Nikon D750 'Service Advisory' - Long Exposure Noise Reduction

May 23, 2018  •  7 Comments

Recently, I discovered a weird 'glitch' caused by Long Exposure Noise Reduction on the Nikon D750.  Note, this only applies to astrophotographers and night photographers taking images with stars in them.  If your image does not have stars, you shouldn't notice any problems.

When LENR is enabled, the processing causes stars to lose their true colors and turn green, yellow, cyan, and purple.  After doing numerous tests, I've confirmed this problem is directly related to the LENR processing.  The comparison below should give you a good look at the problem.  

 

LENR OFF
LENR ON

 

 

What is Long Exposure Noise Reduction?


When LENR is turned on, your camera will take 2 exposures.  The first is a normal photo, the second is a dark frame.  This dark frame is mainly used to find Hot Pixels.  Once the camera has both images, it will automatically remove the Hot Pixels and save one final RAW or JPEG image.  You can try this process manually.  For example, take a normal photo.  Immediately after that finishes, put your lens cap on and take another image with the same settings.  This will be your dark frame.  Using various Astro photo stacking programs, or just Photoshop, you can subtract the Hot Pixels from your normal image.  However, this might not always work well.  Therefore, I like using LENR in-camera.

 

 

 

Temporary Fix


As of now, there are 2 ways to fix this problem in Post-Processing.  The first requires Nikon's Capture NX-D software.  Once you open your RAW image, find the Noise Reduction panel on the right-side.  You should see an Astro Noise Reduction checkbox.  Once clicked, it should automatically remove any hot pixels and fix the color shift issue reasonably well.

Of course, using Capture NX-D isn't part of most photographers' workflow.  Personally, I use Adobe Camera RAW to initially edit my RAW photos.  By increasing the Color noise reduction slider, and decreasing the Color Detail slider, the color shift is drastically reduced.  However, this may negatively impact real colors in the image. 

 

 

Latest Updates


After speaking with the Nikon Reps in 2018, I got a final answer from the Nikon techs.  "...the result is that the original images exhibited no color aliasing, therefore, it is assumed that using Photoshop had some impact that resulted in the color changes seen after processing."

In other words, Nikon is saying that the problem comes with the RAW processor's debayering process.  However, I tested multiple RAW processors (Nikon Capture NX-D, Camera RAW, RAW Therapee).  They all exhibited the same problem, with varying degrees of intensity depending on which debayering process was used.  This problem was only visible in images taken with LENR ON.  Therefore, I would argue the problem is still related to the camera itself, not the RAW processor you use.  Unfortunately, it seems Nikon has no interest in fixing this problem.

If you have a Nikon camera with the ability to enable LENR, do you notice this problem?  In order to test, take an image of the night sky with LENR turned off.  Then take another image, same settings, with LENR turned On.  I noticed this issue mainly on my wider angle lenses, like the Sigma 35mm and Nikon 14-24mm.  If you have a star tracker, try using that to take a longer exposure, which might help exaggerate this issue.

If you find this same problem on another Nikon camera model, please let me know!  Either mention it in the comments, or send an email to [email protected] with sample images.

 


Comments


Archive
January February March April May June July August (1) September October (1) November December
January (1) February (1) March April (2) May (1) June (1) July August (1) September October November December
January February March April (1) May June July August (1) September October (1) November (2) December (1)
January February (2) March April (1) May June July August September (1) October November December
January February March (1) April May June (2) July August September October November December
January (1) February March April May June July August September October November December