My top three bits of advice are:
- Shoot fast lenses (2.8 or faster)
- Shoot RAW format (and underexpose if necessary)
- Use third-party noise reduction (such as Neat Image or Noise Ninja) to offset high ISOs

I'm sure others will have more ideas.

> (Tele?) zoom... (whats the dif tween the 2, btw?)

Telephoto means it magnifies (e.g. like binoculars do). Zoom means it has a *variable* focal-length... some, all, or none of which might be telephoto. You can have a "wide zoom" which is all wide-angle. You can have a "tele zoom" which is all telephoto. etc...


> Have only been consistently disappointed by my shots in low light situations...... How do i improve them...?

In addition to the (excellent!) suggestions already made, I'd ask: what's disappointing you? Without knowing what's wrong, we're only able to give general syuggestions...


- Steve

Shoot raw.

Use a tripod or monopod if you can.

Use a fast prime. The 50mm f/1.8 is dirt cheap and worth its weight in gold

Boost the ISO.

Learn when you can trust the camera's meter, and when you should go manual.

Use strobes if you can.

Patti's suggestion to get a 50mm prime lens is a good one. Shooting RAW is good. Using a soft box is also good -- the bigger the better and the closer the better (within reason). You can get some very nice results using just the one strobe and a large softbox. For even better results add a reflector or another strobe to open up shadows. I use Alien Bees and like them a lot. Get a book on studio lighting for ideas on how to set up your lights. Get a flash meter and take multiple readings: directly at the camera, directly at the main light with the fill light blocked off, and directly at the fill light with the main light blocked off. Make sure there's little or no extraneous lighting or at least make sure it's accounted for and adds to the effect. And lastly, preview your shots.

Hi stefographer,

Let me set what I can add beyond what others have posted ...

You'll want to learn as much as possible about your camera. Here's a nice review encapsulating a lot of it's capabilities:
www.dpreview.com/news/0801...os450d.asp

One of the first things to note about your camera is the more limited Field of View (FoV), or "cropping factor" of it's optical system. What this means is that only some of the circle of light which exits the lens (the output image circle) is used by the image sensor in the camera. So, only a fraction of the view is ever used by your camera. In the case of your camera, it's about 60%. Such a smaller reduced view is effectively a form of magnification ... your whole frame records less of the total view through this lens, because the camera works kind of like looking at it closer. The FoV for your camera is 1.6X. This is a magnification factor that will affect the possible view through every interchangeable lens you place on the camera. Any interchangeable lens on a film SLR, or a full-frame DSLR would not have this phenomena, and so this cropped view effect must be considered in composition. An artifact of the magnification is that it's beneficial for longer telephoto views, but much less favorable for wide-angle views. So, effect of FoV tends to "shift" the possible views, because a lens at a shallow wide-angle can instead become a "normal" view, and a truly wide-angle view becomes impossible. All these things enter into what you can compose in the frame.

For quick comparison, here's an equivalent focal length table for your camera, given fairly common lens lengths:
LENS On an XSi:
24mm 38.4mm
28mm 44.8mm
35mm 56mm
50mm 80mm
70mm 112mm
85mm 136mm
100mm 160mm
135mm 216mm
200mm 320mm

What are some things to observe about this table? Well, the 28mm and 35mm "wide-angle" work pretty close to the "normal" view of 50mm in the 35mm camera system. The 50mm or "normal" lens becomes a telephoto view of 80mm. For lenses like 70mm and longer, you definitely have a telephoto view. Nicely enough, this "extra reach" comes at zero optical loss. You are just looking at less of the full view through the lens, but f/2.8 is still f/2.8 in that view.

You asked several questions, let me deal with low-light first.

First, I'll echo Patti's comment that the "Nifty Fifty," or 50mm f/1.8 lens, is worth it's weight in gold. I just picked one up as a secondary lens for low-light work. If you ultimately have the budget, however, I cannot recommend enough the 50mm f/1.2L, the 85mm f/1.2L II, and the 135mm f/2L lenses, for low-light work. They are just fabulous, and it's the only way I've found one can get great belly dance photos in typical crappy club lighting.

One specific technique for using camera raw that was mentioned was underexposure. Maybe going into that more could be helpful? Basically, you take photos while deliberately setting the exposure downward, so that you can have (most likely) a better shutter speed to freeze movement, or (sometimes) to have more choice in aperture.

The reason this can work so well is that your camera records camera raw data at 14-bits per color. For the full red, green, and blue (RGB) array of pixels for each frame, this is basically 42-color, or 3 x 14-bits per pixel, after a transformation step is done to the camera raw data. The actual tiling of pixels for camera raw is not really 42-bits in it's native form, because it is a mosaic of individual 14-bit colors, one pixel for R, G, and B. Various available software transforms the tiled array, or mosaic, of camera raw pixels into a real photograph having all three colors stacked on top of each other as RGB. This transformation step is called demosaicing, an essential step in obtaining JPEG, PNG, or TIFF photos from your camera data. There other optimizations possible aside from demosaicing using camera raw conversion software, and these are actually valuable things for the aesthetic results you can obtain. This would be things like choices for sharpness, and the particular tone range curve to be used. But the main thing is conversion of a mosaic into a conventional digital photo, where all of the three primary colors are stacked on top of each other for each pixel.

For XSi camera, if you output from camera raw a lossless data format like TIFF, you can obtain 42-bit color, each primary color having a 14-bit range. But you can also output well-exposed JPEGs (which are only 24-bit color, 8-bits per RGB), using under-exposed camera raw.

For low-light work, the secret to this technique is that camera raw has enormously large dynamic range. Each bit of data represents approximately a 2:1 ratio of light. In photographic terms, this ratio is virtually the same as 1 stop of exposure. But with 14-bits available per color, the full dynamic range possible for each color is 2^14 : 1, or 16384 : 1. This is a very large dynamic range. Some high-quality black-and-white films have a dynamic range of approximately 10 stops, which is 2^10 : 1, or 1024 : 1. Things are not quite this simple, because a linear (straight-line) dynamic range is not really how people see in the world, and in fact, the response curve for how we would see equal changes in highlights and dark tones is not a straight-line, it's curved.

But the point here is that you tremendous LATITUDE with camera raw on your DSLR, because if expose downward one stop, you lose perhaps 1-bit of 14-bits, leaving 13 of 14 available bits free to represent your image. You just tell the conversion software to brighten up your image one full stop, and suddenly it's "well-exposed," not "under-exposed" anymore. So, that's a low-light camera raw conversion trick, in a nutshell.

The idea is certainly extendable, but only to a point. If you underexpose two stops, you effectively lop-off 2-bits of 14 total. This leaves 12-bits, which was the full dynamic range of the DSLR camera generation just before your Rebel XSi. But several vagaries about the quality of demosaicing (literally!) enter into the picture if you underexpose this aggressively.

Others views may differ here, but my experience is that the most you really want do is 1 and 1/3 stops of underexposure, to still retain very high image quality after camera raw conversion with software-adjusted brightness increases. Try it and see!

Some comments on a portrait lens, versus telephoto. "Telephoto" is a view, whereas a portrait lens is an application for a lens. The "telephoto effect" starts at about 85mm, and certainly almost anything about 100mm is going to be a telephoto lens (except for certain life-size or macro lenses in this focal length). A good portrait lens has several qualities, but an important one is sufficient magnification to bring (for example) the face into the frame in a flatter way, generally without distortion. A lens with a telephoto view is a good way to do that, and a 50mm lens is NOT a good way to do that. If you try to get someone's face to fill the frame with a 50mm lens, even after including the 1.6X FoV factor, you still have to move uncomfortably close to them in order to obtain that view ... and the result is a bulbous, distorted face. Not too glamourous. If instead, you use a 100mm or large focal length lens, the view is much less distorted.

I consider my 135mm prime lens to be my "portrait lens," because it provides the telephoto effect at a sufficient distance where people's features are undistorted. It's really sort of amazing ... we do some work with 2 models in the studio using a view with a 24mm-70mm zoom lens, and then we do the same exact poses again using the 135mm view ... and the view is instantly much more intimate and close. It really (literally) changes the perspective. For example, two laughing girls seen with a 135mm view can look like they're sisters, but with a 50mm view, you might be wondering what the photo is even about, given the same pose.

On my cameras, the FoV is either 1.0X (full frame), or 1.3X; so the 135mm works out as either 135mm, or as about a 175mm (on the 1.3X FoV camera). One result is that I sometimes will use a 70-200mm zoom at the 200mm end as a stand-in portrait lens.

Recommendation for a good portrait view is that with a 1.6X FoV camera, you have a telephoto lens with at least 100mm, which would yield 160mm equivalent focal length. Could this already be in the zoom range of the kit lens you have now? My stronger recommendation for a good portrait lens (as distinct from a view suitable for portaits) is to use a prime lens that is long enough. For your camera, a 100mm prime lens would work well, as would a135mm.

Studio lighting is a completely separate matter from low-light photography, you might consider looking at some of the past dialog on a lighting tribe many of us here also habituate (or ask more detailed questions):
tribes.tribe.net/strobelighting

One quick little studio lighting tip I'll offer is that the softest light from a softbox is immediately in front of the softbox. You can see this directly as you create shadows with one hand blocking the light to another ... and then slowly moving away from the softbox. Instantly, shadows will appear, and they will become more and more distinct the farther away you get from the softbox. So a good little trick is to place the softbox near your model, but just fractions of an inch outside the view of your camera. That will be the softest light possible, with that setup.

But low-light also does come up in the studio context. For example, you might find yourself wishing to photograph a scene with candles or another dim source for main illumination. In that case, all the forgoing comments on this thread apply no less. If you have a stationary model, and you are using a Canon zoom lens with internal Image Stabilization (IS), you can get incredible candle-light photos handheld at very slow shutter speeds.

Stef ... hoping these notes will add to your pool of know-how, and build on what the other talented photographers offered here.

Blessings and Light,
M