High ISO Questions

[smoke665]

@astroNikon Sports and Wildlife were the two areas I first thought of when looking at the high ISOs. I'm not really up on the Nikon but in the case of the Pentax they went with a slightly different sensor, a new CPU, and system software. As manufacturers develop new technology I honestly believe they will one day mimic or even surpass the human eye in the ability to capture light and dynamic range.

 

[dasmith232]

"Practical use" is certainly a novel idea. Will it be useable for documentation? Maybe, I guess depending on what you're using the images for....

There are lots of decisions and compromises to be made in photography. ISO Speed is one of the them, and it's important, but so is getting the shot. And for that matter, just getting out there and getting any shots. Especially for PJ, an image with some noise is a whole lot better than motion blur or camera shake with no noise. I think?

I find that if I'm reasonably careful about controlling exposure (beyond the ISO setting), I don't really find myself worry about noise all that much. Then put the power of post-processing to work, and things usually work out pretty well.

I'm doing something wrong. I go out and take pictures and really don't over analyze stuff.

Yeah, you're just crazy...

Oh, and the other thing, if you're not shooting everything at ISO 100, then ... uh, nevermind.

 

[JasonC]

In my film days I considered anything over ISO 100 to be too grainy. In fact, ISO 64 was my film speed of choice in my AE-1 and ISO 100 was regularly wound in my little pocket Olympus XA. However, with digital technology and full-frame cameras, the noise of ISO 1600 seems to be comparable to the grain found in between ISO 100 and 400.

Just this weekend I was at an outdoor car show with overcast skies and was forced to take hand-held pics at ISO 1600. Surprisingly, the pics, which were taken with a Canon 760 D and a cheap kit lens (18-55mm) were not what I would consider excessively noisy. Indeed I was pleasantly surprised!

Best,

Jason

 

[greybeard]

My D750 allows me to set my camera to Aperture priority, set the ISO to AUTO, set the minimum shutter speed to whatever (usually 1/125-1/250) and set the Aperture to whatever (usually) f/5.6-f/8. Because of the camera's ability to handle high ISO (6400 and above) I don't worry about noisy photos and because my shutter speed is set to minimum 1/125-1/250 etc. I don't worry about motion blur. Because these cameras have shutters that operate at 1/4000 and above I don't worry about overexposure either. When it is necessary, I can go full manual and have complete creative control.

My 1st cameras didn't have a light meter so I learned using a handheld light meter. I shot most of my stuff on Kodachrome 25 (that's ISO (ASA) 25) . Today's super high ISO cameras represent freedom to me.

 

[smoke665]

For those who've weighed in that they are using the higher ISO settings, have you noticed any significant differences in noise between focal lenght and/other distance to the subject?

 

[astroNikon]

FYI, the D750 (d6x0, d5x00, d3x00) shutter speed only goes to 1/4000.
whereas the D500 (and D7x00, d8x0, etc) goes to 1/8000.

99.9% of the time I'm in manual where I can set the exact Aperture, and the exact Shutter Speed I want and in sports/wildlife I use AUTO ISO. In other scenarios I manually set ISO.

I've never liked the results of allowing the camera to select either the Aperture or Shutter speed because when I see a scene I know what Aperture (DOF) I want and what Shutter speed is needed (in kids sports as they get bigger the speed go up, and the camera doesn't know that). The computer can just take a guess and if you set your lower limits high enough you'll get what you want most of the time.

 

[JasonC]

For those who've weighed in that they are using the higher ISO settings, have you noticed any significant differences in noise between focal lenght and/other distance to the subject?

No I haven't, but then I've only just begun experimenting with "high" ISOs. Great question, though.

Best,

Jason

 

[Destin]

Alright so I just got my D500 from Santa in a brown UPS suit. I'll put together a real life test in a separate thread in the coming days, but for now, here's a quick preview compared to my D7100. Both images were captured as compressed RAW files, SOOC and converted to .jpg in lightroom. No adjustments applied.

D500@ 51,200:

D500 @ 12,800:

D7100 @ 12,800:

 

[smoke665]

@Destin you can see the difference at least on my tablet. Looking forward to reading about your experiences.

 

[Ysarex]

A little fly in the ointment here that has to be considered: Processing software matters.

I'm not a low light shooter and it's the odd occasion that you find me taking photos into the ISO 1600 range, but I do take extreme contrast high DR photos and then squeeze the max out of them. When you get down to the bottom of your sensor data it really matters how you process that data. Too busy right now to shoot an example but here's a instance where I had to take some low-light photos (some event snaps recently for some friends). Not the best example but should serve. Here's the snap:

Now we're going to zoom in to the darkest area of the photo were we can watch a tone transition into black:

First: Default open in both Adobe and Capture One with default noise processing. The difference is subtle but subtle matters when you're trying to squeeze out that last little drop. Adobe's transition is splotchy by comparison. What you're seeing is the difference between floating point precision versus whole numbers. Adobe has made a compromise choice for speed over precision. Capture One has made the same compromise in the other direction.

NOTE: This is not to be critical of one over the other -- both choices are good choices and we can't have it all. Personally I endorse Adobe's choice in this case and when people ask me what software to use the word Lightroom is out of my mouth before they finish their sentence.

We live in an Adobe dominated world and so when we evaluate our camera's performance Adobe is typically the default arbiter. In those instances where you're trying to use a camera sensor underexposed by as much as 5 to 7 stops Adobe may not be your best friend.

Joe

 

[smoke665]

Good point Joe. I found that processing my raw files in Pentax proprietary utility, will consistently yield better results, but it's such a PITA that it's not worth the effort. The presets I use in LR and PS are close enough.

 

[TCampbell]

The advent of the mega ISO cameras being released has prompted me to read up on the subject, which in turn brings more questions.

1. In the past there has been some discussion as to the effect a higher ISO has on dynamic range. My understanding on the new, mega high ISO cameras while not necessarily practical from a use standpoint at the maximum level, have actually raised the bar on the lower levels, making practical use of 50,000 ISO possible. While there have been improvements in the sensors, most of the increase seems to have come in the CPU and image acceleration software, which combined, limits downstream/upstream noise. My question is has the dynamic range capability improved in addition to the downstream/upstream noise reduction, or has the noise control simply created an appearance of an improvement.

This depends on the sensor. Some sensors don't do any (or at least not much) "upstream" amplification and rely entirely on "downstream" amplification. Downstream amplification results in a loss of DR (1 stop of DR is lost for every 1 stop of ISO boost).

If the sensor does "upstream" amplification, then that happens before the analog to digital conversion (ADC) and you can get some ISO gain without much of a loss in DR.

Noise can be randomly anywhere on the sensor. But since the noise represents a spike in reported values for that photo-site (as compared to it's neighbors) and bright regions of your image already have very high values, the noise doesn't stand out as being so different than it's background. But when this happens in dark regions of your images, the contrast difference is very noticeable.

When you apply global noise reduction, you try to compare the value of a pixel as compared to it's neighbors and "average them out". This process, unfortunately, also results in a softening of the image (which most people don't want.)

You can use software to selectively apply noise reduction based on the brightness of it's neighboring pixels so that it more aggressively goes after noise in shadows and does very little for noise in highlights. Another technique some software uses is to detect edges of contrast and generate a mask that blocks the noise reduction from happening in those areas... but allows noise reduction in areas with little to no contrast.

So while you may not be able to stop the noise from being in the image straight-out-of-the-camera... you can do something about it once you get the image onto the computer.

 

[Derrel]

Destin,
One thing to note is that the D500's shot at ISO 12,800 is considerably "brighter" than the ISO 12,800 shot made with the D7100. I did not look at EXIF information, but one would want to compare the two cameras in two ways: at the same or identical f/stop and shutter speed setting for both cameras, and also at the same level of "exposure" in terms of brightness/histogram, even if that means slowing the shutter down for one camera (in this case the D7100) in order to achievce that same level of "exposure".

It looks to me that at 12,800 the D500 has improved over the D7100's sensor. The one knock against the D7100 was that pattern banding that can been seen in deeply under-exposed areas that are later brightened in software; THAT is the single biggest difference ebtween the D7100 and the D7200--the pattern banding in deeply under-exposed parts of D7100-shot frames has been eliminated in the D7200's images.

The whole question of High ISO is plagued by viewpoints that are 15,12,10,9,5, 3 years behind the current state of the art in sensor tech.

Ask a Nikon D200 shooter about High ISO issues and then ask the same question to a Nikon D500 owner; ask a Canon 5D-Mark II shooter about underexposing by 5 stops and lifting the shadows later in software and then ask a Nikon D750 owner the same question and you'll get a huge, "No way!" and a, "Yeah, sure, if I want to, I can do that."; ask a Nikon D2x user what ISO 1600 is like and then ask a Nikon D500 shooter the same issue; sensor technology AND camera electronics AND software processing tools--all three of these things have changed, markedly, within the past five years. How people view High ISO, and how people view AUTO ISO, and how people view a number of things related to this depends, in many instances, on viewpoints that were set, or developed, with what is now outdated technology.

A lot of shooters hold ideas that are four,three,two camera generations behind what is possible with TODAY'S sensors, camera's and their internal electronics, and today's software tools.

 

[smoke665]

A lot of shooters hold ideas that are four,three,two camera generations behind what is possible with TODAY'S sensors, camera's and their internal electronics, and today's software

Think you nailed It!

 

[Ysarex]

This depends on the sensor. Some sensors don't do any (or at least not much) "upstream" amplification and rely entirely on "downstream" amplification. Downstream amplification results in a loss of DR (1 stop of DR is lost for every 1 stop of ISO boost).

If the sensor does "upstream" amplification, then that happens before the analog to digital conversion (ADC) and you can get some ISO gain without much of a loss in DR.

This has me a little concerned. DR is a function of exposure period. The engineering implementation in the camera that processes the sensor signal can employ a boost to the analog signal prior to ADC, digital scaling in the ADC and/or a hybrid combination of both. Regardless of implementation the method used to process the signal will do a better or worse job of retaining the sensor data but not create data. Exposure creates data. The light sensitivity of the sensor is in no way altered by the methodology engineered into the system to process the sensor output. A full sensor exposure = maximum DR. Any reduction in exposure = reduction in DR.

Joe

Pass the popcorn.