[figmental1978]

Here's the EXIF data from flickr.

[Steve Beswick]

I have all that you have mentioned (check my signature).
What I think I need is a fast 70 or 100 like you have.
I prefer to avoid flash if possible.

I don't have one yet, but a Jupiter-9 is an 85mm f2.0 screw mount lens. They are really inexpensive for an 85mm lens and it tends to get really good reviews. Supposedly it has wonderful bokeh.

[marcdsgn]

... using lightroom I can recover quite a bit from the shadows. ... Intentionally shooting to the left (underexposure) will allow faster shutter speed.

A good example of this:

A macro shot of a tiny frog (15mm from snout to butt) taken with the K10D with two things against nailing the shot:

1) It was windy, so it was hard to keep the subject in focus; and
2) I couldn't get close enough for a tight crop.

So I closed the aperture to extend the depth-of-field on a high shutter-speed, sacrificing sufficient light in exchange for less blur.

LEFT is the unprocessed shot. RIGHT is the post-processed result (In PHOTOSHOP, not LIGHTROOM).

[Canada_Rockies]

A good example of this:

A macro shot of a tiny frog (15mm from snout to butt) taken with the K10D with two things against nailing the shot:

1) It was windy, so it was hard to keep the subject in focus; and
2) I couldn't get close enough for a tight crop.

So I closed the aperture to extend the DOF on a high shutter-speed, sacrificing sufficient light in exchange for less blur.

LEFT is the unprocessed shot. RIGHT is the post-processed result (In PHOTOSHOP, not LIGHTROOM).

Nice work, Marc. Lightoom will do the same exposure adjustments and crops as CS. What is missing is a whole lot of things, but those two are there.

[GerryL]

I think the principle is the same as "pushing" the ISO in film.
You have an ISO 200 for example but set the setting to ISO 400 for a push of 1 stop.
The film becomes a bit grainy and there's more contrast to the highlights and shadows.
In digital, you just do it in post processing (software).

[marcdsgn]

Lightoom will do the same exposure adjustments and crops as CS.

Yes, I assumed as much. I just thought I'd better be pedantic about my methods, in case people thought they needed Lightroom to do this. I suspect many (if not most) basic PP programs will also do the same thing. In PHOTOSHOP, the Lightness/Contrast palette is usually all I need for this kind of adjustment.

[er1kksen]

Sometimes flash is acceptable, sometimes it is inappropriate.

When shooting without flash, underexposing and "push" processing is often necessary even with a relatively fast lens (f/2.8). Well, maybe not so much with the K20D if you use ISO 6400, but for cameras that max out at ISO 1600 or ISO 3200, it can be. Depends on whether your subjects are being still enough that you can get away with slow shutter speeds like, say, 1/15".

I don't know about you, but I'd really rather not use my K20D at ISO 6400, as there's severe blue banding on the top and bottom of the frame. Underexposing at ISO 1600 and pushing sounds like a safer strategy (I believe 6400 is actually just pushed 3200 in the K20D, so shooting at 3200 is out as well).

[Marc Sabatella]

I've seen that severe banding you refer to, but on the other hand, I've also seen ISO 6400 images from the K20D that are pretty amazing. Noise levels are very dependent on lots of factors that aren't always obvious. The more you can "expose to the right" the better, of course. A really well-exposed image at ISO 6400 may look as good as a push-processed ISO 1600 image. A poorly exposed ISO 6400 image will look bad, but if it's poorly exposed at ISO 6400, it's going to need a *ton* of push processing at ISO 1600 and will probably look just as bad if you do it that way. You really want to get exposure to be "good" at ISO 6400, or get within two stops (preferably *well* within!) of "good" at ISO 1600 - in which case, I'd expect fairly similar results either way.

But when I said there ar non-obvious factors at work too, I've also noticed (with my K200D, anyhow, but it should apply to all digital cameras) that color of light plays a big role. Strongly colored red light - as is often the case with spotlights - means you are using only the red pixel sites on the sensor; the others ones might as well not be there since there is no blue or green in the image. That means you're magnifying noise by a factor of 4, since red pixel sites make up only a quarter of the total. "Yellow" or "Orange" light just means you get to use the green pixel sites too, but at a much lower level, meaning you're still dealing with a lot of noise.

Then there is the contrast inherent in the scene. High contrast measn you have both light and shadow areas to deal with, and noise loves shadows. If it's a scene where you can get away with a strong S surve to keep the shadows dark, then you can hide the shadow noise. But if the interest is in the shadows, you can't use that trick, meanign the nosie will be noticed.

The kinds of texture and detail in the scene matters too. Some types of texture show noise more than others, and some kinds of details are more affected by noise as well. Lots of fine detail will often be obliterated by noise. But completely flat areas will show nosie more prominently as well. Wood, cloth or even skin can often be good, because the noise can disguise itself as ordinary texture making it less noticeable.

All of this is why posting shots of unrelated scenes as a means of comparing noise levels between ISO settings - or between cameras - is absoutely pointless. Difference in noise levels that are caused by the differences in the scenes themselves can often be larger than the differences caused by the differences between cameras.

[er1kksen]

That's all very true. My experience with photographing stage performances, which are probably not dissimilar in lighting to the situations the OP is interested in, is that they are exactly the sort of situations that you refer to as causing the worst noise problems. Mostly very warm lights, and dark backgrounds that show swaths of blue banding like visible clouds that might not be noticeable in a more evenly-lit scene, leading me to suspect that pushing an ISO 1600 exposure, while it might even result in a higher overall level of noise throughout the scene, may be a way to avoid that banding (which is more disruptive to an image than ordinary noise, imo).

[kristoffon]

I don't know about you, but I'd really rather not use my K20D at ISO 6400, as there's severe blue banding on the top and bottom of the frame. Underexposing at ISO 1600 and pushing sounds like a safer strategy (I believe 6400 is actually just pushed 3200 in the K20D, so shooting at 3200 is out as well).

Underexposing ISO 1600 and pushing IS JUST THE SAME AS SELECTING A HIGHER ISO AND EXPOSING PROPERLY.

Whatever the ISO you select the sensor captures the EXACT same data. ISO just selects the range from the sensor's latitude that gets translated into viewable pixels.

That's why ISO 100 has lowsy highlight recovery capability and why I almost never shoot ISO 100 - the sensor is nearly saturating. Shooting ISO 1600 -1 EV you're just fooling yourself - the data is the same as if you shot ISO 3200.

[Chwisch87]

However, wouldn't ISO 3200 create more digital noise?

I always shoot in raw since as a format it is very easy to correct a mistake in exposure of what would otherwise be a useless shot.

[Ash]

Deja vu from the ETTR thread concurrently going on...
All of this is true as long as the sensor's dynamic range is not exceeded

[pedrojoper]

You will get a lot of noise, but far better than the JPEGs. Consider trying DxO optics to convert the RAW files, their noise reduction is pretty awesome...