Any advantage to higher ISO when shooting in RAW?

[mcap1972]

The digital range is higher when shooting in RAW.

 

[photoguy99]

In practice the differences are quite small.

If you do have large patches of very similar tones you'll get posterization if you shoot low ISO and brighten in post. This is trivial to fix. Technically, what you've done is oversample in the spatial dimension, so you can simply low-pass the result and get the same result as shooting at bigger ISO.

The problems arise when you have fine detail with very subtle differences in tonality. This is the kind of detail that is hard to see anyways, and tends to be destroyed by resizing and compression anyways.

With appropriate technique in post, you can shoot everything in RAW at base ISO and get a very very close to the same results as with shooting at higher ISO. You get a somewhat different noise profile which you may or may not find easier to deal with or more pleasing.

If JPEG is in the mix, use an appropriate ISO all the time.

 

[Ysarex]

Also assume 100% quantum efficiency of the sensor and zero noise.

Yep, every weekend I work in the garage to come up with a practical method to shoot my camera while immersed in liquid nitrogen. Haven't made a lot of progress. If you figure that one out before me let us know.

As photoguy says, in practice there's nothing there to wet your pants over and I'd go one way or the other simply based on how convenient it was when shooting my camera. For example, right now I'm shooting Fuji X system cameras which are nostalgically designed with knobs and such for us old guys to use and for the young guys to look at and think it makes them retro-cool.

To change the ISO on my camera I have to push a button. Then I have to jog wheel through a list to the right ISO and then I have to push another button to set the change. On the other hand there's a nice big knob right on top of the camera which when turned directly alters the exposure in 1/3 stops. So if I encounter low light and don't have my tripod I can get the faster shutter speed I need by turning that one knob -- fast. Or I can start pushing buttons and jogging. I've taken the photo before someone trying to change the ISO has finished jogging. That time delay can matter.

In practice the differences are quite small.

If you do have large patches of very similar tones you'll get posterization if you shoot low ISO and brighten in post. This is trivial to fix. Technically, what you've done is oversample in the spatial dimension, so you can simply low-pass the result and get the same result as shooting at bigger ISO.

The problems arise when you have fine detail with very subtle differences in tonality. This is the kind of detail that is hard to see anyways, and tends to be destroyed by resizing and compression anyways.

With appropriate technique in post, you can shoot everything in RAW at base ISO and get a very very close to the same results as with shooting at higher ISO. You get a somewhat different noise profile which you may or may not find easier to deal with or more pleasing.

If JPEG is in the mix, use an appropriate ISO all the time.

I would endorse this as exactly right. So my camera is right here on my desk and it's real nice out so I ran out on the front porch. Here's two quick photos:

My camera was as always set to base ISO and it was getting dim so I twisted that knob to -3 and got a shutter speed of 1/27th sec at f/7.1 (I grabbed the tripod on the way out).

Then I put the exp. cmp. dial back to zero and ran up the ISO to 1600 and took the photo at 1/27th sec. and f/7.1. Identical exposure for the sensor each time. One version is amplified (ISO 1600) and one isn't. Processing in PhotoNinja is identical for both photos with the only sole exception that the ISO 200 version got a +3 exp. cmp. If I wanted to role up my sleeves I could get serious about noise filtering either one to best advantage and you're going to need a seriously powerful pixel peeper to tell them apart.

Joe

 

[robbins.photo]

Also assume 100% quantum efficiency of the sensor and zero noise.

Yep, every weekend I work in the garage to come up with a practical method to shoot my camera while immersed in liquid nitrogen. Haven't made a lot of progress. If you figure that one out before me let us know.

As photoguy says, in practice there's nothing there to wet your pants over and I'd go one way or the other simply based on how convenient it was when shooting my camera. For example, right now I'm shooting Fuji X system cameras which are nostalgically designed with knobs and such for us old guys to use and for the young guys to look at and think it makes them retro-cool.

To change the ISO on my camera I have to push a button. Then I have to jog wheel through a list to the right ISO and then I have to push another button to set the change. On the other hand there's a nice big knob right on top of the camera which when turned directly alters the exposure in 1/3 stops. So if I encounter low light and don't have my tripod I can get the faster shutter speed I need by turning that one knob -- fast. Or I can start pushing buttons and jogging. I've taken the photo before someone trying to change the ISO has finished jogging. That time delay can matter.

In practice the differences are quite small.

If you do have large patches of very similar tones you'll get posterization if you shoot low ISO and brighten in post. This is trivial to fix. Technically, what you've done is oversample in the spatial dimension, so you can simply low-pass the result and get the same result as shooting at bigger ISO.

The problems arise when you have fine detail with very subtle differences in tonality. This is the kind of detail that is hard to see anyways, and tends to be destroyed by resizing and compression anyways.

With appropriate technique in post, you can shoot everything in RAW at base ISO and get a very very close to the same results as with shooting at higher ISO. You get a somewhat different noise profile which you may or may not find easier to deal with or more pleasing.

If JPEG is in the mix, use an appropriate ISO all the time.

I would endorse this as exactly right. So my camera is right here on my desk and it's real nice out so I ran out on the front porch. Here's two quick photos:

My camera was as always set to base ISO and it was getting dim so I twisted that knob to -3 and got a shutter speed of 1/27th sec at f/7.1 (I grabbed the tripod on the way out).

Then I put the exp. cmp. dial back to zero and ran up the ISO to 1600 and took the photo at 1/27th sec. and f/7.1. Identical exposure for the sensor each time. One version is amplified (ISO 1600) and one isn't. Processing in PhotoNinja is identical for both photos with the only sole exception that the ISO 200 version got a +3 exp. cmp. If I wanted to role up my sleeves I could get serious about noise filtering either one to best advantage and you're going to need a seriously powerful pixel peeper to tell them apart.

Joe

You just need to duct tape a bannana to the camera before using the liquid nitrogen. Simple really

 

[spacediver]

Yep, every weekend I work in the garage to come up with a practical method to shoot my camera while immersed in liquid nitrogen. Haven't made a lot of progress. If you figure that one out before me let us know.

oh c'mon, it was a thought experiment for the sake of understanding the theory of the physics and electronics. An often used device in many teaching methods. Figured you'd catch on to that

 

[Ysarex]

Yep, every weekend I work in the garage to come up with a practical method to shoot my camera while immersed in liquid nitrogen. Haven't made a lot of progress. If you figure that one out before me let us know.

oh c'mon, it was a thought experiment for the sake of understanding the theory of the physics and electronics. An often used device in many teaching methods. Figured you'd catch on to that

Figured I did catch on to that.

It's like tests of dynamic range. DX0 labs has tested the dynamic range of my camera. Their result is utter nonsense. It's a lab test and has almost no relevance to taking photographs which is what I do with my camera. Photoguy implied this: the test that matters is the print on the wall and by the time you get there the lab test results were long ago lost.

You did originally ask the question in a manner that implied practice: Is their any advantage to using a higher ISO when shooting RAW? Can you see a pragmatic difference in those two photos I posted? Can you even tell which is which?

Joe

 

[spacediver]

Ah I see the source of confusion. I should have made my post clearer. The RAW part was just to de-emphasize the notion that shooting in high ISO may have provided gain for the image (e.g. someone who doesn't shoot in RAW might want a lazy way to brighten an image by using higher ISO values).

The motivation behind these questions has potential implications for a project I'm currently working on, using slow shutter speeds, involving display characterization, where optimizing signal to noise ratio is of paramount importance. The involved luminances in this project are well below those encountered in normal photography (they're below 0.002 cd/m2, which is starting to approach the threshold of human sensitivity to light). As I'm analyzing the RAW files directly in matlab, I'm less concerned with how an 8 or 10 bit rendered image looks than about the statistics of the 14 bit pixel values.


I'm not claiming any practical benefit for photography, I was purely interested in what exactly increasing ISO does in terms of information. I answered my own question with the help of discussion - I originally didn't understand how increasing ISO could provide any advantage that RAW post processing couldn't take care of. I now understand why increasing ISO may have a legitimate role in providing scene detail that would otherwise be invisible in the encoded image. Your last post, along with photoguy's suggests that it's rare to actually encounter a situation where increasing ISO actually helps, and I'm not in a position to disagree, having never owned a camera

I appreciate those photos (and they look good too!)

 

[Ysarex]

Ah I see the source of confusion. I should have made my post clearer. The RAW part was just to de-emphasize the notion that shooting in high ISO may have provided gain for the image (e.g. someone who doesn't shoot in RAW might want a lazy way to brighten an image by using higher ISO values).

The motivation behind these questions has potential implications for a project I'm currently working on, using slow shutter speeds, involving display characterization, where optimizing signal to noise ratio is of paramount importance. The involved luminances in this project are well below those encountered in normal photography (they're below 0.002 cd/m2, which is starting to approach the threshold of human sensitivity to light). As I'm analyzing the RAW files directly in matlab, I'm less concerned with how an 8 or 10 bit rendered image looks than about the statistics of the 14 bit pixel values.


I'm not claiming any practical benefit for photography, I was purely interested in what exactly increasing ISO does in terms of information. I answered my own question with the help of discussion - I originally didn't understand how increasing ISO could provide any advantage that RAW post processing couldn't take care of. I now understand why increasing ISO may have a legitimate role in providing scene detail that would otherwise be invisible in the encoded image. Your last post, along with photoguy's suggests that it's rare to actually encounter a situation where increasing ISO actually helps, and I'm not in a position to disagree, having never owned a camera

I appreciate those photos (and they look good too!)

OK got it then, you are the lab guy. That's great as we need lab guys (and gals). My son is a research physicist so I have to support the lab guys. And I agree it's important to understand how things work. I'm the college photo prof. and it drives me nuts that all my students think raising ISO on their cameras is increasing light sensitivity. They get that from the bleep Internet and from the legacy understanding that came over from the analog world. I think I often ignore ISO as much to be able to show them what's really going on as just a matter of convenience. Photoguy is right that there are subtle differences between amplify versus don't amplify but for the sake of understanding, I sure wish the industry hadn't transferred over ISO from film to digital.

Joe

 

[robbins.photo]

Yep, every weekend I work in the garage to come up with a practical method to shoot my camera while immersed in liquid nitrogen. Haven't made a lot of progress. If you figure that one out before me let us know.

oh c'mon, it was a thought experiment for the sake of understanding the theory of the physics and electronics. An often used device in many teaching methods. Figured you'd catch on to that

Ya, ya.. blah blah.. theories/methods etc - but seriously, do I look like I'd care if there were not a banana involved somehow?

Lol

 

[CameraClicker]

The digital range is higher when shooting in RAW.

The camera always shoots in raw. If you ask, it will give you the raw data, or it will process the raw data -- throwing out most of the data, to give you a JPEG. The first step is always the collection of data from the sensor, from there it may be amplified and/or processed before being written to memory as a file.

 

[spacediver]

Ya, ya.. blah blah.. theories/methods etc - but seriously, do I look like I'd care if there were not a banana involved somehow?

Lol

haha, tell u what, I'll go shoot a dark banana on a moonless night in low and high ISO, and we'll see who goes hungry

 

[robbins.photo]

Ya, ya.. blah blah.. theories/methods etc - but seriously, do I look like I'd care if there were not a banana involved somehow?

Lol

haha, tell u what, I'll go shoot a dark banana on a moonless night in low and high ISO, and we'll see who goes hungry

Lol.. well that's the advantage of being the 800 lb gorilla in the room, you are never the one to go hungry. You do have to put up with some uncomfortable silence now and again though, especially when you wear pink - lol