Motivation is a photography blog that discusses the creative aspects of photography. The posts will include thoughts about images and their interpretation, photographers and their work, technique, workflow, my ongoing projects, and perhaps even the occasional off topic rant.

Topaz DeNoise 6 And Canon 5DsR Noise

Back in October I wrote about a five minute exposure I had made at Pete's Lake in Michigan's Upper Peninsula using my Canon 5DsR camera. That exposure was quite noisy because of its duration and if I wanted a usable image I had to leave the island quite dark, almost black, in order to mask the strong noise. I didn't really like the way it looked and thought the image might not be able to be saved.

Recently, Topaz Labs came out with a new version of their DeNoise product/plug-in. I already own a couple of noise removal programs, including the prior Topaz version, but decided to upgrade to their newest.  This version is a whole different ball game.

There is an incredible amount of control available within the program.  An excellent 40 minute video is all you really need to get a good grasp on how it works. So I watched it and gave DeNoise 6 a try with this immensely challenging image after I wasn't happy with the results that other programs gave me (see that original post).

The ground rules:

  • I duplicated and reprocessed the image from the same starting point but did noise reduction first and then processed it the way I wanted it to look in terms of the tonality of the trees and island.
  • I took the original, noisy image from that post back in October and, without applying noise removal, used a curve and a bit of saturation, to bring it to just about where the newly processed version was in terms of appearance.
  • The crops below are 100% but converted to 8 bit jpeg and sRGB color space; I didn't notice any dramatic changes resulting from this. 
  • Disclaimer: there were residual white specks following the DeNoise process that were present on the original (ie, they were not artifacts induced by DeNoise). I would not call the processing a success if these remained or if I had to work for hours on manual removal. I tried the Photoshop dusk and speck removal filter and it took care of nearly all these specks in a second.

So here is the newly processed image:


Pete's Lake, 5 Minute Exposure    © Howard Grill

Directly below is a 100% crop, as described above, from the upper left hand corner of the image without Topaz DeNoise processing. 

Upper Left Corner, Before Topaz DeNoise 6

Here is the same crop after noise removal. I was pleased that you could see the gentle tones of the cloud streaking. Removal of the noise actually made it more evident.

Upper Left Corner, After Topaz DeNoise 6

Below is a 100% crop from the water.  It's even worse than the upper left corner in terms of noise. NASTY!!!

Section Of Water, Before Topaz DeNoise 6

And below is the DeNoise 6 version of the same location in the image:

Section Of Water, After Topaz DeNoise 6

Finally, we come to the very distant trees in the shadows, which I significantly lightened compared to how the photograph came out of the camera. Note that at baseline there is some lack of sharpness, but this is explained by the fact that the trees are probably a half mile away and that the exposure is five minutes long with a gentle breeze blowing those branches.

Here is the original version:

Trees On Distant Isalnd, Before Topaz DeNoise 6

And the DeNoise version (after Photoshop dust and scratch removal was also used; see above).  It appears to me that there is perhaps a very, very slight softening of the image in comparison, which is pretty impressive for this degree of noise removal. In my opinion it has taken the image to a state where it is quite usable. Given that this is a 50 MP file the size of those trees on even a large print is going to be small.

Trees On Distant Island, After Topaz DeNoise 6

Definitely a worthwhile purchase in my opinion!

Canon 5DsR And Long Exposure Noise

In my last post, I presented one of my favorite images from my trip to Michigan's Upper Peninsula.  As much as I really like the image there is a problem with it.  It is a problem that is not visible to anyone looking at the image on the internet or at a small image size, but it is one that I suspect will limit my ability to make a very large print of the photograph as it is currently processed, which is something I was hoping to be able to do.

So what is this problem?  The answer is digital noise during the five minute long exposure. I will confess to being a bit of a 'pixel peeper', but really only am concerned about issues that would affect the final print. But before I discuss anything, allow me to show you what I mean.

Color and Luminance Noise In The Brightened Shadows

Color and Luminance Noise In The Brightened Shadows

The portion of the image above is a 100% crop from the dark island/tree area of the photo AFTER noise reduction using Imagenomic Noiseware, and one can still see both color and luminance noise that is fairly marked. Now, it is true that I raised (brightened) the shadows a bit, which would tend to accentuate noise, so I will also show you a 100% crop from the straight out of camera image without brightening the shadows. Here the noise is minimal.

Pete's Lake Crop Right Out Of Camera

Pete's Lake Crop Right Out Of Camera

Perhaps I was expecting too much to be able to brighten the trees in order to to show some detail and color and perhaps the noise is purely related to the 5DsR's pixel pitch (small) and dynamic range as opposed to the length of the exposure.  That will take some experimentation on my part to figure out but, while that may well be the case, there is still luminance noise in the sky visible in this 100% crop after noise reduction:

Luminance Noise In Processed Sky

Luminance Noise In Processed Sky

As one would expect, this brighter area is certainly less noisy than the shadows, but is still noisier than I would expect and noisier than I have seen in my shorter duration exposures.

So what does all this mean?  These are just early thoughts and so I am not really sure yet.  It will require some further experimentation to sort out the effects of long exposure vs the sensor's dynamic range.  I find this interesting, as it is something fellow photographer Cole Thompson has written about in a post entitled "5DSr Noise Issues".  

Of course, this doesn't necessarily mean I can't make a large print. The noise might not be visible in a moderate sized print. I could make the print with less processing and leave the shadows darker than I had in my processed version, which would hide the noise.  There are possibilities, those they are slightly different than was my vision for the final image.  Also, when doing long exposures I could expose to the right more in order to eliminate noise, assuming that the scene is not as contrasty as this.

Please don't get the wrong idea.  I am extraordinarily happy with my 5DsR as it is able to record amazing detail and color with ample pixels to crop as needed, but I suppose one has to obey the laws of physics and smaller pixels are going to deliver more noise.  Unless of course you have a Nikon. Or a Sony........only kidding. Sort of :)