Adobe recently rolled out one of its biggest Lightroom feature updates in quite some time: AI noise reduction. Promising to vastly out-perform the previous system, the new functionality has some stiff competition from many other options. How does it perform in practice? This great video overview takes a look at what you can expect.
Coming to you from Mark Duffy, the awesome video takes a look at the new AI noise reduction functionality in Lightroom. I have had a few weeks to check out the feature as well, and I have to say I have been quite impressed so far. It is certainly a major step up from the old feature. Although it might not edge out some of the results from competitors, it is close enough that the convenience has mostly won me over. Having the functionality right there where I do most of my cataloging and editing work makes it much more convenient and efficient than round-tripping to another program and makes me more likely to apply it on more borderline photos that I might have let slide before. Still, you should find the right balance of convenience and results for your needs and workflow. Check out the video above for the full rundown from Duffy.
Presentation felt a bit rushed and rough around the edges.
Isn't that why i stated "unplanned video" at the very start. This was a reaction video, not a tutorial. I briefly used the AI the previous night and left it to figure it out on camera. Prime example is when i realised the button to deactivate a panel has been replaced with a click and hold view button which is totally dumb to me. Sorry you didnt like the video but it was never intended as a tutorial.
Requiring the generation of a new file and excluding the upscale option isn’t useful. Surely, the team making this ai tool is connected deeply with the Lr core team meaning they could make it much smoother such as combining the raw and new DNG/tiff into a PSD package where turning off the Denoise layer shows the original file. Same with upscaler. Could all be packaged as a single file with proxy viewing. Why ditch topaz software at this point?
There doesnt seem to be an option to sync across other photos like with the new masks as of late. So with me working on M1 Mac Mini, thats 2 minutes per photo, selected individually which means i cant just go grab a coffee while it works away. I'm sure they'll integrate it better for our workflows soon enough.
I think your issue is the Mac Mini. I am using a basic Apple Studio with 32 GB of RAM and getting processing times of 25 seconds with Canon R5 46mp files.
Until the image is viewed on a different RAW viewer i guess the test is incomplete. I recently did a test on a series of astro shot and on other RAW viewers I had a ton of hot pixels peeping out. But LrC preview showed none. There were issues with all denoise in some ways or other e.g. too much sharpening or halos etc. But over all DxO and Topaz results are pretty good. Topaz probably got best result fast. On1 was softening the images quite a bit. PhotoLab was quite demanding when processed in a batch but most consistent result in my test.
Topaz out of DXO Pure Raw 3 and Lightroom Classic has the worse AI on its noise reduction. It just makes far too many mistakes.
I do hope Adobe offers this with CR too.