dpreview iso comparison

boyfalldown said:
the histograms?. They were as close as I could get them, the main difference is in the blue channel in the highlights,
Click to expand...
it was in reply to Phil's comment about underexposure. the histogram shows both shots are exactly same exposed.

here's a theory, the extra noise at lower lighting/longer shutter could be due to sensor heat during the longer exposure? no idea how to test this......
 
boyfalldown said:
am ashamed to admit I don't know how to do that. Give me a quick pointer I will
Click to expand...

I'm not sure if it can be done in Lightroom, but in Photoshop: Filters>Noise>Reduce Noise, then select Advanced, then pick blue (and set preserve details down to zero).


woof woof said:
I've just tried spot and the histograms look different. I didn't try setting a different shutter speed to what the metering set to try and get the histograms closer together.

Is this all going too far? :wacky: :D
Click to expand...

It's different because when you use a dimmer switch the filament temperature changes, changing the spectral mix of the light emitted as well as the intensity (skews to longer wavelengths). You get more red, less blue (to simplify).
 
Phil Young said:
Its either my phone or my eyes or my head but the second looks slightly brighter.

The keys appear to have more shine...is it just me?!
Click to expand...

That's not what I'm asking.

I'm asking whether you've seen a simple third stop change in exposure ever have that effect on noise.

Same ISO, same aperture, just a little longer SS. Because that's what you're claiming if you're attributing the change in noise to the difference in exposure settings.
 
Phil Young said:
The keys appear to have more shine...is it just me?!
Click to expand...

Thats cause the light was brighter on the ceiling, above things. Nothings blown. Really Phil, I ran exactly the test you wanted and now you wish to move goal posts :nono:
 
Ok,

Here's what I say about the test....

The 2nd looks better due to being slightly brighter.

I'm sure if you put 100 or even 200% crops into the thread (great that a 36mp camera was used for this btw) I'm fairly confident there won't be any actual difference in noise but our eyes are "tricked" into thinking there is because of the difference in brightness.

Either way, pop the 100% crops up showing the dark areas and the light areas and let's see the difference :)
 
Rapscallion said:
I'm lost, what are we arguing about...

!
Click to expand...

Didn't it start out as "under exposed = more noise"?

If that was it (I'm lost now too) I'd agree and say always get the exposure smack on or ETTR and back it off in post.
 
OK, noise reduction on only blue channel

4293-EDIT.jpg


which also has an interesting effect on the histogram

Screen-Shot-2012-09-25-at-15_58.jpg


anyway I've a feeling everything worth saying has been said in this thread
 
As you mentioned earlier, you have to be careful with artificial light differences between shots as the wrong kind can make noise much more visible.

Personally I think that drawing the curtains is a better way :D
 
woof woof said:
Personally I think that drawing the curtains is a better way :D
Click to expand...

Sorry to be picky but closing the curtains will totally change the nature of the lighting. You're basically putting a huge diffuser in front of your lights source. I would speculate that is why you cannot get your histograms to look alike.
 
Yes, it made it next to impossible to get the same histogram and maybe it would be less different with artificial lighting but with artificial lighting sometimes you get... noise.

If anyone can get the histogram to match and not introduce more noise I'd say they've done very well :D

PS. No need to be sorry about being picky. A thread like this couldn't exist without being... picky. :)
 
why not just use a flash as key lighting, and set its power to 1/8 and 1/2 for 2 stop of difference?

i may do one myself tonight :D i'm sure got my meerkat model Aleksandr will happy to comply.
 
gad-westy said:
Sorry to be picky but closing the curtains will totally change the nature of the lighting. You're basically putting a huge diffuser in front of your lights source. I would speculate that is why you cannot get your histograms to look alike.
Click to expand...

Don't forget one side of the argument from Scott is that sunlight vs artificial will affect noise.
 
I'd love to contribute to this thread in a more valuable way, but I'm off out to do some photography..... ;0)
 
wuyanxu said:
why not just use a flash as key lighting, and set its power to 1/8 and 1/2 for 2 stop of difference?

i may do one myself tonight :D i'm sure got my meerkat model Aleksandr will happy to comply.
Click to expand...

I agree, but the "dark yet high contrast church" scenario I mentioned earlier usually comes with a Vicar saying "no flash" and a Bride expecting perfect images
 
Has nobody got a strobe handy with a controllable modelling light? That would be the best way to test it at this hour, no outside ambient or flash peaks to interfere with....don't make me get my stuff out of the cupboard
 
tiler65 said:
Has nobody got a strobe handy with a controllable modelling light? That would be the best way to test it at this hour, no outside ambient or flash peaks to interfere with....don't make me get my stuff out of the cupboard
Click to expand...

Do it!
 
So far we need:

A strobe test where the artificial light is adjusted to match the speed.

A daylight & artificial test.

I can do the strobe tonight.

We need a test with curtains so one where the light comes from daylight and another with artificial light.

As long as we take crops with the same exposure in what we are testing that will be enough to prove weather the light source dictates the amount of noise from the sensor.

I can do the strobe test and that will prove weather a very fast shutter speed gives less noise than a slower one.

Are we happy with that?
 
I have my results!
Controlled conditions at ISO 12800 with shutter speeds varying from 1/1600s to 1/3s.
The bottom line is that I can't see any difference in noise levels or fine detail.

Really sorry, but I have to go out and will post the evidence tomorrow.
 
DuncanDisorderly said:
I have my results!
Controlled conditions at ISO 12800 with shutter speeds varying from 1/1600s to 1/3s.
The bottom line is that I can't see any difference in noise levels or fine detail.

Really sorry, but I have to go out and will post the evidence tomorrow.
Click to expand...

<Rubs fingers together>

Excellent....

Look forward to seeing that Duncan.
 
scottthehat said:
just because i said outside and inside phil, come on mate you seem like a bright boy, light in general affects iso.
Click to expand...

Am I misunderstanding what you said Scott...here's how I interpreted it:

If you take a shot in daylight at a fast shutter speed you will get less noise compared to indoor light at slower speeds...

????
 
OMG!!!

I have read some pretty pointless threads on here but this one seems to take the prize (at least for this year!).

Just to add to the discussion here are 2 pics I took 4 years ago with my 350D at 1600 ISO and underexposed by two stops to get an approx 6400 ISO.

I then put them through Neat Image using profiles I prepared myself (100% Quality and 100% match as far as I remember):


805.jpg


804.jpg

​

Nowadays I regularly use 3200 ISO on my Canon 1Ds MkII and process them in the same way to get rid of the noise.

So much easier than worrying about the effects of shutter speed or aperture on the noise!

And I am very happy with the resultant pics.

So much easier than histo-gazing !

Anyway good luck with your deliberations.

.
 
Last edited:
petersmart said:
OMG!!!

I have read some pretty pointless threads on here but this one seems to take the prize (at least for this year!).

Just to add to the discussion here are 2 pics I took 4 years ago with my 350D at 1600 ISO and underexposed by two stops to get an approx 6400 ISO.

I then put them through Neat Image using profiles I prepared myself (100% Quality and 100% match as far as I remember):





Nowadays I regularly use 3200 ISO on my Canon 1Ds MkII and process them in the same way to get rid of the noise.

So much easier than worrying about the effects of shutter speed or aperture on the noise!

And I am very happy with the resultant pics.

So much easier than histo-gazing !

Anyway good luck with your deliberations.

.
Click to expand...

Lol I'm sure we all feel pretty embarrassed to be contributing and going through with tests but its also a good point to make to those that aren't aware:

Bottom line being, expose well and noise isn't a problem.
 
I don't really understand the need to post and say it's a pointless thread. If it's pointless don't post. Simples :D

I found/find it interesting and on a day when I couldn't really go out (biblical rain) or do much of anything really (suffering with a strain/sprain) I needed something interesting :D
 
woof woof said:
I don't really understand the need to post and say it's a pointless thread. If it's pointless don't post. Simples :D

I found/find it interesting and on a day when I couldn't really go out (biblical rain) or do much of anything really (suffering with a strain/sprain) I needed something interesting :D
Click to expand...

Amen to that! I know it's sad. It's seriously sad in fact but I find this stuff interesting too. I've just done my own experiment. Set up a shot with the lights on then turned them off. ISO 25,600. Used on camera flash as my only light source. Varied it a bit to get two totally different amounts of light on my subject but from the same direction. Must admit I can't see much difference between either shot and the histograms look near identical.

I do have some very firm conclusions:

1) ISO 25,600 can create a lot of noise even with a nailed exposure. And these actually look far better than a typical gritty low light exposure that I would expect. I rarely venture north of 6400 in uncontrolled conditions.

2) Operating a camera in the dark isn't that easy

3) 1/1 flash power straight at your face can make your eyes go funny

Here's the science bit, I don't know how you guys post your fancy histograms or 100% crops so you'll just have to put up with the full images.


1/250 by G.A.D, on Flickr


1/45 by G.A.D, on Flickr

As straight out the camera as I can get them though both have been uploaded to flickr via aperture 3 but with zero noise correction applied.
 
Last edited:
woof woof said:
I don't really understand the need to post and say it's a pointless thread. If it's pointless don't post. Simples :D

I found/find it interesting and on a day when I couldn't really go out (biblical rain) or do much of anything really (suffering with a strain/sprain) I needed something interesting :D
Click to expand...

Absolutely.

Call it sad or passionate but when I can't take photos, I like to talk about it.

Its in my nature to know about my subject so I'm always up for learning more or proving to myself what I already know. For me, that is the point to this thread.

I'm Sher others have similar or different reasons which are important to them.
 
I've uploaded the full set of test images to a gallery.
You can download full size JPEGs if that floats your boat.
http://www.wild-landscapes.co.uk/Other/2012/ISO-Tests/25573977_QL64L3
I think I've worked out how to turn off sharpening on linked images. The full size JPEGs will definitely be unsharpened.

Right - Here's the crops.
One is 1/1600s, the other is 1/3s - all other settings are unchanged.
That's roughly 9 stops :|
If you have been following this thread, then it should be obvious which one is which.
The images were taken in RAW and processed using LR4 with no tweaks or settings.

20120925-184927-I39A3586-crop-X3.jpg


20120925-145906-I39A3579-crop-X3.jpg


It's pretty clear that ISO 12800 is never going to make pixel peepers happy.
It was supposed to be an extreme test :shrug:
But I've printed ISO 12800 to A3+ and the prints look great, so this level of noise works for me!

As for differences between the two?
Sure the colour is different, that's what time of day does; but I don't think it is significant here.
Noise and fine detail look spookily similar and well within margins for test conditions.

Edited to add: Here's what the full frame looks like.
20120925-145906-I39A3579-S.jpg


No doubt my experimental technique is going to come under attack.
I'm up for it - fire away :D
 
Last edited:
Phil Young said:
Absolutely.

Call it sad or passionate but when I can't take photos, I like to talk about it.

Its in my nature to know about my subject so I'm always up for learning more or proving to myself what I already know. For me, that is the point to this thread.

I'm Sher others have similar or different reasons which are important to them.
Click to expand...

What he said (y)
My sentiments exactly.

Quoting myself from a thread I linked to earlier on testing lenses...
"P.S. I don't aim for every image I take to be critically sharp - that would be crazy. Taking images is a compromise needed for real world conditions. But at least I know what effect the compromise is having on my finished image, and most importantly, whether it will affect an A3+ print."
 
You must log in or register to reply here.
Back
Top