

Yes! You got it! Alan is talking about read noise. Shot noise is in the signal and with today's modern cameras it's shot noise that you're seeing. I'm going to put up another post with some examples.

Joe

I don’t think the analogy is the same. Even with no music, if I turn my stereo all the way up, I can hear the noise. Joe made a great example by turning up his ISO to the max and still was able to show no apparent noise.


Sent from my iPhone using ThePhotoForum.com mobile app

 

[Ysarex]

Thanks designer. I feel like I understand that the sensor sees the exposure dictated solely by aperture and shutter speed. ISO setting doesn't alter the data as recorded by the sensor during the exposure.

YES! And the scales fell from his eyes!

The point is entirely academic, though, right?

No! Got a little more work to do then.

Even though the gain is applied after the capture, you still have to set the ISO prior to exposing, and it affects the brightness of the image we see in 1/3 stop increments just like ss and f stop.

Let's try and clean up a couple of things going on here and pull in some stuff from other recent posts. There's a little confusion here about the source of noise. There are multiple sources of noise in a digital photo but most of us only have to worry about shot noise and read noise. The noise Alan has described in a couple posts is read noise. Let's have a look at it: To do that I had to reach back a decade and dig an old camera out of the drawer. I used to shoot a little 1/1.7 sensor Samsung compact. So I just took this photo with it at ISO 1600 which is pretty high for 10 year old tech and a sub-postage stamp size sensor:

Not bad for ISO 1600 and that downsized image is noise filtered. But now let's look at the image at 100% with no noise filtering. We're also going to look at a second image side by side. The second photo was shot without moving the camera at the exact same exposure but with the ISO set back to 80. Both raw files. The second photo was brightened in post to match the first shot at 1600:

There's a difference in the noise -- the ISO 80 shot that was brightened has worse noise. Raising ISO suppresses noise in many cameras and especially older cameras that generate read noise. So raising the ISO to 1600 helped reduce noise. The ISO 80 shot is noisier (color trouble as well) because the 10 year old tech in the electronics generates some read noise which the ISO gain otherwise helps suppress.

In that case sitting there is my new compact camera a Canon G7Xmkii. It is virtually free of read noise. They have engineered it out of this camera. The same goes for many modern cameras now especially from Nikon, Sony and Fuji. They are virtually free of read noise. Let's use the G7 then and repeat the above test.

I metered the scene for ISO 1600 and took that photo and then left the exposure the same and re-took the shot at ISO 125. Here's that side by side:

The noise is the same in both. Unlike above with the Samsung, ISO 1600 provided no advantage and the data in the two raw files will render the same final image when the ISO 125 shot is brightened in post. Read noise was a reality we had to pay attention to a decade ago. The Wiki article Alan cited is dated. Read this instead: what's that noise. Buy a camera today and especially tomorrow and read noise is rapidly becoming a worry from the past.

Zulu! This is where you pay attention again. Is this just an academic exercise? NO! When the ISO function in a digital camera brightens the image it does so before or during ADC (analog to digital conversion) which is before the raw file is created and written. That makes what ISO does permanent in a raw file. And when ISO brightens the data coming from the sensor it reduces DR. Look above at the jar of rice behind the camera (photo from the G7) and the specular highlight on the jar just below the rim. That's the photo brightened from the ISO 125 shot which produced a nearly black JPEG. Now look at that same highlight in this shot:

Specular highlights should clip and in the ISO 1600 shot directly above that highlight is clipped -- it's clipped in the raw file and there's nothing I can do about that the data is lost. That highlight was clipped by the ISO brightening of the sensor data and we in fact call that ISO clipping. The DR in the raw file is reduced and a histogram of the raw file will show the clipping. But in the ISO 125 shot that highlight is not clipped. It's not even close to clipped and I was able to retain it in processing as a diffuse highlight. That's not academic -- that's a new capability we now have with cameras that are functionally free of read noise.

Joe

 

[zulu42]

Joe, thanks very much for the great explanation. I do see how it isn't entirely academic in that- know your sensor and use ISO as needed to reduce noise. Often that means raise the ISO, especially in cameras that aren't real new.

I shoot a ton in low exposure situations. Not just low light because even in daylight I'm pumping ISO up to get the shutter speeds I like for shooting birds. It was actually a struggle to get over avoiding ISO at the cost of underexposing. Still, shooting birds seems like constantly at or near the limit of low light. Low light Cooper's Hawk

If there's any way to apply this knowledge to get my d800 a few more stops worth of nice images... should I try to ETTR more as the ISO goes up?

 

[donny1963]

Okay, the exposure triangle works as us noobs were taught, just remove the misconception that ISO causes noise. Low light causes noise, and underexposing makes it worse.

ISO has nothing to do with exposure, but other than that, you're right.

The term applied gain from your ISO setting implies that the gain is applied to the data after capture. Ergo; the ISO setting is not involved with exposure.

As I have written on here before; the ISO setting (gain) is applied to the data in order to generate a JPEG file and present it as a viewable image.

Correct, ISO has nothing to do with Exposure, people think it does because if they turn up the ISO the picture looks like it's getting better exposure, all it's doing is applying gain..
By the way what the ISO does happens after the picture has been taken,, yeah people think it's like FILM, It's not!!!
the ISO takes place after the picture is taken and then the camera's processing writes the picture to the data card.

basically a DSLR camera is nothing more then a Computer with a lens attached to it..
Some People just don't understand, the Lower the iso you use the better quality your picture will be providing you took the correct exposure and all.
Always a rule use the Lowest ISO number you can use, other wise the higher you turn it up the more noise and the more of your picture information you lose..

the noise your seeing from high ISO , some people think it's the same as Grain on Film, IT'S NOT!! it's garbage and degraded data of your picture, you lose contrast, dynamic range, Color all that, it gets broken up and looks like crap.. Film grain is the old days where you would use a higher ISO / ASA film the chemicals embeded in the film is created that way, and it's not noise and garbage like digital..
Taking pictures with a digital camera is not the same as using film,
Back in the film days i use to shoot slides , FUJIFILM Fujichrome Velvia 50 the ISO i used was 50, i use to get the sharpest pictures using that.
Best i ever used, it was mainly designed for landscape photography, but i did use it for portraits as well..
the color was saturated a bit, but in a good way, and very very sharp, because of the low ISO.

The only drawback to using that was (Light) very unforgiving shooting in low, and even during daylight you still had to be careful doing portraits, you had to use a flash at all times otherwise it would pickup shadows very easy, and ruin your shot, unless you use the fill flash..
just the shade from a subjects jaw or hat the shadows would be very dark because of the LOW ISO.
for portraits i mainly used a 50 M 1.4 Lens, shooting at an aperture higher then 3.5 was difficult with ISO 50