Originally posted by vancmann Nice! The first image, the D7000's, its hard to rate because although no noise, there are no fine details in that image.
The second one, if it is indeed the 51,200, looks dam good. I messed around with that image using just NN and while I removed the noise, I found it hard to keep any details. Had trouble finding the "sweet spot" of less noise and more details.
What would be your reccomendation of software to get in addition to NN for noise removal?
Thats a bit harsh on the first image, Most of the detail is still there, at least with what there was to start with.
Topaz Denoise 5 is the best NR utility..... John concurs on this. Although it does not remove blotchy noise, it rebuilds detail by analyzing noise patterns....Its a very clever program but easy to use. If you have NN then you will know its a smearfest like Noiseware. Which has its uses for background, flat surface work. to remove blotches. But not for the heaviest of noise or in detailed areas. The algorithms work best at finer noise patters (read anything up to 12800 on K-5)
protect highlights by applying partial masks. quick mask/normal mask.
Here is the 25600 Version, this is very easy to do. Its simply Open in ACR and apply chroma NR to liking. Apply small amount of sharpening. Apply Topaz Denoise, apply some more slight sharpening (bigger radius this time but less strong)
I tried to get a similar result using the Pentax Camera Utility instead of ACR. Because ACR builds the image with much finer grain, it was easy to NR, the Pentax utility was not!, but it produces nicer colors.
I just redid the 51200 version. simpler, and faster. As I used DCraw for my first attempt, this time I used ACR.
I found out something new with the way Topaz behaves. If I sharpened the image first, it picked up the noise easier at a lower setting whilst retaining detail . I then masked the image to use Noiseware on the background. (like Noise Ninja) I then added noise to the same masked parts so as to make the image look grainy, but not so noisy.
Here is the result, its better than the other 51200 image I did a few weeks back. live and learn.
Again, this time it took me 15-20 minutes. If I had knows what I was doing beforehand I would have done it in 5-10.