New Canon 5D owner - disappointing night photography results?

[anonymous]

I've been a night photographer for a few years now and have been shooting only E-6 film and doing scans on an imacon scanner. But recently, the last lab here where i live closed (i live in las vegas) so I started looking for a digital camera that I could shoot for low light work. I ran across this website and liked the images, so after reading your info on your equipment (canon 5D) I purchase one. But after doing a number of night shot, I've been really disappointed in the printing results.

I mostly shoot at 100 ISO at F-22 for about 10-30 seconds, but i'm getting some grain that shows on my prints. (I have an epson 7800) My output print size is 20x16. I've been trying to read more about the 5D, but I'm wondering if this is the best the camera can do. If anyone has any info that could help me get better image output, I would be glad to hear from you.

[McQ]

What is the weather like where you are shooting? The noise levels in long exposures are highly sensitive to temperature.

Since the exposures are only 10-30 seconds at ISO 100 (does not seem like full night conditions to me), you can always take two shots (with the camera in an identical position) and average them. This can help quite a bit. Just make sure the tripod is on a really good surface and that you use a cable/remote release. The technique is described more on this website's tutorial about image averaging noise.

Another consideration is with the nature of the noise (ie, is it random, banding or fixed pattern noise). Banding and fixed pattern are much more problematic. Averaging only really works with random noise, dark frame subtraction works with fixed pattern noise and banding noise is more or less uncontrollable unless you intentionally overexpose (or use a different camera).

Also, are you shooting in RAW? (you should be if not). Try overexposing the image (without blowing highlights unless these are streetlights) and then add a negative exposure compensation upon RAW conversion. This will help a ton with the shadow noise (esp if this is banding noise).

Hope these help get you started.

[Poljazz]

I think that f22 seems to be a bit too small an aperture to use because of diffration. You may also be underexosing a lot which will increase noise a lot.

[kevinf]

The 5D has fantastic low-light performance. One of the best digital cameras out there for that particular use. I would try bumping your ISO up to at least 400 and halve your aperture and halve your shutter speed and your results should improve with lower noise and higher sharpness.

[Keith Reeder]

I agree absolutely with Kevin about the 5D - it should be able to produce spectacular images in these circumstances.

Poljazz makes a very good point too about f/22 being too small an aperture because of diffraction.

Increasing ISO and/or opening the aperture (thereby reducing exposure time) will help in another way - long exposures generate heat in the sensor, which means additional noise.

The 5D has in-camera long exposure NR - you might wish to activate it.

[_GUI_]

10-30 seconds shots should not be a problem in noise because of heating. I would consider the fact that you could be not reaching an adequate degree of exposure in your RAW files. A fairly high enough degree of exposure is even more important than ISO for noise reduction.

If your night scenes have such a high DR that you cannot afford more exposure, I would take 2 shots, one correct and one overexposed (2 or 3 f-stops overexposed) and blend them so that the second shot can enrich your shadows reducing noise in them.

I would also consider to open a bit your aperture. f22 seems too exagerated (although it is true I know some arquitecture photographers that use it with the 5D, despite the danger of diffraction).

BR

[Keith Reeder]

Originally Posted by _GUI_

10-30 seconds shots should not be a problem in noise because of heating.

The 40D manual (which I admit might not be the same as the 5D in this regard) suggests that Long Exposure NR is useful for exposures of one second or more...

[_GUI_]

Originally Posted by Keith Reeder

The 40D manual (which I admit might not be the same as the 5D in this regard) suggests that Long Exposure NR is useful for exposures of one second or more...

That refers to black frame substraction, and yes, can be useful at any exposure time, but this does not mean the improvement will be visible. In my 350D it activates at 25 or 30s I think, so I consider the 1s a bit exagerated.
Long exposure times where heating noise is a real problem involve exposures of several minutes, and astrophotographers know very well that on those conditions noise is a real problem.

[Keith Reeder]

Originally Posted by _GUI_

That refers to black frame substraction, and yes, can be useful at any exposure time, but this does not mean the improvement will be visible. In my 350D it activates at 25 or 30s I think, so I consider the 1s a bit exagerated.
Long exposure times where heating noise is a real problem involve exposures of several minutes, and astrophotographers know very well that on those conditions noise is a real problem.

I see - thanks for the clarification.

It's not a style of photography I do, and I was just working from the manual.

[Antonio Correia]

Originally Posted by McQ

What is the weather like where you are shooting? The noise levels in long exposures are highly sensitive to temperature....

McQ, do you think that switching off the camera for some minute/s between shots would produce less noise ?

This procedure would cool the sensor decreasing noise ...

[McQ]

Originally Posted by Antonio Correia

McQ, do you think that switching off the camera for some minute/s between shots would produce less noise ?

This procedure would cool the sensor decreasing noise ...

In newer cameras, such as the Canon 20D and onwards, these have several features which actually disable some of the electronics to minimize sensor heat and conserve the battery during a long exposure. This helps a lot, and is a big reason why the 20D improved on the 10D by as much as it did.

On the other hand, your question relates to decreasing the sensor temperature *between* exposures, with the assumption that the sensor temperature will not get as high during the exposure as a result. It really depends on how long it takes for the sensor temperature to plateau once the exposure begins, which is a function of many interrelated factors. If it takes a long time, then sure, turning off the camera may help, but likely only in older cameras which did not have all the most modern heat reducing features. Even then, there would likely only be a particular range of exposure times where it would be noticeable.

Strictly speaking, if turning off the camera decreases sensor temperature a tad, and sensor temperature correlates with image noise, then yes, turning off the camera will reduce image noise. However, I doubt this is noticeable and even if it is, I would question whether the extra trouble warrants the marginal improvement, if any. It would be neat for someone to give it a try though.

[Antonio Correia]

Originally Posted by McQ

...It would be neat for someone to give it a try though.

I don't want to promice in order not to break the promisse but I will try to do the experiment.
Care to make a sugestion in terms of exposure and gaps between shots ?

Using the 20D at night with CPF and f/5.6
A - One 5 minutes shot with ISO 1600
As soon as the camera writes in the card re-shoot the same picture with the same settings.

B - One 5 minutes shot with ISO 1600
As soon as the camera writes in the card, switch off the camera for 5 minutes
Re-shoot the same picture with the same settings.

Would it be OK ?

[McQ]

Well, the real unknown here is how long it takes the sensor temperature to plateau. If the plateau time is short, then having the second exposure be a long exposure will not show any difference between (A) and (B) above. You might want to try a relatively short exposure to magnify the effect.

[Antonio Correia]

Originally Posted by McQ

Well, the real unknown here is how long it takes the sensor temperature to plateau. If the plateau time is short, then having the second exposure be a long exposure will not show any difference between (A) and (B) above. You might want to try a relatively short exposure to magnify the effect.

Will you please be a little more specif ?

Thank you

[McQ]

A - One 5 minutes shot with ISO 1600
As soon as the camera writes in the card re-shoot the same picture with the same settings.

B - One 5 minutes shot with ISO 1600
As soon as the camera writes in the card, switch off the camera for 5 minutes
Re-shoot the same picture with the same settings.

The presumed benefit of turning off the camera/sensor prior to starting the exposure is that it will mean that this exposure will start at a lower temperature than it would otherwise. However, if the camera sensor heats up really fast and reaches it's (higher) equilibrium temperature with the surrounding air in a short amount of time (let's say 20 seconds, the "plateau time"), then the majority of your proposed 5 minute exposure will be spent with a sensor temperature that is no different than if you had not turned the camera off in the first place. This would make it much more difficult to distinguish any difference between your proposed (A) and (B) tests.

Note that I am talking about the second exposure, not the first (which should still be long...5 minutes is probably plenty). You would presumably be throwing out the first exposure in both (A) and (B) as these would be identical...

[WKHC]

Originally Posted by Keith Reeder

The 5D has in-camera long exposure NR - you might wish to activate it.

I always set the custom function for Long Exposure NR to Automatic but I am not sure when it kicks in.