[dosdan]

an image pushed 6-10 EV always have more noise than one pushed only 2-3 or not pushed at all

True, if you're pushing from base ISO, but not if your pushing from above the "plateau", but it needs further comment. Assuming that you're use ISO80 as the base ISO, 6 stops (2^6) = ISO5120, while 10 stops = ISO81920. 6 stops is just usable, but the later is too far for any APS-C camera and for most FF cameras too, regardless of what method you use.

1. If there is no NR, the difference in pushed vs. upping-the-ISO operation should only be a max. of 0.3 stops (see the 3rd point in first message in this thread - this is from Guillermo Luijk - Falk Lumo's figure is about 0.5-0.6 stops).

2. With the K-5 there is on-sensor-chip NR above ISO1600. So, for example, a 5 stops push from ISO80 -> ISO2560, will probably have a slightly greater difference due to the introduction of this NR into the mix.

Now some don't mind this NR; others do. Since it occurs on the chip before demosaicing, the visual degradation from its use is reduced. However if you want to avoid this on-chip NR completely when operating above ISO1600, just push from ISO1600 (underexpose & boost the EV in PP). This will give a slight SNR improvement over pushing from base ISO and will get around the limited EV adjustment range of some current raw converters. (Note: all above-ISO1600 settings in-camera are digital multiplications anyway.)

So, instead of selecting ISO3200 or ISO6400 when shooting raw, leave it on ISO1600 and use +1EV or +2 EV boost in PP instead. This will then allow you to apply your own type of NR in PP.

By the way, the 0.3 stops is the maximum difference for a push from ISO100 to ISO1600. So in smaller pushes like a 2 stop push from ISO100 -> ISO400 or from ISO400 -> ISO1600, there is even less difference. So while this sensor benefits slightly from upping the ISO, the difference is not much at all.

Dan.

---

Last edited by dosdan; 04-07-2011 at 05:49 PM.

[Pompous Moronox]

By the way, the 0.3 stops is the maximum difference for a push from ISO100 to ISO1600. So in smaller pushes like a 2 stop push from ISO100 -> ISO400 or from ISO400 -> ISO1600, there is even less difference. So while this sensor benefits slightly from upping the ISO, the difference is not much at all.

It's about 2/3 stop of DR what ISO 1600 buys over ISO 100 (dxomark).

However, I don't know how pattern noise (the little that there is with this sensor) is influenced, if at all, by the change of ISO. When doing supermassive pushing, any change in this can make or break the image.

[simico]

It's about 2/3 stop of DR what ISO 1600 buys over ISO 100 (dxomark).

However, I don't know how pattern noise (the little that there is with this sensor) is influenced, if at all, by the change of ISO. When doing supermassive pushing, any change in this can make or break the image.

In practice I couldn't really notice much difference... maybe it's just my eyes.

Full size examples (not for those still on slow internet):
Shot at ISO 1600
Shot at ISO100 and pushed 4 stops (equivalent to ISO1600)

I also have some old night cityscape shots and pushed shots aren't really much worse than "proper" shots.

[Gimbal]

I miss the ISO in a way. I tried some astro shoots back in January and couldn't go longer on shutter time due to earth rotation so I figured maybe if I try to up the ISO? But na, what's the point.

---

Last edited by Gimbal; 04-12-2011 at 10:54 AM.

[Smeggypants]

In practice I couldn't really notice much difference... maybe it's just my eyes.

Full size examples (not for those still on slow internet):
Shot at ISO 1600
Shot at ISO100 and pushed 4 stops (equivalent to ISO1600)

I also have some old night cityscape shots and pushed shots aren't really much worse than "proper" shots.

The pushed image is clearly noisier, but not to any huge degree.

however the issue is how much leeway do you have for post processing in the pushed image compared to the "shot at 1600" image

[falconeye]

The pushed image is clearly noisier, but not to any huge degree.

One has to be careful in pushing ISO in PP. One trap in LR is black level. If it isn't set to zero prior to pushing then the result is ugly. Moreover, Pentax clips the black at higher ISO levels which results in darker (but less noisy) shadows. Eventually, when pushing a lot, one would need a per-color option to clip the blacks which LR does not offer. W/o this option, dark tones tend to become tinted.

If done properly, there shouldn't be a visible difference between a pushed and high ISO image with the K-5.

So, the perfect PP software for "ISOless" sensors has yet to appear.

[Smeggypants]

So, the perfect PP software for "ISOless" sensors has yet to appear.

So has the perfect sensor

[simico]

One has to be careful in pushing ISO in PP. One trap in LR is black level. If it isn't set to zero prior to pushing then the result is ugly.

I used an early alpha version of Raw Therapee 3, every parameter was set to zero. The only change was to set +4 EV for the pushed image.

Coming from the K-m and having seen (and developed for comparison) many K-7 raw files, I'm happy to take either pics - they are unbelievably clean for ISO1600. I seldom use minimal NR even on ISO3200 shots.

[Class A]

W/o this option, dark tones tend to become tinted.

Doesn't the "Shadow tint" slider in the "Camera Calibration" panel address this?

[falconeye]

Doesn't the "Shadow tint" slider in the "Camera Calibration" panel address this?

Maybe. Frankly, I don't know what this slider does.

[Smeggypants]

Hey Falk, some thing maybe you can answer ...

Regarding light as packets ( Photons ) - Given that light is simply a range of frequencies of electromagnetic energy then why aren't radio waves, or X-rays measure in packets?

[philbaum]

One has to be careful in pushing ISO in PP. One trap in LR is black level. If it isn't set to zero prior to pushing then the result is ugly. Moreover, Pentax clips the black at higher ISO levels which results in darker (but less noisy) shadows. Eventually, when pushing a lot, one would need a per-color option to clip the blacks which LR does not offer. W/o this option, dark tones tend to become tinted.

If done properly, there shouldn't be a visible difference between a pushed and high ISO image with the K-5.

So, the perfect PP software for "ISOless" sensors has yet to appear.

Falk, thank you for sharing that.

Can't also the camera settings result in tinting in the shadows. Say for example, one normally sets a -2ev to ensure unclipped highlights, then inadvertently underexposes the picture. Where this unexposure clips red for example, more than blue or green, wouldn't subsequent processing produce unnaturally tinted pictures, even wtih the perfect software?

[falconeye]

Falk, thank you for sharing that.

Can't also the camera settings result in tinting in the shadows. Say for example, one normally sets a -2ev to ensure unclipped highlights, then inadvertently underexposes the picture. Where this unexposure clips red for example, more than blue or green, wouldn't subsequent processing produce unnaturally tinted pictures, even wtih the perfect software?

You mean clipping in the black?

The dark tones do not exactly clip. They become very noisy. But you are right, if 2^-14 is underrun frequently enough, there could be a tinting effect when say green is pushed back to visible while red and/or blue stay zero. But I'd say that such surfaces are too noisy and should never become visible anyway. The tinting is more likely to occur if dark levels are clipped above zero.

[Class A]

Given that light is simply a range of frequencies of electromagnetic energy then why aren't radio waves, or X-rays measure in packets?

AFAIC, there's nothing "simple" about light, if you want to get to the bottom of things.

Quantization (e.g., regarding light as photons) applies to all electromagnetic waves. The wave/particle duality furthermore also applies to particles, such as electrons. Even they can behave as waves (-> de Broglie wavelength).

[Eruditass]

Anyone notice what the X100 does above ISO1600?


Fujifilm X100 CLEVER ISOs above ISO1600

The good thing here is that unlike in most cameras, the Fuji X100 shooting above ISO1600 doesn't care of digitally mutiplying the RAW data by the corresponding factor of 2 before saving the RAW file. In this way, ISO3200 on this camera is just ISO1600 plus some metadata telling the RAW developer 'hey! this is supposed to be ISO3200, so display it to the uninformed user 1 stop brighter than it really is'.

What is the good news? the X100 RAW shooter above ISO1600 gets a correct exposure in the camera display and in-camera JPEG, but keeps in the RAW file all the highlights information that other cameras stupidly ruin just for multiplying the RAW data to achieve the illusion of high ISO.

Also,

An interesting trick for faster shutter speeds might be:

Set sensor to base ISO.
Take a reading
Determine maximum sensor reading
Reset analog gain (ISO) so that the maximum sensor reading is just below full scale. (maybe 95%?)

Many of us wish for such a metering mode that simply doesn't highlights (or in the case of a shot with the sun, lets no more than one metering segment be clipped)