Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
High ISO noise vs under exposure correction (RAW)
#21
Now I'm confused!

You wrote "I haven't tried it on my camera but I suppose underexposing and then pushing up in RAW converter produces less noise than ISO boosted images."



Ciao, Walter
#22
[quote name='Walter Schulz' timestamp='1299433333' post='6560']

Now I'm confused!

You wrote "I haven't tried it on my camera but I suppose underexposing and then pushing up in RAW converter produces less noise than ISO boosted images."



Ciao, Walter

[/quote]



Well I'm confused of confusing you and Allan <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='Big Grin' />... Sorry for that.



I supposed so (as you've quoted above), but then I tried it and have seen that pushing up exposure in ACR creates a lot of noise. Also creates too saturated high contrasted images, and the DR is less than the ISO boosted image. The pushed up results of course include manipulation of ACR itself in terms of other IQ settings (twisted color profiles), but I assume, if I had used an untwisted version of ACR, the results would be still worse than the ISO boosted image.



Walter, I've checked the links you've sent. Thanks a lot. I've found some untwisted profiles for D700 and placed them under the related folder to be used in ACR. But apart from the difference regarding the "recovery" slide, the result was more or less the same. I did not have much time for a thorough check but at first glance it seems so...



Serkan
#23
[quote name='PuxaVida' timestamp='1299487256' post='6565']

Well I'm confused of confusing you and Allan <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='Big Grin' />... Sorry for that.



I supposed so (as you've quoted above), but then I tried it and have seen that pushing up exposure in ACR creates a lot of noise. Also creates too saturated high contrasted images, and the DR is less than the ISO boosted image. The pushed up results of course include manipulation of ACR itself in terms of other IQ settings (twisted color profiles), but I assume, if I had used an untwisted version of ACR, the results would be still worse than the ISO boosted image.



Walter, I've checked the links you've sent. Thanks a lot. I've found some untwisted profiles for D700 and placed them under the related folder to be used in ACR. But apart from the difference regarding the "recovery" slide, the result was more or less the same. I did not have much time for a thorough check but at first glance it seems so...



Serkan

[/quote]



I dont know what you mean by untwisted profiles but in my test images, the contrast/brightness/saturation/recovery/blacks sliders were exactly the same for both images - I didnt use 'auto'.
#24
[quote name='allanmb' timestamp='1299498793' post='6570']

I dont know what you mean by untwisted profiles but in my test images, the contrast/brightness/saturation/recovery/blacks sliders were exactly the same for both images - I didnt use 'auto'.

[/quote]



The "twisted color profiles" is an expression Walter pointed out, which is used for standart color profiles of ACR. When I push up exposure in ACR, it's not only the brightness which is adjusted. ACR does other manipulations on the image, and maybe this is why I've seen a big difference between two images with the following exif data:



f/5.6, 1/640, ISO 800

f/5.6, 1/640, ISO 200



On the image with ISO=200, the exposure was pushed up (+2) in ACR to have the same brightness with ISO 800 image (and nothing else was touched). The resulting image was clearly worse than the ISO 800 image in terms of IQ. It has more noise, over-saturated colors with high contrast and also has a less DR.



Serkan
#25
[quote name='PuxaVida' timestamp='1299500610' post='6572']

The "twisted color profiles" is an expression Walter pointed out, which is used for standart color profiles of ACR. When I push up exposure in ACR, it's not only the brightness which is adjusted. ACR does other manipulations on the image, and maybe this is why I've seen a big difference between two images with the following exif data:



f/5.6, 1/640, ISO 800

f/5.6, 1/640, ISO 200



On the image with ISO=200, the exposure was pushed up (+2) in ACR to have the same brightness with ISO 800 image (and nothing else was touched). The resulting image was clearly worse than the ISO 800 image in terms of IQ. It has more noise, over-saturated colors with high contrast and also has a less DR.



Serkan

[/quote]



I doubt it has anything to do with the colour profile. Its just the fact that its better to get the image exposed correctly before the A/D converter than afterwards. Its the same as an analogue music cassette, those with low recording levels would sound a little muffled and had more hissing noise when played at higher volumes. Those that were recorded at a better level would sound better all round.



(Also, I meant to say that you will notice a bigger quality difference in low light, ISO 800 f/5.6, 1/640 is quite bright and will give more defined light/shadow transitions. I expect the settings I did it at will show the quality difference a little more due to the lack of general light)
#26
[quote name='PuxaVida' timestamp='1299500610' post='6572']

The "twisted color profiles" is an expression Walter pointed out, which is used for standart color profiles of ACR. When I push up exposure in ACR, it's not only the brightness which is adjusted. ACR does other manipulations on the image, and maybe this is why I've seen a big difference between two images with the following exif data:



f/5.6, 1/640, ISO 800

f/5.6, 1/640, ISO 200



On the image with ISO=200, the exposure was pushed up (+2) in ACR to have the same brightness with ISO 800 image (and nothing else was touched). The resulting image was clearly worse than the ISO 800 image in terms of IQ. It has more noise, over-saturated colors with high contrast and also has a less DR.



Serkan

[/quote]

It's difficult to expect a linear outcome when most things aren't linear.



Sensors for example have 3 colour channels (as we know) and all three should have different response curves (efficiency that varies with the amount of light going in). A "decent" amount of light is necessary for the "ordinary" response case that we know of but if the image is underexposed (to be pushed in post later), I'm sure the response will not be mathematically proportional to reduction in actual exposure.



The above will also lead to a change in colors (hue and saturation, regardless of brightness) depending on the exposure (because the 3 channels have independent response curves). So there can be colour shifts depending on the level of exposure and maybe ACR corrects these in a way which results in images that are slightly different to properly exposed shots. Or maybe ACR doesn't correct them at all. Either way, you're bound to see a difference.



GTW
#27
[quote name='genotypewriter' timestamp='1299550000' post='6592']

It's difficult to expect a linear outcome when most things aren't linear.



Sensors for example have 3 colour channels (as we know) and all three should have different response curves (efficiency that varies with the amount of light going in). A "decent" amount of light is necessary for the "ordinary" response case that we know of but if the image is underexposed (to be pushed in post later), I'm sure the response will not be mathematically proportional to reduction in actual exposure.



The above will also lead to a change in colors (hue and saturation, regardless of brightness) depending on the exposure (because the 3 channels have independent response curves). So there can be colour shifts depending on the level of exposure and maybe ACR corrects these in a way which results in images that are slightly different to properly exposed shots. Or maybe ACR doesn't correct them at all. Either way, you're bound to see a difference.



GTW

[/quote]



Thanks for the explanation. It's obvious that increasing ISO is the better alternative.



Serkan
  


Forum Jump:


Users browsing this thread:
2 Guest(s)