Oh, OK I get it now. When you say “if you don’t have access to the original RAW file” you pretty much mean if you only have jpegs and didn’t shoot in RAW. So if I have a library of RAW files that get directly converted to TIFFs, that would be ideal. That’s what you meant by access to the original RAW file, not actually opening the RAW file directly in DeNoise AI. But that said, the optimum method would have been to open the RAW file directly with DeNoise AI.
Maybe that’s a feature Topaz can add: direct transfer of the RAW file from Lightroom to DeNoise AI. It seems like it is possible because other software (e.g., DxO PhotoLab) lets you do it by going to File–>Plug-in Extras–>Transfer to [DxO PhotoLab 3]. Then again, it’s no different than opening the RAW directly in DeNoise AI and making a TIFF and then importing that TIFF into Lightroom. It’s just that I do basic global adjustments (esp. exposure and white balance) to the RAW file first before sending it to DeNoise AI.
The little experience I have had with processing photos with color noise matches with what you say. It does seem better to set the Color Noise Reduction slider in the RAW converter first. The fact that the Chroma Noise Reduction sliders are outside the Auto adjustments seems to indicate there is no AI involved with that yet.
i found that if I export image from lightroom with default 25 in color noise reduction and 0 in sharpening, it will have artifact easily with DeNoise model, especially for the human face, lower the color noise slider in lightroom can help to reduce artifact generated by DeNoise model
if i turn off the noise reduction and sharpening in lightroom, and use the AI Clear model, it always produces over softened image
it’s difficult to have a satisfy result with image exported by lightroom
The masking feature of DeNoise AI is quite appealing I must say. The RAW processing for Olympus ORF files is however rather disappointing, to the point of being unusable I am afraid. Lens corrections being totally absent for one thing has a huge impact, especially now that it has become part of the lens design to have PP corrections applied. Then the colors and even overall exposure require significant tweaking. So if one uses Oly RAW, I strongly recommend to process first in Oly Workshop, Olympus proprietary RAW editor to generate a TIF or 100%JPEG as an input to DeNoise AI. It is still up for debate if one should apply Oly Workshop NR or not before DeNoise AI. I have much further testing to do before I draw that conclusion.
@Roger250 I agree with you on the Olympus ORF raw files. My workflow with these files begins with DxO PhotoLab, including its Prime denoise tool. Then there’s no need for DeNoise AI. I tend to use DeNoise AI as a plugin in Topaz Studio 2 for special situations that may arise.
Doesn’t PL spawn these processes in the background? If I’m not mistaken? I’ve noticed it’s not super-fast, but not slow enough to where it causes a problem for me (obviously, on my system). I’m not sure if Prime-NR uses the GPU. Anyway, I use both.
As mentioned above in the opening post, DeNoise first then use Sharpen AI instead of LR sharpening. Note that the image will need to be a TIF (preferably) passed to DeNoise from LR with no noise reduction applied.