An attempt at Noise Reduction

[ajohnw]

The little clump is the eagle Christina. The big spike is the sky. There will be some eagle in the big spike as well. If the exposure had been correct the eagle aspect of the histogram would be broader. The sky just about uses up all of the dynamic range and the eagle very little. In a back lit shot like that a decent exposure for the eagle is bound to clip the sky as the dynamic range needed to get both is well over what the camera can handle. How much to compensate on the meter readings in this situation depends on the camera.

The dark end, dark subjects and low light are difficult but if it's a situation where there are no highlights to clip or that doesn't matter some degree of over exposure is the best option. The camera histogram can give you a rough idea of the scope for this - use up some of the spare space on the right or decide how much to clip off.

In both cases the idea is to get the subject out of the noise region. In the case of the eagle I don't thin noise would be a problem if the exposure had been corrected. Saying it can't be seen when the eagle shot is reduced to web size doesn't really help as the eagle still wont look right,

Personally when things are that dark I don't think it's a good idea to just increase the exposure. If you look carefully at the shot I posted you will see some colour artefacts. That's not noise it's what happens when dark areas are lifted up. The point to stop at is prior to when they become obvious.

I don't have a shot such as the one you requested but I do have something from a jpg which is essentially what can be done from raw using an S curve and then processing that. i have just brightened up 2 areas as far as they can go. The shot was taken to get zero clipping in the sky.



Using and insane curve out of curiosity from raw I extracted as much as possible but as you can see the shot is then unusable.



It's all a case of getting the subject in the usable range of the camera.

The final version was like this one which looks to me like I over sharpened a bit.



John
-
PS just noticed from the clouds that I have made those from 3 different shots - you'll just have to trust me.
-

[Brownbear]

Thank you for sharing... The little clump is the eagle (: but I'm delighted to know that because it will help me do better next time around.

Your explanation is very informative and helpful. Beautiful image. I do. Thanks again.

The little clump is the eagle Christina. The big spike is the sky. There will be some eagle in the big spike as well. If the exposure had been correct the eagle aspect of the histogram would be broader. The sky just about uses up all of the dynamic range and the eagle very little. In a back lit shot like that a decent exposure for the eagle is bound to clip the sky as the dynamic range needed to get both is well over what the camera can handle. How much to compensate on the meter readings in this situation depends on the camera.

The dark end, dark subjects and low light are difficult but if it's a situation where there are no highlights to clip or that doesn't matter some degree of over exposure is the best option. The camera histogram can give you a rough idea of the scope for this - use up some of the spare space on the right or decide how much to clip off.

In both cases the idea is to get the subject out of the noise region. In the case of the eagle I don't thin noise would be a problem if the exposure had been corrected. Saying it can't be seen when the eagle shot is reduced to web size doesn't really help as the eagle still wont look right,

Personally when things are that dark I don't think it's a good idea to just increase the exposure. If you look carefully at the shot I posted you will see some colour artefacts. That's not noise it's what happens when dark areas are lifted up. The point to stop at is prior to when they become obvious.

I don't have a shot such as the one you requested but I do have something from a jpg which is essentially what can be done from raw using an S curve and then processing that. i have just brightened up 2 areas as far as they can go. The shot was taken to get zero clipping in the sky.



Using and insane curve out of curiosity from raw I extracted as much as possible but as you can see the shot is then unusable.



It's all a case of getting the subject in the usable range of the camera.

The final version was like this one which looks to me like I over sharpened a bit.



John
-
PS just noticed from the clouds that I have made those from 3 different shots - you'll just have to trust me.
-

[Richard Lundberg]

Won't work.

You need to open-up to keep the shutterspeed up, and with a long lens and wide-open aperture the DoF is modest, at best.

Eg

- 200mm lens @ F2.8 - with the bird at 30m, DoF is only 3.8m, and the hyperfocal distance is 471m.

- With a 200mm lens at F8 the hyperfocal distance is still 167m, and a good shutterspeed of, say, 1/2000th @ F2.8 will have dropped to 1/250th (or alternatively you'll end up increasing the ISO from 100 to 800, which will start to eat into required dynamic range considerably.

Look at the difference in the focus between subject and background; no way even at F8 you'll get any significant detail at the working distances you'd need if focusing to infinity.

I'll be damned. So that's why no one can get a decent picture of a UFO.

[Colin Southern]

I'll be damned. So that's why no one can get a decent picture of a UFO.

You do the math then tell us how few pixels would be representing the bird at the closest distance it's in focus when focused at infinity. Assume 200mm lens.

[ajohnw]

Bird photography often involves working near pixel peaking levels so actual dof wont be as much as web calculators suggest. Best option is to focus on it and pray.

This touches on something mentioned in another thread 200mm is plenty on m 4/3 for bird photography in the garden. I suppose some might have albatrosses flying about but often they are tiny at shooting distances. Eg a nvg shot of a robin at 20m at the most - 600mm equivalent on 35mm. This is near on 100% resolution. Reduction was 3/4 from memory.



Cold and wet so shot through the window at an angle mainly to see what low reduction in size means in terms of pp. There would be some scope for reduction if I had switched the 2x converter on but not much. I'm pretty impressed with the Oly 75-300mm going on this shot and the feather spikes at the base of the bird. Suspect 1 stop less would have been a good idea.

John
-

[Geoff F]

A little more general information about Blend Modes, Christina, and explained better and more fully than I could manage. But there are several pages about the subject!

http://ronbigelow.com/articles/blend1/blend1.htm

[Brownbear]

Thank you Geoff. Easy to understand and very interesting, ie; blend modes

A little more general information about Blend Modes, Christina, and explained better and more fully than I could manage. But there are several pages about the subject!

http://ronbigelow.com/articles/blend1/blend1.htm

[Richard Lundberg]

You do the math then tell us how few pixels would be representing the bird at the closest distance it's in focus when focused at infinity. Assume 200mm lens.

Math class is in the other building, this is Photography 101.

[Colin Southern]

Math class is in the other building, this is Photography 101.

Then I suggest that if you're going to be one of the teachers of Photography 101 then you don't lead the students astray with bad advice, including shooting BIF with focus set to infinity and at F8.

[Richard Lundberg]

I'm not the teacher, I'm auditing this class. Christina probably can't do what she needs to do without an ISO that looks like the cruising altitude of an SR-71.
Nice bird, John. I guess you thought the background was blown out a bit. i think it looks good the way it is.

[Colin Southern]

I'm not the teacher, I'm auditing this class.

To be blunt Richard, you're not qualified to audit this class. You'd do a lot better keeping quiet and learning from it though.