Noise Reduction Tools: A Review
But noise? HDR at ISO 1600 is usually trash unless you own a seriously good camera. I have 4 noise reduction tools. Here's what I got:
Topaz Labs DeJPEG
This works really good with highly detailed shots such an HDR image of a rock texture or forest floor...anything that doesn't have plain areas or gradients. You get excessive banding and artifacting in gradients if you apply a severe noise reduction. Take care, the software boosts saturation a bit too much by default. For me, this software is barely useful. This works really nicely for casual purposes like saving your noise-filled mobile phone shots. But not useful for commercial photography because it will give you rejections if you pull it to the limits. Generally, this software isn't worth going for at all (for me).
A surprising piece of software. It literally deletes the noise and keeps the details intact. I've never seen anything as flexible and efficient as this. I use this most of the time without worrying about details at all. But this suffers from the same problem as the Topaz Labs DeJPEG. If you apply excessive noise reduction for HDR images, you'll get huge artifacting in plain or gradient areas that wouldn't go even with a blur of 5 px. Why? Because this software has an extraordinarily high sensitivity for edge detection. It picks up the CMOS leakage current between pixel blocks and keeps them intact. Since the leakage current across blocks increases by over 40 times at ISO 1600 than in ISO 100, Imagenomic thinks it is "details" and leaves it, leaving the image full of artifacts.
But for general noise reduction (upto ISO 400 or 800 on compact cameras and ISO 1600 on DSLRs), it is extremely efficient and useful.
My favorite for severe noise reduction. It removes fine details from all plain areas but leaves no sign of noise even at 400%, which gives your HDR images a flat, painting like look. For natural looks you can set the reduction levels to less severe and you will get good enough photos with almost no noticeable noise. The softness of image can be a problem but the edges are sharp, it is just the plain areas that look smooth or "washed out". But it doesn't make it feel too bad. Better than the other software, that's for sure. The software has an undo brush and some channel viewers which I find very useful for deciding whether a channel selective noise reduction would help. This is definitely the winner with high ISO shots.
Canon Digital Photo Professional
The BEST for general shots. A low noise reduction would get you a clean and great photo with no noticeable problems at all. You cannot even see this software at work, it works so cleanly. But then, when you make the settings severe for extremely high ISO shots, you start to get problems. Especially at high luminance noise reduction settings, it blurs out the photo a bit too much. The blur radius exceeds the sharpen radius settings and the sharpening ceases to have any noticeable effect. I don't particularly like blurry edges. So I vote for Noise Ninja for high ISO cases.
General use of DPP is good for me when shooting in RAW. For severe noise I use Noise Ninja (on 16 bit TIFFs) on which I have noise profiles installed for all my cameras. The auto profiling tool is very useful too at times. For JPEGs coming out of compact cameras or DSLRs, using Imagenomic is a good idea (for moderate noise situations). For high noise level JPEGs, Noise Ninja is better.
I hope this article helps people stuck and frustrated with noise problems. The above software saves most photos from being resized due to excess noise. Apologies to the Nikon users. I have never used a Nikon camera so far and have no idea about the noise performance or the RAW processing tools you get with your Nikon.
The views in this article are from my experience with both compact cameras and DSLRs (Canon). It is not supposed to criticize any software. It is just something to help you get results without trial and error and the resulting frustration.
Related image searches
Review related image searches
This article has been read 2113 times. Photo credits: Pratik Panda.