ISO 100 vs. 1600 vs. 3200

NikonUser · Post by **NikonUser** » Sun May 24, 2009 1:55 pm

In another thread:
HERE
sagamartha wondered about image quality at ISO 100 vs. ISO 1600 for the Nikon D90.

This American Lappet Moth (length about 3cm) visited last night and stayed all day.

Full frame, D90 + 100mm AF MicroNikkor @ f11 -fine large JPG, unprocessed except reduced to 800px and <200K
Top: IS0 100 1/100sec pop-up flash
Middle: ISO 3200 1/4000 sec no flash
Bottom: ISO 1600 1/2000 sec no flash

rjlittlefield · Post by **rjlittlefield** » Sun May 24, 2009 3:35 pm

Phyllodesma americana? A very pretty moth.

At this scale, the most obvious difference is much more noise in the OOF background regions.

Any chance of seeing some actual-pixel crops of both in-focus and out-of-focus regions?

--Rik

PS. Because this is one of the technical discussion forums, there's no limit on images-per-day as long as you say something about each of them. A bare-bones label is sufficient.

NikonUser · Post by **NikonUser** » Sun May 24, 2009 7:20 pm

800px selections from the non-processed JPG's.
Focus was manual, so the 3 frames may not be all in exactly the same focus plane.
At this level the high ISO's obviously show more noise than the 1ow ISO.

rjlittlefield · Post by **rjlittlefield** » Sun May 24, 2009 8:23 pm

Thanks, NU.

I think you're right about the focus being slightly different in each frame. There's also a bit of difference in appearance that I think is due to flash in the first case but not the second two.

Looking for comparable areas, it seems to me that the wing in upper left is about equally focused in each case.

If that's true, then what I'm seeing is about what I'd expect based on my earlier study using my now ancient Canon 300D: no significant loss of resolution for high contrast detail, slight loss of resolution for low contrast detail, and significant roughening of what should be smooth textures such as OOF background.

What do other people see?

--Rik

sagarmatha · Post by **sagarmatha** » Sun May 24, 2009 11:05 pm

So, FINALLY my question could be resolved. Thanks Nikonuser. I agree with Rik but it seems that it's possible to do good macro shooting in ambient light with a good camera.
The only question now is whether it's a good idea to wait for the D90s successor

. The analysis of the coming D400 regarding noise will reveal that.

DaveW · Post by **DaveW** » Mon May 25, 2009 12:34 am

Unfortunately cameras are now simply computers with lenses, and like computers only have an average 18 month dealer shelf life before they are replaced with something hopefully better. You can wait therefore to buy one forever, because regarding being up to date they have a very limited "best camera available" lifespan.

Their longest lifespan at the top is if you buy at their highest price when they were just released. If you wait for their price to fall their reign at the top will be more limited, and if the makers are offering cash back that is a sign they are clearing dealers shelves for a new model, so the day after you buy it may be announced and you wish you had waited, although the new one will usually be dearer.

However none of this effects the camera itself. It is just as good the day after it is replaced by a new model as it was the day before, plus things like extra pixels on the new one do not always produce better pictures or better high ISO performance, sometimes the opposite.

DaveW

sagarmatha · Post by **sagarmatha** » Mon May 25, 2009 12:41 am

Yes but a new developed sensor and the same amount of MP will maybe reduce the noise even further. One drawback of the D90 that I'm not so fond of is the "simple" video function. Surely this function must be much better developed in the next model. It seems that we are in a transition period regarding this feature. This is why I'm waiting, otherwise I would have bought a D90/D5000 or something else yesterday.

DaveW · Post by **DaveW** » Mon May 25, 2009 1:02 am

"Yes but a new developed sensor and the same amount of MP will maybe reduce the noise even further."

In your dreams! Most noise reduction circuits simply hide noise by softening or degrading the image, not really what macro photographers want I would have thought?

In the three images above, even disregarding any focus variations, the 100 ISO one is obviously the best if you compare the noise (grain) on the background and that will not be influenced by focus.

Even if they improved high ISO noise performance in a new camera the difference between the cameras performance at high and low ISO's will still be there, and the higer ISO's will be degraded when noise reduction kicks in. Some people don't mind noisy images, but I personally prefer the sharpest least noisy (grainy) image I can get.

DaveW

rjlittlefield · Post by **rjlittlefield** » Mon May 25, 2009 10:05 am

sagarmatha wrote:Yes but a new developed sensor and the same amount of MP will maybe reduce the noise even further.

I'm not optimistic about pixel noise going down very much farther.

The problem, as I understand it from various readings, is that a) noise in modern sensors is already dominated by sampling uncertainty in the photon counts, and b) most of the available photons are already being counted.

A 1998 SPIE article writes that "Quantum efficiency (QE) is the fraction of photon flux that contributes to the photocurrent in a photodetector or a pixel. Increasing QE improves sensor signal to noise ratio and dynamic range." That article notes that at that time, quantum efficiencies were running around 0.37. A more recent article in 2006 SPIE describes cell phone sensors as "The peak quantum efficiency is 0.65 and the pixel fill factor is 50%."

Assuming that I'm interpreting the latter numbers correctly, what this means is that 1 out of every 3 available photons is already contributing to the pixel data. Of course that leaves 2 out of every 3 not contributing, which seems to offer 3X opportunity for improvement. Unfortunately, sampling uncertainty only drops as the square root of the counts, and the square root of 3 is only about 1.7.

In other words, even having a perfect sensor would reduce noise by less than a factor of 2. Anything beyond that will come from clever subterfuge, usually called "noise reduction" or "filtering" in advertisements and publications.

It's possible, of course, that I've misinterpreted the literature. In that case I would greatly appreciate being corrected and shown where the error lies.

--Rik

sagarmatha · Post by **sagarmatha** » Mon May 25, 2009 11:03 am

Rik, thanks for clarifying the physical limits regarding this matter.

lothman · Post by **lothman** » Mon May 25, 2009 2:49 pm

@rik

may I ask you as a god of stacking software

could grainy structures due ti high ISO affect stacking algorythms, for example that they would interpret it as an sharp area of the pic?

rjlittlefield · Post by **rjlittlefield** » Mon May 25, 2009 3:28 pm

could grainy structures due to high ISO affect stacking algorithms, for example that they would interpret it as an sharp area of the pic?

Absolutely. Random pixel noise is especially unfriendly to depth map algorithms because it causes errors in determining the plane of best focus. It doesn't have so much effect on the accuracy of pyramid algorithms, since levels above the base are progressively less affected. But because the pyramid algorithms tend to find and preserve "detail" wherever they find it, they tend to enhance the visual effect of noise. Having images that look noisy to start with means that you get images that look very noisy in the end.

As a practical matter, it seems like this should be an issue only for stacks shot in the field. On the bench, you can add more light or extend the exposure time to allow using low ISO. Or have I overlooked something?

--Rik

DaveW · Post by **DaveW** » Tue May 26, 2009 12:46 am

Which brings us back to the old recomendation "always use the lowest ISO possible for the situation". High ISO's are only for situations when a lower one cannot be used!

Many amateur photographers tend to stick their cameras on high ISO's since they think they can then use the fastest shutter speeds all the time and avoid camera shake. They never find out what is the lowest hand holding speed for their various lenses where shake is not obvious so therefore lower ISO's can be used for better image quality.

I always try and use ISO 100 as far as I can, and usually can take insects in good available light at around 160th second hand holding a 70-180 Micro Nikkor, even in the UK where the light intensity is not as high as in the US. Admitted these are not in flight. This one was was taken hand held in that manner and it is a crop from a larger frame:-

http://www.photomacrography.net/forum/v ... php?t=3490

DaveW

sagarmatha · Post by **sagarmatha** » Tue May 26, 2009 12:51 am

In my case it's a question of using or not using a flash to get the image. I think that with VR and ISO 400 one can get rather far without using flash.