Guide on ISO-speeds, suggestions anyone?

[Drake]

That's right, shots ISO 100-200 from my 10MP Rebel XS are about 8-9mb (RAW). Take the ISO to 1600, and the output is about 10-11, sometimes even 12mb.

But I think you are missing the dynamic range loss. Depending on the camera, jpeg/raw and the testing procedure the results might be different, but just to give you an idea - a page from the test of a Rebel XS, by a very reputable Polish site.

Test Canon EOS 1000D - Zakres tonalny - Test aparatu - Optyczne.pl

I don't expect you to read the text, but just look at the chart

The black dots stand for the total DR.
Green dots - good quality
Red dots - very high quality

As you can see, the 'good quality' DR may shrink from 6EV to 3EV, which is a lot.

Sorry for the link in Polish, couldn't find one fast enough in English.

 

[kasperkjaerphoto]

That's right, shots ISO 100-200 from my 10MP Rebel XS are about 8-9mb (RAW). Take the ISO to 1600, and the output is about 10-11, sometimes even 12mb.

But I think you are missing the dynamic range loss. Depending on the camera, jpeg/raw and the testing procedure the results might be different, but just to give you an idea - a page from the test of a Rebel XS, by a very reputable Polish site.

Test Canon EOS 1000D - Zakres tonalny - Test aparatu - Optyczne.pl

I don't expect you to read the text, but just look at the chart

The black dots stand for the total DR.
Green dots - good quality
Red dots - very high quality

As you can see, the 'good quality' DR may shrink from 6EV to 3EV, which is a lot.

Sorry for the link in Polish, couldn't find one fast enough in English.

I don't think I'm pro enough to quite understand what you mean, but as far as I can interpret you are agreeing with me that high ISO-speeds give low quality images (if an image can be achieved without blur at low ISO-speeds).

 

[kasperkjaerphoto]

@erose86:
I am not trying to discourage it at all costs, but a friend of mine says that she always shots at ISO1600 because she thinks it gives better images (the forementioned reasons) it is this common misconception I am trying to get rid of.
I am aware that I sound against high ISO-speeds, that is because as a general rule I avoid them, unless I have problems with camera shake, then I have no choice.
But I think everyone would agree that if you can snap a photo without blur at ISO100, you wouldn't choose ISO400 or higher?

 

[Drake]

That's right, shots ISO 100-200 from my 10MP Rebel XS are about 8-9mb (RAW). Take the ISO to 1600, and the output is about 10-11, sometimes even 12mb.

But I think you are missing the dynamic range loss. Depending on the camera, jpeg/raw and the testing procedure the results might be different, but just to give you an idea - a page from the test of a Rebel XS, by a very reputable Polish site.

Test Canon EOS 1000D - Zakres tonalny - Test aparatu - Optyczne.pl

I don't expect you to read the text, but just look at the chart

The black dots stand for the total DR.
Green dots - good quality
Red dots - very high quality

As you can see, the 'good quality' DR may shrink from 6EV to 3EV, which is a lot.

Sorry for the link in Polish, couldn't find one fast enough in English.

I don't think I'm pro enough to quite understand what you mean, but as far as I can interpret you are agreeing with me that high ISO-speeds give low quality images (if an image can be achieved without blur at low ISO-speeds).

I am no pro either, but from what I understand it's more or less that when you shoot the camera (Rebel XS in this case) at ISO 100, you get about 6 stops of good quality image, and about 1,5 stop in addition of the things you can recover - highlights (at the price of some weird colors, posterization, color clipping, you name it) and shadows (at the price of some heavy noise). That adds up to a total of about 7.5 EV total dynamic range.

At ISO 1600 you get only about 3 stops of the high quality image. The total dynamic range is similar, about 7,5 stops, but you have to recover the remaining lower quality 4 stops first.

But I think everyone would agree that if you can snap a photo without blur at ISO100, you wouldn't choose ISO400 or higher?

No doubt about that, but sometimes when you don't have time to check the LCD and you want to be 100% positive you got it right because you won't have the opportunity to get a second try, you might want to set the ISO a little bit higher, just to be sure.

 

[kasperkjaerphoto]

But I think everyone would agree that if you can snap a photo without blur at ISO100, you wouldn't choose ISO400 or higher?

Oh, I agree with you... I just saying, make sure you're presenting the information in such a way that says to the newbies "use the lowest ISO possible" not in a way that they would interpret it as "High ISO bad. ALWAYS!"

Good luck on your article. :sillysmi:

I think I will write at the end that it is a trial-and-error thing, and when in doubt and snapping a really important photo that you can't do over, use automatic ISO, untill you have mastered manual ISO.

 

[Garbz]

Well Electrical engeneering 101. The higher iso, the more the power consumption to an extent is true. The way digital camera's sensor gets higher sensetivity is by sending more power to the sensor.

That's not entirely true. A typical CCD / CMOS sensor effectively reacts like a reverse biased PN junction. ... Actually they are a variant of a reverse biased PN junction. The bias part causes a small base current to flow through the junction, this exists regardless of light hitting it and is called the dark current. Now what is true that for changes in amplification scheme (resistive load across the sensor) the dark current changes. However this does not change the sensitivity much at all. The majority of this is done at the amplifier and analogue to digital converter.

Now the other power consumption of any type of photo cell requires a photon with energy greater than the bandgap energy to excite the depletion region and create an electron-hole pair allowing electrons to flow through the device.

Errr now in english: Light hit sensor, bit of current moves forward.

For two pictures of the same brightness this current is equal if the amplification on the other end is the same, but it's not. In fact if you amplify a weaker signal the sensor can actually use LESS power since there's less electrons to flow, and the sensor isn't energised for as long reducing the dark current power demands.

To judge the power consumption you need to consider how does the power consumption of the sensor being on for less time, at a different bias level (affecting the bandgap energy slightly too which is where the other truth from your statement that a sensor can be more sensitive), compare to the increased demands on the amplifier, and analogue to digital converter (and if you shoot JPEG, the additional delay due to noise reduction processing which can be significant if you're squeezing every last drop of power out of a microprocessor).

I think in truth this not only depends on each camera design, but also how each picture is taken.

That's my take on the theory, in practice I haven't seen a difference when shooting night shots at ISO100 vs ISO800. My camera chews a battery in roughly an hour regardless of how I set up the shot.

So, a higher ISO value may give you a better exposure capturing more detail which in turn could mean a larger file size? Is that the line of thinking?

All things being equal the compressibility of an image depends on how well the detail reacts to the algorithm. High-ISO images aren't magically larger, take a low-ISO shot of a noise source and you'll have just as large (infact larger since high-ISO NR probably won't be applied) images.

The size of an image is exclusively determined by the content. High-ISO adds noise. The difference in size between a high ISO and low ISO image of a blue sky will be quite big. The difference in noise between a high ISO and low ISO image of something with a complicated fine and contrasty texture that is in perfect focus will be non-existent.

 

[kasperkjaerphoto]

Thank you for all the responses, I have finished my article and posted it on my blog.

Kasper Kjær Photography: ISO-speeds: What is the difference?