Topaz Denoise AI mini review/test
Posted: Mon Feb 15, 2021 9:46 am
When I first became aware of Topaz Denoise a couple of years ago, I downloaded a trial and tried it. I could not see much difference to using the built in noise reduction in Adobe Lightroom. That was then. This week I downloaded and tested a trial copy of the latest version Topaz Denoise AI.
Here are the results.
When taking subs with my Pentax K1, I often wind the ISO up high and take a quick test image at ISO12800 30s just to check the overall appearance and image density/fog. I then wind down to ISO1600 and set the sub-exposure time to 4 minutes. It's not just noise but more importantly the dynamic range. Dynamic range decreases almost 1eV with each 1 stop increase in ISO.
I have taken one of these 30 second ISO 12800 subs to test Topaz.
This is a full sized original. It shows the segment has been cropped to display the results at the 100% pixel for pixel level below. Many noise reduction programs distort and blur the original image leaving aliasing to disguise the noise. The latest Topaz AI does a much better job.
This is the noise reduced image below. I have overlaid part of the unprocessed image inside the box for direct comparison. There is a mottling of faint nebulosity which is a bit exaggerated over the mottling visible in the ISO1600 x 4 minute 20 sub averaged image I produced the same night.
I don't see this doing away with lower ISO multi-sub deep sky imaging and replacing it with high ISO capture. But I can see applications in some shooting situations where multiple subs may not be possible:- aurorae, fast moving comets, some nightscape photography.
I'll post the unadulterated final image in the deep sky section.
Joe Cali
Here are the results.
When taking subs with my Pentax K1, I often wind the ISO up high and take a quick test image at ISO12800 30s just to check the overall appearance and image density/fog. I then wind down to ISO1600 and set the sub-exposure time to 4 minutes. It's not just noise but more importantly the dynamic range. Dynamic range decreases almost 1eV with each 1 stop increase in ISO.
I have taken one of these 30 second ISO 12800 subs to test Topaz.
This is a full sized original. It shows the segment has been cropped to display the results at the 100% pixel for pixel level below. Many noise reduction programs distort and blur the original image leaving aliasing to disguise the noise. The latest Topaz AI does a much better job.
This is the noise reduced image below. I have overlaid part of the unprocessed image inside the box for direct comparison. There is a mottling of faint nebulosity which is a bit exaggerated over the mottling visible in the ISO1600 x 4 minute 20 sub averaged image I produced the same night.
I don't see this doing away with lower ISO multi-sub deep sky imaging and replacing it with high ISO capture. But I can see applications in some shooting situations where multiple subs may not be possible:- aurorae, fast moving comets, some nightscape photography.
I'll post the unadulterated final image in the deep sky section.
Joe Cali