Pages: [1]   Go Down

Author Topic: Long exposures with canon DSLR's  (Read 3488 times)

Khurram

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 488
Long exposures with canon DSLR's
« on: September 19, 2006, 12:19:12 pm »

I wanted to get some advice for shooting long exposures digitally with Canon DSLR's, as I'm mainly concerned about digital noise.  

I have a 1Dmk IIN and just picked up the rebel xti.  I'm going to be shooting at the Antelope canyons in Page over the next few days and wanted to know if the camera's noise reduction should be turned on, as my exposures will probably be between 30 sec to 2 min.  Is it better to use the canon in camera noise reduction or to deal with any noise using other software afterwards.  Is the settings on DPP up to the task??

I dont' have any experience using any noise reduction software or settings, so any advice would be helpful.
Logged
----
[url=http://www.pbase.com/kssphotog

pobrien3

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 320
Long exposures with canon DSLR's
« Reply #1 on: September 19, 2006, 11:55:02 pm »

If you're shooting long exposures anyway, then keep the ISO as low as possible which will minimise noise.  Shoot at ISO100 and turn the camera's noise reduction ON.  Your alternative is to shoot your own dark frame and subtract it later in Photoshop.  I'd let the camera do it.  The camera will effectively take a second 'dark' exposure after the shutter has closed so you won't see your histogram until it's done that (takes as long as the first exposure), but you can use it for another shot while you're waiting.

For long exposures I set to 'bulb', shoot at ISO100, NR turned on, dark slide over the eyepice, mirror locked up (not critical for longer exposures), using the cable release.  On a tripod, of course.  Process RAW files in ACR, apply noise reduction in Photoshop using Noise Ninja.  If there are any points of light in the scene (distant buildings, etc.) I'll often bracket as it's hard to predict the effect of these in the final image, aso I may end up blending several exposures.

Experiment; go outside tonight and shoot a 2 min exposure with NR on, and one with it off, and see if you feel the noise is acceptable in either method.  DPP won't really help you remove the noise you get with NR off, and Noise Ninja etc will soften it too much.
Logged

Phuong

  • Full Member
  • ***
  • Offline Offline
  • Posts: 113
Long exposures with canon DSLR's
« Reply #2 on: September 20, 2006, 12:51:38 am »

with film, opening too long causes reciprocity failure
with digital, opening too long, the sensor will be hotter, there's a possibility that some pixels will get too hot (called hot pixel). these pixels will read (255,255,255) and there's no way to recover them in PS. doing dark frame subtraction is basically recording a map of them, therefore easily recover them (in camera)

so, the answer is yes, yes, yes
Logged

John Sheehy

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 838
Long exposures with canon DSLR's
« Reply #3 on: September 20, 2006, 08:29:51 am »

Quote
with film, opening too long causes reciprocity failure
with digital, opening too long, the sensor will be hotter, there's a possibility that some pixels will get too hot (called hot pixel). these pixels will read (255,255,255) and there's no way to recover them in PS.

"Hot pixels" is a term usually used for pixels that are bright immediately after being reset, I believe.  The effect that occurs with long exposures is gradual, as analog pixel intensity grows over time.  I like to distinguish them by calling them warm pixels.  Some grow faster than others, and you will see them faster at higher ISOs.  They don't have to be anywhere near 255 to stand out, especially in a dark image.

You can filter them out in software, rather than trying to recover them.  Running a Median filter, and then fading the result to taste can work.  Converting from RAW in recent ACR versions eliminates the most obvious ones.  This is best done in RAW, really, because the warm pixels are only one pixel in size in the RAW data; in the conversion their influence would be spread over about 9 pixels, and then they may be more likely to be indistinguishable from real spotted detail in the subject.
Logged
Pages: [1]   Go Up