ISO 2500 vs 160

[orjanf]

Advertisement (gone after registration)

Since it seems to me that any setting other than the 'native' (presumably optimal) ISO of a sensor involves just processing it occurred to me that there should be little difference between setting the camera to 2500 ASA or simply underexposing 4 steps and using the raw converter to pull out the detail.

A simple test pointing the camera at a corner with the ISO set at 2500 ASA and then taking another picture setting the same shutter speed manually and turning the ISO back to 160, opening in the raw converter and setting the exposure to x4 should give similar results except that I would expect the built in process to apply some 'smart' noise suppression and thus produce a slightly better result at 2500.

The initial test surprised me however and the result of using ISO 160 and 4x appears substantially better in terms of both noise and detail (handheld at 1/15 so the detail is arguable) than using ISO 2500. In fact the image rivals my Canon 1D Mk2 at the same ISO.

I guess I've missed something? And no, I didn't get he images mixed up. I have adjused the white balance (same value for both) and set the black level to 0, other settings are standard default for lightroom. Havent tried in Capture One yet.

100% crops of roughly the same area,

I would much appreciate if someone has the time to verify (or refute) my result.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[davidada]

Wow - you may have hit on something- I will have to try this myself!

[c6gowin]

Amazing results. I don't understand why there is such a difference, but it is dramatic. The pushed 160 shot looks much better in terms of noise and color. The sharpness difference could have been affected by handholding, but it is not unreasonable to expect improvement in sharpness based on the other improvements.

 

My experience has shown the M8 handles the shadows very well, but I never thought of going to this extreme.

 

Mark Gowin

[c6gowin]

After looking at the photos again, I wonder if the 2500 example would look better if the exposure is backed down a bit - it may be slightly overexposed. Am I wrong?

 

Mark Gowin

[barjohn]

Well, I gave it a try and you are right, post processing really is the way to go. Here is the same shot taken at ISO 2500 at 1/45 sec followed by using ISO 160 at 1/45 sec. First processed in LR with black point set at +2 for ISO 2500 and ISO 160, then with black point set at +2 for ISO 160 and finally in C1 4 Beta at exp +2.5 and brightness +37

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[farnz]

Orjan,

 

There have been several posts about this topic in the past and it appears to stem from the M8's compression algorithm from 14-bit to 8-bit that favours the shadow detail.

 

Personally I prefer Shadows and Highlights in PS to reveal the shadow detail because, to me, it produces slightly less noise and more detail than compensating with Exposure but YMMV.

 

I haven't often needed to use the full 4 stops because of Leica lens speed and as a rule I find I get the best results if I don't exceed 2 and a half stops underexposure.

 

Pete.

[Jamie Roberts]

Advertisement (gone after registration)

Been saying this for a very, very long time: shadows at low ISOs will push; at high ISOs you need to nail the exposure and expect a drop in DR.

 

(BTW--both photos are underexposed--the white paper isn't anywhere near white. This is why the high ISO looks crummy).

 

And the "four stops" in LR or C1 isn't actually four EV So there *is* a reason to shoot at high ISOs--but (does this sound familiar yet) you need to expose "to the right" at ISO 1250 or 2500.

 

Otherwise, you'll just get more noise and less DR.

 

@Orjan, properly exposed and processed, your M8 will *kill* your 1d2 (or 1ds2) at low ISOs. It has at least a stop--maybe two--more stuff in the shadows (believe me, I have both, and a 5d too!).

[barjohn]

I have now followed up with first using LR and getting best I could obtain from just adjusting exposure, brightness and black point, first at ISO 2500 then 160. The two after that are using C! first at ISO 2500 then at ISO 160 WB was set to the same value.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[barjohn]

I think the difference is due to the way the signal is amplified. In the camera, when you select a higher ISO you are increasing the gain (and I am betting it is the analog gain) and the amplifier is noisy, hence more noise in the image. C1 and LR amplify the signal in the digital domain which is far less noisy and hence the cleaner images. C1 appears to have better amplification algorithms.

[lars_bergquist]

----------------

(BTW--both photos are underexposed--the white paper isn't anywhere near white. This is why the high ISO looks crummy).

--------------------------------------------

 

Jamie, a white paper is a diffuse highlight. The reflectance of white coated paper is usually somewhere around 85–88%. There must of necessity exist a margin to full 'highlight burnout'. If there would be full burnout, how would you handle a specular highlight which might well be occur in an area of diffuse white? Example: a field of snow with point reflexes.

 

An image with burnt-out detail-less diffuse highlights is overexposed – that is the very definition of overexposure!

 

The old man from the Age of Decent Exposure

[sm23221]

Let's talk real world and practical. You will get a better quality handheld photo at a higher ISO exposed correctly with a faster shutter speed than you will if you used a low ISO at a slower shutter speed and pushed it. If you're using a tripod the point is moot, but it's an interesting assessment anyway.

[farnz]

Let's talk real world and practical. You will get a better quality handheld photo at a higher ISO exposed correctly with a faster shutter speed than you will if you used a low ISO at a slower shutter speed and pushed it. If you're using a tripod the point is moot, but it's an interesting assessment anyway.

Michael,

 

I guess it depands on what you mean by "better quality". Up to, say, ISO 640 I expect what you say is true but beyond ISO 640 noise becomes a significant factor and pushing ISO 160 gives a less noisy image.

 

The deliberate underexposure will naturally compensate for shutter speed, which should be similar in both instances.

 

Pete.

[barjohn]

Let's talk real world and practical. You will get a better quality handheld photo at a higher ISO exposed correctly with a faster shutter speed than you will if you used a low ISO at a slower shutter speed and pushed it. If you're using a tripod the point is moot, but it's an interesting assessment anyway.

 

You missed the point. They were both shot at the same shutter speed! You measure the exposure at ISO 2500 and shutter speed, then shoot at ISO 160 with the shutter speed manually set to the same shutter speed.

[Maggie_O]

Interestingly enough, Mitch and I have been having a very similar conversation regarding the D-Lux 3.

 

Pushing lower ISOs really does seem to work.

 

Of course, it takes all the fun out of chimping!

[Jamie Roberts]

Yes Lars,

 

Of course if we're talking polar bears in the snow, then you need to expose--with some allowance--for highlights and speculars.

 

But this is still underexposed--assuming of course that the white peice of paper is not the subject of this completely pixel-peeping scene (would anyone actually care about such a shot? Can you tell me what the subject is?).

 

So, first of all, the paper is red. Way too red, BTW, unless it's red paper. Check it and see. Like using the wrong film, for you guys from the neolithic age of glass plates

 

White balancing (actually grey balancing) this image shows that the paper is actually pretty underexposed; the blue channel is well into the realm of mid greys; the red is higher, but not anywhere near a specular (or even a highlight).

 

So I stand by what I said, Since I shoot people in black tuxes and white wedding dresses all the time, and need to hold both the shadows and the highlights, I know my way around exposure with the M8

[Guest malland]

Interestingly enough, Mitch and I have been having a very similar conversation regarding the D-Lux 3.

 

Pushing lower ISOs really does seem to work.

 

Of course, it takes all the fun out of chimping!

And with the D-Lux 3 this method of "pushing" is more important because ISO 1600 is unusable because even in RAW there is in-camera smothing that "smears" the image, as ISO 800 is a vert much hit-or-miss propostion aa ro whether it will be usable for anyone shot — even in B&W for someone who likes grain.

 

—Mitch/Bangkok

[barjohn]

Jamie,

 

My WB is pretty close to on; however, I did have to adjust the WB for the ISO 2500 shot as it was completely different than the ISO 160 shot (One of the Leica M8 goodies) both shot with the exact same light.

[barjohn]

Finally, to show another product, I have converted in BibblePro, latest version. First is the picture at ISO 160 as it comes out of the camera and then the ISO 2500 adjusted in BibblePro and then the ISO 160 adjusted in BibblePro. I find this exercise rather telling. I also tried the older version of C1 3.7.7 and I could not get as good results. These are all at pixel peeping 100% as are the former. At normal size they all looked pretty good.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[Advertisement]

Hallo,

Look here: Leica M

[Jamie Roberts]

Jamie,

 

My WB is pretty close to on; however, I did have to adjust the WB for the ISO 2500 shot as it was completely different than the ISO 160 shot (One of the Leica M8 goodies) both shot with the exact same light.

 

Yes, your WB looks ok to me too. I was talking about the OP's shots.

[orjanf]

Hi,

Thank you all for comments.

Sorry about delay, went to bed after posting - timezone difference.

I apologise for the lack of quality of the images. I am glad to see that better ones are now available, thank you John

 

After studying your much better and carefully made images, it seems to me that the basic observation still holds.

An analogue amplifier should maintain the same signal to noise ratio regardless of the gain setting. Since it seems unlikely the signal path is different at the ISO 160 setting or that the A/D converter would be much better at lower input signal levels, it seems to me likely and as pointed out by Pete that this is a consequence of the compression algorithm.

 

Jamie

All your comments about exposure and white balance in regards to the samples I provided are correct.

The original exposure at 2500 was at auto covering a much wider view than the 100% crop. The background (shown) is underexposed. I then just pulled the colour temp slider left until they went from horrible to merely very bad

quote:

"And the "four stops" in LR or C1 isn't actually four EV"

 

I wonder if you would mind expanding on this. I can confirm the general abscence of free lunches in asmuch as a highlight section of the foreground (not shown in crop) which in the 2500 exposure is just below being blown (96,97,96), is in the 160x4 blown. Thus I guess I need to use higher ISO where maximum detail in highlights is required