Shooting Indoors with poor available light

[dcheung]

What lenses do you guys generally use for this kind of application?

 

For a zoom 2.8 is the fastest ur gonna get, but most of the time it's

way way way too slow for me. eg, poorly lit gyms shooting sports,

shooting dinner parties at fancy resturant, shooting house parties,

shooting dances at ball rooms.

 

That's why I'm gonna start building up my fast prime collection.

Going to NYC in 2 weeks. Gonna pick up the 85 1.8 for sure (for

indoor volleyball on a 20D), probably pickup a used 50 1.4 as well.

 

What i'm afarid though is that even these fast apertures won't be fast

enough for most indoor environments such has people's houses or

classrooms ... etc.

 

The question is, for you available light indoor shooters, do you find

a 1.4 or 1.8 aperture to be fast enough for your uses? For example, I

feel even the 85 1.2 will not be fast enough for a ball room. Those

places are usually pretty dark. Given that, I will probably like to

beable to stop down by 1 or 2/3 of a stop to increase sharpness, ....

what am I left to do? use ISO3200? I don't like to tolerate

anything over ISO800 (I have a 20D). I use noise reduction software

for ISO400 and above already.

[dcheung]

maybe this is obvious, but i should point it out anyways:

 

I'm trying not to use flash, if anything, i'll have the flash output be less than the available light, so use it as fill in i guess. (FEC -1.3 to -2 in Tv or Av mode)

[awindsor]

Give up on photography for now and wait for future technology or change your expectations.

 

Shoot wide open, use ISO 3200, and deliberately underexpose if necessary to get a fast enough shutter speed. Use noise reduction software to remove the chroma noise or convert to B&W. Leave most of the luminance noise alone. If the subject is in motion you will get motion blur (you will be lucky to get handholdable shutter speeds without trying to get shutter speeds that will freeze motion).

 

If you can use flash you should. You still need a fast lens and maybe higher ISO to get a reasonable ambient exposure but then the flash will expose the main subject. Gel the flash to balance for tungsten lighting (no need to balance perfectly, having a warm background is fine). Bounce or use a diffuser. Use a flash bracket to ameliorate red-eye.

 

Oh and be very very thankful you weren't photographing in the age of film.

[dcheung]

to be honest, I wouldn't be too surprised if in 2.5 years, ISO 3200 noise levels are like that of ISO 800 on a 20D.

 

Semiconductor technology in this field is proliferating extremely fast.

[dcheung]

In the future, low light photography will be improved not because of better/faster lenses, but because of much better sensors. Think about how much sensors have improve from say the d60 to the 20D,.... or if you look back even further. There really hasn't been that many years. Now, we are on the onset of digital point and shoots with low noise even at ISO1600. I'm hoping the new elan digital (30D) will have crazy low noise profiles when it comes out.

[panos_voudouris]

If your problem is motion blur in volleyball there is only that much you can do without

flash.

 

But...unless you are visiting Dracula's dungeon, people's houses will work

fine with a 28/1.8. I get 1/125@f/2 in my well lit kitchen and 1/45@f/2 in my less than

well lit living room at ISO 640. 1/45 is more than enough for handholding it and if you get

a monopod you'll be able to go down to 1/15.

[bennyboy]

50 f/1.4 works well indoors for me, and that's without having to use a ludicrous ISO - but I'm using fairly shallow DOF.

[bens]

you can get decent results for indoor sports with the 85 and 50 -- using iso 400-1600 i have regularly shot at 1/200-1/400 shutter speeds without flash, depending on conditions. couple of basketball shots in my candid portraits portfolio taken with the 85 if you are interested.

[mark u]

Here's a small exercise in the physics of light:

 

The full well capacity of a typical DSLR pixel is of the order of 64K (or 2 ^16) electrons (the precise number doesn't matter to the analysis greatly, but it keeps the math simple). That corresponds to a saturated pixel with an exposure of 1/100th at f/16 and ISO 100 (sunny 16 rule for EV 15 basis 100 ISO). The overall efficiency of capture of photons that make it through the lens to the sensor to be converted into electrons is in the order of 30%. For available darkness - say candlelight it's about 11-12 stops darker, so if you open up the lens 6 stops to f/2 you are still capturing only 1/64th (2^-6) of the light of a sunny 16 exposure at 1/100th for a bright white subject. That's just 1K (2^10) electrons. Now light photons obey Poisson statistics in terms of their arrival time, so the standard deviation of the number of electrons generated in each pixel is the square root of the average number - or about 32, some 3% of the signal. For a mid tone, you are capturing 1/4-1/8th as many photons, and the photon shot noise becomes even more prominent. Now how about those shadows?

 

Of course, you could lengthen the exposure to capture more photons, or you could make the pixels bigger at the expense of resolution (astronomers do both with their CCDs, and high speed films effectively behave as if they had large pixels). Maybe you could improve the overall efficiency of capture, but that is only going to gain about 1 stop.

 

Note that the only attributes of the sensor we have considered in this discussion is pixel size and capture efficiency. Camera generated noise hasn't been added to the mix at all. Despite that, you are going to end up with some very noisy images. Remember that all using a high ISO does is to amplify whatever signal the pixel has captured, which will amplify the photon shot noise too. There's no way round the quantum physics of photon shot noise, so I think you hopes that sensors will magically make that possible in the future are a little optimistic.

[pbizarro]

Try a simple thing:

 

1. Go to one of the places where you typically have to shoot under low light. Take your f/2.8 zoom;

 

2. Meter the light at f/2.8 at several ISO's, make note on the different shutter speeds;

 

3. Convert previous shutter speeds to the corresponding shutter speeds for f/1.4 and f/1.8 apertures.

 

That will answer your question.

[dcheung]

^ I did that, and it wasn't good enough. That's why I'm asking people who do indoor available light photography as to what they do in situations like this. OR is there simply no solution due to technological drawbacks.

[danny_lee2]

So you are asking how to do photography in the dark. this is photography. Photo=light graphy=write.

 

how you gonna do photography with no light ?

 

what you are asking is just plain silly. its either high iso, or long shutter speed, just pick one.

[john_stark]

I shoot in some very dark gyms all the time using a 1D MKII N or old 1D and never shoot higher ISO than 1600 with Canon 70-200mm USM IS f2.8 lens.

You can normally pick up one or two stops by getting above the the action and shot down getting the light reflecting off the floor. The IS really helps though, and even more at floor level. Some motion blur won't hurt the photos, and in most cases helps give a sense of action that completly freezing the action loses.

Knowing PS really helps when doing post processing.

JS<div></div>

[panos_voudouris]

" I did that, and it wasn't good enough."

 

So what shutter speeds do you get at what apertures? 1/2 at f/2.8? That is still 1/8 at f/

1.4 and a monopod will be sufficient to give you just about enough stability for a 50mm

lens. Again, there is a

minimum of light you need for taking a photo. Where exactly are you shooting? What are

the conditions? f/1.4 is more than enough for indoor low lighting at ISO 800/1600.

 

If you cannot get a suitable shutter speed at f/1.4, chances are there is not enough light

for AF to work and it is too dark to focus manually as well. So either use a flash or sit back

and enjoy the conversation/game/whatever.

[mark u]

Simple summary of my previous post in this thread:

 

You can't beat the laws of physics.

[lester_wareham]

to be honest, I wouldn't be too surprised if in 2.5 years, ISO 3200 noise levels are like that of ISO 800 on a 20D.

 

I don't know how true that is. I understand the current sensors are quite efficient in quantum terms. ie photon counting noise is already significant.

 

Check out some of Roger Clark's stuff. http://clarkvision.com/imagedetail/index.html

[awindsor]

There is low light and LOW light. Most interior lighting is fine for our eyes but quite low for photography. A fast lens and a moderate ISO is normally sufficient. I like to photograph musicians in jazz clubs. Here the lighting is poor for your eyes and very difficult for film photography. Digital photography is much more practical since I can operate at the equivalent of ISO 6400 (actually it is ISO 8000 since Canon's ratings are off) and get very usable images.

[eric merrill]

50/1.4 for low light. Neopan 1600 at 1600.

 

If there's really no light at all, the Canon stays in the bag and the Leica gets all the use.

[glogower]

3 tips for shooting in this environment...1) use a tripod. 2) use a tripod. 3) use a tripod.

[glogower]

Taken in low light with a tripod...

[dcheung]

John Stark >> I love how u can see the meat on her arm compress on that volleyball shot. I'm surprised how target motion blurr free that image is considering you are shooting at 1/250th. I try to kick it up to 1/1000th because I like to get the spikes frozen. Maybe it's cuz that shot is of female volleyball and the ball speed is not as fast as the males. I was shooting male volleyball for my university (university of alberta, and our team beat top all the top rated NCAA teams we've invited and we're number 1 in canada). The ball speeds in our games are alot faster. 1/1000 is prefered and that's when i was shooting at ISO3200 and was about 1.3 stops underexposed. That was with F4.0 so with an 85 1.8, I should beable to shoot in that gym no problem. The real problem comes in people's house parties. People are always moving around so I'm desiring shutter speeds of 1/250th or 1/320th of a second....but poorly lite environments don't allow that even at ISO3200. A lot of the times the problem arises at home where i'm just playing with my gear in my poorly lite living room during TV commercials. In that room, at F2.8 and ISO3200, i need shutter speeds of like 1/15th or longer for proper exposure (depending on target) That's fine if the targets aren't moving, but if the target is walking around, it's not good enough.... and that's one of the main reasons I made this forum thread.

[panos_voudouris]

"I'm desiring shutter speeds of 1/250th or 1/320th of a second....but poorly lite environments don't allow that even at ISO3200. A lot of the times the problem arises at home where i'm just playing with my gear in my poorly lite living room during TV commercials. In that room, at F2.8 and ISO3200, i need shutter speeds of like 1/15th or longer for proper exposure (depending on target) "

 

On parties it is usually very dark for any camera/lens/ISO/whatever. In that case you need to get a flash and learn how to bounce/diffuse it so that it looks good. Also, if you cannot freeze motion, use it to your advantage.

[john_stark]

Glad you like shot, however these girls are no slouches and the speed is pretty darn close to what you get at college level (in fact these kids will be in college next year) I have stopped high speed action, but use strobes to do it. You could mount strobes up high and it won't affect the players. It will cut your ISO and noise WAY down too!

 

As for your parties, go with a flash and bounce it as said above. Manual focus if needed.

 

I am waiting for Canon to ship the new "S" focus screen for low light...that will be a big help with us 1D MKII N users who shoot low light a lot.

 

JS

[dcheung]

Mark U >>

 

What makes you believe that the sensor noise levels we are working with is close to the possion distribution of the quantized emission of photons? If that was the situation, then there would be no point in cooling down the sensors for long exposures. Lost of people do this for astrophotography. The point of cooling it is to reduce the thermal energy that causes electrons to beable to jump into the conduction band. We want only photon energy to kick these electrons into the conduction band. To improve sensors, we can make ones that are not as suseptable to thermal noise. I do agree that quantized emission of photons do contribute to some of the noise....and perhaps it even contributes to a significant amount of noise at ISO100. We'll give u the benefit of the doubt and say that quantized emission of photons cause 100% of the noise at ISO100 (which is not true at all). Given that, I commented on improving sensor noise at high ISO. CMOS sensors have the amplifiers built right into the semiconductor sensor itself. What we can do to improve high ISO noise is to reduce the noise introduced by the amplification stage. Yes at ISO 800 we will simply amplify the quantum shot noise, however, that's not all we see. The amount of noise at ISO3200 is more than double the amount we see at ISO1600. This means that it's the amplification stage that introduced more noise. Noise increase is not a linear function to ISO setting. In another words, even if we are at the quantum shot noise limit at ISO100 (which we are not), we still have much room for improvement for high ISO settings.

[mark u]

You appear to have completely failed to understand my post. I am starting from the assumption that you have a sensor with virtually zero noise (i.e. with the technological improvements you envisage). My point are simply that current sensor technology isn't so far off that ideal if you are trying to use shortish shutter speeds. Given a level of lighting and subject, a particular lens aperture will focus only so many photons per square micron per millisecond onto the sensor. Photon shot noise is a function of that number, and INCREASES as a proportion of the signal the lower the level of light (you seem to think that noise increases with higher levels of illumination, which is simply not true). My calculations are based on estimating the number of photons per pixel that would be captured assuming candlelight and an aperture of f/2 with a shutter speed of 1/100th. Those numbers in no way depend on the ISO setting of the camera. The only place that comes into play is the degree of amplification of the electrons in the pixel.