Jump to content

Banner.jpg.b83b14cd4142fe10848741bb2a14c66b.jpg

What ISO?

Andrew*

By Andrew*
in Imaging - Discussion

Recommended Posts

dark knight

dark knight

Posted
Posted

I think I totally misunderstood the ISO function in modern DSLR's. The mistake I made was thinking it would perform the same function as the 35mm film camera's ISO settings. Maybe it should be referred to as a Gain setting and not ISO at all. Is my interpretation now nearer to the mark....

Link to comment
Share on other sites
  • Replies 66
  • Created
  • Last Reply
Sam

Sam

Posted
Posted
Why? Hot pixels, amp glow etc. can be removed by subtracting dark frames ... If you have a sky glow problem, that is less tractable but anti-pollution filtering may help. Those with serious skyglow problems usually resort to shooting with narrowband filters on a monochrome CCD camera.

Brian, it's because a canon's circuits apparently heat up a bit after a while and eventually you get to the point where it's introducing so much noise you gain no benefit from the longer exposures. Some folks reckon that's about 5-6 minutes. If you could cool the camera directly then this problem would go away. That's how I understand it.

Link to comment
Share on other sites
brianb

brianb

Posted
Posted
it's because a canon's circuits apparently heat up a bit after a while and eventually you get to the point where it's introducing so much noise you gain no benefit from the longer exposures. Some folks reckon that's about 5-6 minutes.

I've been over 20 mins with an unmodified 40D ... without deliberately cooling the thing either. Without dark frame subtraction it's horrid but with dark frame subtraction it's OK. Incidentally dark frame subtraction doesn't work properly until the camera temperature has stabilized.

The best (or worst) way to treat a DSLR is to turn on live view, the continuous readout really does get the chip hot & noisy. I work with the review screen switched off completely.

Link to comment
Share on other sites
narrowbandpaul

narrowbandpaul

Posted
Posted

hi.....i have subsequently tested my 300D to try to answer the various questions about which ISO to use.

also included is a discussion in to the implication for imaging

see the thread below

http://stargazerslounge.com/imaging-discussion/87450-experiment-into-camera-isos-imaging-folows-what-iso-thread.html

i will also post it here.....

Camera Properties and ISO.doc

Link to comment
Share on other sites
Andrew*

Andrew*

Posted
  • Author
Posted

Thanks for sharing that bit of work. Most enlightening. From that, it seems ISO 200 is best for the 300D - most DR and Full well depth, and not too much read noise.

I wonder how the 450D would fare. All that maths is certainly too daunting to try this myself!!

Andrew

Link to comment
Share on other sites
narrowbandpaul

narrowbandpaul

Posted
Posted

thank you guys.

dark knight, you now see that the ISO is just an amplification, and that at high ISO, your image may be of worse quality......

andrew, the maths isnt hard. Software programmes spit out average and standard deviation value. All you have to do is find the average, standard deviation, then square it, and plug it in to the equation.

newer models should show lower read noise, so it would be advisable to check for your own camera......

glad you enjoyed it....

Link to comment
Share on other sites
brianb

brianb

Posted
Posted

Nice piece of work!

newer models should show lower read noise

Also it varies considerably with temperature - like dark current, approximately doubling with every 7C rise - that's why the better dedicated astro cameras are cooled ...

Link to comment
Share on other sites
narrowbandpaul

narrowbandpaul

Posted
Posted

brian, I doubt strongly whether this is true.

can you point me to an article that states this.

the reason that astro CCD's are cooled is to supress dark current, and charge leakout from traps

i have read the charge measurement chapter from James Janesicks Scientific Charge Coupled Devices, and the read noise chapter in his book on Photon Transfer. No where is it ever mentioned that read noise shows a temperature dependance. It may vary slightly with temperature, but read noise is not a kind of dark current that shows such a temp dependance.

i am not having a go at you brian, I share your thoughts regarding ISO, but somewhere you have read that read noise doubles every 7C. So this person has been spitting out bad science, and this sort of thing I would like to see stopped.

So I would be interested to see any article that you have read that states this.

best wishes

paul

Link to comment
Share on other sites
arad85

arad85

Posted
Posted
brian, I doubt strongly whether this is true.

can you point me to an article that states this.

Hi Paul,

Actually, it appears to be 6 degrees.... A quick google pulls out:

Supercooling- Should we just chill out?

Signal to Noise: Understanding it, Measuring it, and Improving it (Part 1) - Article

Special Production ST-4000XCMI

..........

Link to comment
Share on other sites
narrowbandpaul

narrowbandpaul

Posted
Posted

hi arad.

the dark current varies by doubling every 6C.

actually the temperature increase required to double the dark current varies.

i dont think there is an analytic solution to the equation you have to solve. It has to be done numerically. The results are again found in JJ's tome on CCD's.

with regard the second link...I skim read the read noise section, and again nowhere states a temperature dependance.

i think we are getting confused here....... dark current, the spontaneous emission of electrons due to the lattice is strongly temperature dependant.

read noise, the noise introduced when measuring and digitising the charge, is not temperature dependant, at least to first order....

Link to comment
Share on other sites
jgs001

jgs001

Posted
Posted

Paul, thanks, it's very interesting reading, and I'd be interested to know how the 450d pans out compared to your 300d. I can probably cobble the maths itself by using various net tools to work out the stats for me, but... I have a few questions

"Find the average of each of these boxes and note them down"

What values are you using to determine the average ? How do you ensure that the 100x100 pixel box you select is always in the same place on each image (my image edit skills are not very good ? I assume this is to ensure that you are always measuring form the same area of the sensor so that variations across the sensor are removed from the equation. By 50% exposure are you indicating that the peak of the histogram is bang in the middle ? If using Av mode for instance, the camera will aim for the peak being at about 75% (from memory) would this work just as well ?

Link to comment
Share on other sites
brianb

brianb

Posted
Posted
i think we are getting confused here....... dark current, the spontaneous emission of electrons due to the lattice is strongly temperature dependant.

read noise, the noise introduced when measuring and digitising the charge, is not temperature dependant, at least to first order....

Nevertheless my experience with short exposures (sub-1/100 sec) with DSLRs, in a "normal" photographic setting, is that noise increases significantly with temperature, and surely dark current has little or no effect on such short exposures. It's either bias noise or readout noise, and my understanding is that DSLRs with CMOS sensors automatically take a bias frame & subtract it every time the shutter button is pressed, this is necessary because of the characteristics of the sensors used.

With CCD cameras and their (relatively) slow readout, it's usual for the readout noise to vary systematically across the frame, and that variation is definitely quite sensitive to temperature. The variation arises because of "dark current" electrons leaking into & out of the "buckets" whilst waiting to be read, so the physics is the same as that governing dark current temperature dependency.

Whether the doubling rate is 6C or 7C is nit picking, in any case I usually qualify the value with "approximately".

Link to comment
Share on other sites
jgs001

jgs001

Posted
Posted

On the basis of this statement Brian

my understanding is that DSLRs with CMOS sensors automatically take a bias frame & subtract it every time the shutter button is pressed

would I be correct in thinking that actually using Bias frames in the processing is going to cause problems with the results... That would explain some of the oddities I've been seeing.

Link to comment
Share on other sites
brianb

brianb

Posted
Posted
would I be correct in thinking that actually using Bias frames in the processing is going to cause problems with the results...

If you are taking a "bias" frame by setting the shutter speed to 1/8000 and firing the shutter with the lens capped, what you will end up with is essentially a subtraction of two true bias frames, i.e. random noise, but at a fairly low level. Processing that in shouldn't have a particularly serious effect on the image, just reduce the s/n ratio a bit.

Must admit that when imaging with a DSLR I've never bothered trying to shoot & subtract bias frames because my understanding was that the camera does it automatically, in fact I usually let the camera do the dark subtraction as well by enabling "long exposure noise reduction". It seems to work reasonably well, and I'm lazy :)

Link to comment
Share on other sites
jgs001

jgs001

Posted
Posted

I've had a few that went green when I applied the extra bias frames... no clue as to why. I won't use them anymore. I've also stopped using darks... as the 450d seems to work very well and get decent results with only flats applied.

Link to comment
Share on other sites
narrowbandpaul

narrowbandpaul

Posted
Posted

john......

yes use the same 100x100 pixels each time....

most software programmes allow you to choose which pixels you want to crop.

find the average and standard deviation as before.

as long as no pixels are overexposed, any exposure is fine.......overexposing leads to a poor measurement of std dev

arad.....no worries. beer is my vice! oh, and coffee

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

Go to topic listing

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue. By using this site, you agree to our Terms of Use.