Using ISO LOW

[Don Baldwinson]

The sensor in the E-M1 is sometimes mildly complained about by Olympus and other users for noise, even at the default ISO200 setting, and more so at higher settings. Auto ISO also. Most often appearing in shots with an expansive clear sky.
Somewhere on a forum, I was picked up on this about one of my displayed images, and while mild noise has never bothered me too much, I listened to his suggestion - "whenever you can, use ISO Low setting", which I believe is actually ISO100.
Well, I have been following his advice since, and I have to say the images are much cleaner and smoother, not just for clear blue skies. Now I rarely use ISO Auto, and have Noise Control (or whatever it is called), OFF. It also helps if you have to crop heavily.
Worth a try.
Cheers,
Don

[Daniel Bradley]

http://www.dxomark.com/Cameras/Olymp...--Measurements
Actual measured ISO
Hmmmm...

Native ISO on the E-M1 is, as I understand it, ISO 200, which is in fact closer to ISO 125. Below that there is some monkey business going on, but I can't remember the details. All the top guys here on the German forum do not recommend using it. I will check again and see if I can find out more. Native ISO is where you get the most for your money with any sensor.
The best Nikon sensors have native ISO at 50! For comparison, the D-7200 which has about the same pixel pitch as the E-M1, but MUCH cleaner files; less noise and higher dynamic range.
http://www.dxomark.com/Cameras/Nikon...--Measurements

Really curious what the sensor in the D-500 will be like, but first reports suggest a big leap in ISO sensitivity and DR.


Few minutes later...
This from the top guy in Germany (how is your German? )
ISO Low ist sinnvoll, wenn es anders nicht geht und man extrem saubere Bilder bei reduzierter Dynamik haben will. Also bei allen Motiven, die zwar hell, aber ohne große Helligkeitsunterschiede sind. Nebel zum Beispiel.

Translation: ISO 200 makes sense when there is no other way (to get the picture) and you want extremely clean images at the cost of dynamic range. That is, motives that are bright, but do not have any great difference in brightness. For instance, fog.


But then, the link at the top does not seem to corroborate the above. See Dynamic range. But lab tests can be deceptive, as we know....


Noise Reduction, see page 96. Would not be an issue on motorcycles.
Noise is one of the drawbacks we have with cameras (mirrorless) that are always operating in live view. Heat is the problem. DSLRs don't have it unless they are in live view.

[tomsi42]

If I remember correctly - ISO 100 is just a way of overexposing your ISO 200 shot. When there is little DR in the scene; this will work fine. But I think you will get the same effect using ISO 200 and setting the exposure comepnsation to +1 ...

I think that if you use techniques like "Exposure to the right" and be careful with sharpening during post-processing, you can get very clean photos ,also at ISO 200 (and 400). I never judge noise when viewing the photo at 100%, it is better to look at he actual prints or native size on screen.

[Daniel Bradley]

Here is a quick, unscientific test. All sliders in LR set to 0, except for chroma noise at default 25. Taken with my 400 on my E-M1, silent shutter, iPhone release.








Click for full sized.

Proper exposure is THE most important thing. Here I used +0.7 eV taking the exposure from the shiny dart barrel in the center.

Needless to say, with the lenses I use, I am HAPPY to be able to use ISO 200. 400 is pretty much standard, and 800 a LOT of the time.

[RAH]

Now I rarely use ISO Auto, and have Noise Control (or whatever it is called), OFF. It also helps if you have to crop heavily.
Worth a try.

Don, based on the wording in your original post, I am thinking that you were using Auto ISO a lot of the time (until you discovered this Low ISO thing). If this is the case, I just have to say I have always tried to stay away from Auto ISO (even with an upper limit set), because it seems like a recipe for noise. I always use the lowest ISO I can use and still get a good shutter speed.

I know that Auto ISO can be useful with Shutter Priority mode (which I seldom use) to avoid under exposure because of the lack of a fast enough lens, but otherwise, I'd stay away from it. Just MHO.

[Daniel Bradley]

Next little test, this time ISO 200 + 1 eV against ISO 100 at 0 eV.






I see AT BEST, a minute improvement in detail, but a noticeable loss in DR (punch) in the ISO 100 shot. Still, VERY small differences, and certainly not enough for me to use ISO 100. I also agree with Rich, Auto ISO is not a good idea, not with our sensors.


By the way, I am still experiencing problems with this forum in linking the smaller pictures to the larger ones on Flickr. Takes three or four tries to get it right.

[dh202]

I'm on a well calibrated Eizo monitor, and the differences between 100-800 are hardly worth sweating over.
Nothing that cannot be compensated for in post.
ISO800 and up is a whole other ball game, and I rarely ever go that high anyway

[Daniel Bradley]

Agree.
I have to grit my teeth a little when I go up to 800, but it is doable and often necessary. Noise can be managed, but there is loss of detail anyway as ISO gets higher and that means less cropping. Without cropping it is fine, but I have to crop quite a bit. Get the exposure right and you are half way there, but that gets harder as ISO rises due to the reduction in DR.

[dh202]

The noise and DR problem has been been Olys achillies heel with the 4/3 sensors from the very beginning. It has become somewhat better with the incremental advances in the sensors.
They are now using trick software in attempts to help overcome flaws within the output files from the cameras, mostly for lens correction, but who knows what else

[Daniel Bradley]

Tricks? EVERYBODY is doing that! They are all trying to somehow overcome the laws of physics.

[BobT]

I never judge noise when viewing the photo at 100%, it is better to look at he actual prints or native size on screen.

Good point!

[Thonord]

I never judge noise when viewing the photo at 100%, it is better to look at he actual prints or native size on screen.

Excuse me for going OT, but I thought that "100%" was native size on screen. How do I determine the "native size" of my display?

[RAH]

To me, 100% means to zoom in with the software you are using to view the image, so that you are seeing it at full size (100%), with no zoom in or out percent level.

I have never heard the terminology before, but if "native size on screen" doesn't mean 100%, perhaps it means viewing the image at a zoom percent such that it is filling the screen? Considering the context, it sounds like the idea is to view and judge the image in the context of how it will be used - i.e. view it at the print size you want, or for web display view it at the size it will appear on the screen (after downsizing, sharpening, etc). I guess.

[Godfrey]

Excuse me for going OT, but I thought that "100%" was native size on screen. How do I determine the "native size" of my display?

To me, 100% means to zoom in with the software you are using to view the image, so that you are seeing it at full size (100%), with no zoom in or out percent level.

I have never heard the terminology before, but if "native size on screen" doesn't mean 100%, perhaps it means viewing the image at a zoom percent such that it is filling the screen? Considering the context, it sounds like the idea is to view and judge the image in the context of how it will be used - i.e. view it at the print size you want, or for web display view it at the size it will appear on the screen (after downsizing, sharpening, etc). I guess.

Viewing an image at 100% or "native size on screen" means that the pixels are 1:1 ... one pixel in the image corresponds to exactly one pixel on the display screen. How big (and how much of) the image you can see at that resolution depends upon how you have your display configured. My image processing system uses an Apple Thunderbolt 27" display at full resolution, which nets 2560x1440 == 3,686,400 pixels, or about 3.6864 Mpixel.

My three current digital cameras are:


E-M1: 4608x3456 == 15,925,248 pixels or ~15.9 Mpixel
Leica SL: 6000x4000 == 24,000,000 pixels or ~24.0 Mpixel
Leica M-P: 5952x3976 == 23,665,152 pixels or ~23.7 Mpixel


So at 100% or "native" pixel resolution, I can see 23%, 15%, and 15.6% of an image from each of those cameras respectively when I set LR to image them at 1:1 in full screen display mode. (I had a Sony A7 for a year and some as well. I don't have the pixel dimension info at my fingertips just at the moment, but it's right in there with the two Leicas.)


Common print sizes I use that fit the full format images in native proportion are 9x12 inch (3:4, FourThirds) and 10x15 inch (2:3, Leicas). These net 15 and 18 inch diagonal image area respectively, and fit on my display easily. With all three cameras, the pixel resolution is large enough that I just set the print resolution to 360ppi (the highest optimal printing resolution for my printer) and let Lightroom and the print driver interpolate the resolution.

To preview them on the display, the effective scaling on my screen nets a very close approximation of what I'll see on the print if I size a print template to display those dimensions in LR onto my screen. I do the sizing by adjusting the borders having chosen a paper size of 11x17 inches, and measure the size on screen with a ruler.

Even at ISO 6400, I get very clean images from the E-M1 IF I've gotten the exposure right, with no noise reduction or sharpening involved. Getting the exposure "right" means also accommodating the reduction in dynamic range and detail resolution that such a high sensitivity implies. The image quality at this print size is on par with what I was getting out of the Sony A7 at this sensitivity, and is surpassed only marginally by what I see out of the Leica SL and M-P, again at this sensitivity. I've validated that by making prints as well as looking at the images on screen: the fidelity from screen to print is very very close to perfect.

Higher sensitivities induce much more noise, as well as cut resolution and dynamic range; all four cameras diverge at that point and go separate ways in terms of their output image quality, but of course it's visible on screen sized this way as well.

Of course, for image maximum quality, it's much better to drop the sensitivity down to the lower end of the scale. I see little effective difference on any of them between ISO 50 and ISO 400, so at that end of the scale I make ISO decisions based upon desired focus zone depth.

[Thonord]

Thank you both for your answers. Most informative!
Seems I may have to take the "plunge" and upgrade to m4/3. No way will my E30 give me a noiseless image at ISO 6400. I write "may" because "She who must be obeyed" may be of a different opinion.

Again, apologies to Don for hijacking his thread.

[Don Baldwinson]

Thank you both for your answers. Most informative!
Seems I may have to take the "plunge" and upgrade to m4/3. No way will my E30 give me a noiseless image at ISO 6400. I write "may" because "She who must be obeyed" may be of a different opinion.

Again, apologies to Don for hijacking his thread.

No apology needed. I still use ISO low because it gives me clear skies. Then again, it may only be in my head
Don

[BobT]

Noise is one of the drawbacks we have with cameras (mirrorless) that are always operating in live view. Heat is the problem. DSLRs don't have it unless they are in live view.

That's interesting! It has a couple of implications. Firstly, leaving the camera off to just before taking the shot may be beneficial to noise. Secondly, cold weather shooting may also result in less noise. If anyone has the time and inclination, it may be worth checking out to see if there is any significance in this.

By way of a bit of silly brain storming, I've seen gaming PCs with water cooled CPUs. Could we one day see actively cooled sensors? No need to thank me Oly, just send the cheque

[Godfrey]

That's interesting! It has a couple of implications. Firstly, leaving the camera off to just before taking the shot may be beneficial to noise. Secondly, cold weather shooting may also result in less noise. If anyone has the time and inclination, it may be worth checking out to see if there is any significance in this.

By way of a bit of silly brain storming, I've seen gaming PCs with water cooled CPUs. Could we one day see actively cooled sensors? No need to thank me Oly, just send the cheque

Astronomical telescopes have had liquid cooled sensors for a couple of decades now, in order to eke that Nth little bit of light gathering power out of them for long, long exposures.

G

[BobT]

I see little effective difference on any of them between ISO 50 and ISO 400, so at that end of the scale I make ISO decisions based upon desired focus zone depth.

Are you suggesting we may as well generally shoot at ISO 400 (thus reducing camera shake) and only reduce the ISO if we wanted to shoot say a brightly lit portrait with a wide open aperture?

[Daniel Bradley]

As technology progresses I would expect sensors to become less temperature sensitive, so to speak. Certainly happening in computers. My daughters new MacBook Pro doesn't even have a fan in it.
I know that noise is an issue with mirrorless cameras due to heat, but what I don't know is how soon after the camera is turned on it becomes an issue. I would expect within a mater of seconds.

[Godfrey]

Are you suggesting we may as well generally shoot at ISO 400 (thus reducing camera shake) and only reduce the ISO if we wanted to shoot say a brightly lit portrait with a wide open aperture?

That's what I do. My standard ISO setting on all my cameras is ISO 400. I drop it lower when I want more flexibility to use the larger apertures, etc.

[Daniel Bradley]

Same here.

[RAH]

Me too!

[BobT]

Me too... NOW!

[Zensu]

That's interesting! It has a couple of implications. Firstly, leaving the camera off to just before taking the shot may be beneficial to noise. Secondly, cold weather shooting may also result in less noise. If anyone has the time and inclination, it may be worth checking out to see if there is any significance in this.

By way of a bit of silly brain storming, I've seen gaming PCs with water cooled CPUs. Could we one day see actively cooled sensors? No need to thank me Oly, just send the cheque

Maybe something like this?

Here's a link:
http://www.mirrorlessrumors.com/if-y...rther-do-this/

Bobby