When you shoot with a Micro 4/3 camera (mine is a Panasonic Lumix LX100 Mark II), which has a relatively small sensor, you run into problems in low-light conditions. The ISO goes up, the digital noise increases, and you end up losing a lot of detail. In such situations – low light combined with a small sensor – sometimes the results from a smartphone can actually look better. In the documentation I gathered for
Padova decorata (
https://www.padova-decorata.it), there are two or three photos that are practically unwatchable because of how poor in detail they are.
Only AI can really recover images like these, since the information is lost to begin with and no traditional photo editing can bring back what isn’t there. But I don’t have expensive AI software that applies secret algorithms to enhance image quality. I came across an open-source project, freely available and usable in my workflow, called
nind-denoise (
https://github.com/hqhoang/nind-denoise). For those who use Darktable, here’s more info:
https://discuss.pixls.us/t/feedback-needed-integrating-nind-denoise-with-darktable/51393. I tried installing it on my computer – not at all straightforward – but it did allow me to save those two or three images I wanted to include in
Padova decorata.
As a test, I also tried nind-denoise on this photo, and here’s the result: a lot more detail, no noise, and almost everything in focus. But I’m not sure whether I prefer this version, which feels almost
too perfect, or the original one. What do you think? Sincerely, what do you think?
