Max ISO settings with LP filter?

[reddoss]

In the days when I had just an unmoded Canon 450D and no LP filter clip, I found that the maximum ISO setting was 800 without too much noise.

By force of habit I have stuck with this setting since getting a moded camera and a ESO LP filter clip. I have just been processing some subs I took of the bubble with the at ISO 1600. To my surprise they are not too noisy at all.

What ISO settings to other DSLR imagers use with a LP filter clip?

[Shibby]

Does the LP filter affect the noise?

I have a 450d (modded) and tried ISO 1600 out recently. I found the noise level unacceptable but I guess that depends on the temperature at the time.

[themos]

This has been discussed many times!

One problem with subjective estimations of noise in the RAW frame is monitor technology. I think it's just not possible to display the range of brightness levels that the sensor has captured with current LCD screens. So we always end up seeing some kind of "stretched" image that may give a false impression of the amount of noise in an image. What's happening is that we "see" the noise but can't "see" how much stronger the highlights are. Because of this, I have given up trying to gauge, by eye, how noisy a RAW image is. Best to rely on computation to get a Signal-to-Noise ratio.

I have a 450D with a CLS light-pollution clip filter and my default ISO setting is still 800. I think that 400 may well give a better SNR but the difference is small, I believe, maybe equivalent to shooting another sub or two. I meant to finish a set of measurements of noise at different ISO numbers so that I have numbers to back that up but life is short.

Hope that makes sense.

[themos]

I'll add that, theoretically, a light pollution filter will reduce the photon noise of the sky background so in order to remain "photon-noise limited" you would need to reduce your read-noise somewhat which means, other things being equal, INCREASING the ISO setting. But it can very well be that, with a long enough exposure time, you were already so deep in the "photon-noise limited" region that you don' t have to decrease your read noise.

[dph1nm]

I think that 400 may well give a better SNR but the difference is small,

The stuff on the web about Canons shows that the S/N increase with ISO (due to the decreased read noise), so ISO1600 should be better than 400 or 800.

and tried ISO 1600 out recently. I found the noise level unacceptable but I guess that depends on the temperature at the time.

As themos has said, the fact that higher ISOs 'look' noisier is irrelevant. What matters is the S/N.

NigelM

[themos]

The stuff on the web about Canons shows that the S/N increase with ISO (due to the decreased read noise), so ISO1600 should be better than 400 or 800.

Yes, read noise is decreased with higher ISO but shouldn't S/N depend also on the available dynamic range that decreases? After all, that S matters.

[Shibby]

As themos has said, the fact that higher ISOs 'look' noisier is irrelevant. What matters is the S/N.

...which seemed poor for the particular exposure length and temperature that I was working at. I was unable to extract as much detail as I was at 800.

Some interesting graphs here, albeit for a Canon 10D.