LR AI NR Question from Steve's Video

If you would like to post, you'll need to register. Note that if you have a BCG store account, you'll need a new, separate account here (we keep the two sites separate for security purposes).

Watched the video, which was very helpful overall, but it left me with a question. I'll preface by saying...I HATE post-processing, but...I get it...has to happen. Anyway, I'm no expert, but I am able to get what I want to happen. I could probably get a lot more to "happen" if I was more savvy, but...that's another thread. Here's my question. I had read several other LR expert explanations of how to approach AI NR in Lightroom, and they state that it should generally be done first with no sharpening or adjustment to the image. Then, adjust and sharpen the DNG and go from there. That's what I have been doing and it's been pretty amazing. Steve is making some adjustments prior to applying denoise. He does a very good job explaining that this is just one approach, but I'd like to understand it a little better. Do you prefer pre-NR adjustments or not and why?
If you care to contribute to my education...appreciation in advance.
 
He said he does basic stuff like exposure and WB before NR…if the shot is reasonably exposed I just do NR and then those but a little exposure change or WB doesn’t really affect the noise much. I generally run DxO for NR first although sometimes Topaz AI and I do need to try his techniques side by side on a photo or three to see if the improvement is worth the extra time it takes. I wish DxO had an adjustment slider to change the level of NR but you have to buy their more expensive LR equivalent for that.
 
I appreciate that Anjin. I have to say, I've processed some ISO 4000 shots from my D500 that I would not even have considered messing with that came out beautifully, and I didn't have to do anything but apply Denoise. It has been a great addition. I do have DXO, but this is so much better and easier. I'll probably need to replace my computer as it is older and it takes an insane amount of time to do a D850 file. It's worth it though. If I was a professional, I'd have no choice. It takes about 50 minutes per file on a full-size D850. Yeah...I know:confused:.
 
I don't think lightroom users have to worry about doing edits in any order, since lightroom has its own internal order to how it applies the edits, so it doesn't matter what order you do the edits, before the dng or after. However if I understand Steve's point, he gets a better idea as to how much of the NR is needed by doing it his way, so I think it's fine either way. In general one wouldn't sharpen noise, but that would only apply if the rendered file was sent to photoshop. Within lightroom the order won't matter. If you did sharpen pre dng you could dial it back after if needed since the sharpening or any edit isnt baked in until the export.
 
Last edited:
Regarding your 50 minute processing times with your current PC...you MAY be able to speed up processing greatly with the addition of an nVidia 20 or 30 series GPU to your current PC. The base model 2060 and 3060 GPU's include the AI hardware (Tensor Cores) that do the heavy lifting for AI DeNoise processing on both Lightroom and DXO PhotoLab.

There are a lot of these GPU's on the used (and new) market that can be had very affordably and they don't have the massive power requirements of the higher end 80 and 90 versions in each series (20xx, 30xx and 40xx). Of course much will depend on your current configuration, but it's worth considering if you aren't prepared to buy a whole new system and are comfortable upgrading your current PC. A new RTX3060 can be had for under $300...used around $200 in my local area. Gamers always upgrading and getting what they can for their used gear. My SWAG would be that a 3060 would get you down around a 3-5 minutes processing time, but I don't have one to test with. My 3090 is around 10 seconds or less for most Z8 or Z9 files
 
Good video here and a quote from Julianne Kost:

Denoise carries over any enhancements made to the source image. However I would suggest that you apply Denoise early in your workflow so that you’re making adjustments on the cleanest file possible. Otherwise, enhancements that you’ve made (Sharpening and Texture for example), may need to be revisited after running Denoise.

 
Last edited:
Watched the video, which was very helpful overall, but it left me with a question. I'll preface by saying...I HATE post-processing, but...I get it...has to happen. Anyway, I'm no expert, but I am able to get what I want to happen. I could probably get a lot more to "happen" if I was more savvy, but...that's another thread. Here's my question. I had read several other LR expert explanations of how to approach AI NR in Lightroom, and they state that it should generally be done first with no sharpening or adjustment to the image. Then, adjust and sharpen the DNG and go from there. That's what I have been doing and it's been pretty amazing. Steve is making some adjustments prior to applying denoise. He does a very good job explaining that this is just one approach, but I'd like to understand it a little better. Do you prefer pre-NR adjustments or not and why?
If you care to contribute to my education...appreciation in advance.
Conventional wisdom when working with "pixel editors" it do do NR first so that no noise enhancement caused by sharpening, clarity, etc, get "baked in". That's not a concern with LR. If I recall Steve explained why he applies a bit of sharpening before running the AI NR. Paraphrasing it simply gives a better idea for how the noise is cleaning up and how it may affect sharpening overall in critical areas like eyes, etc. He also used masking with the sharpening so that noise isn't affected in BG, etc. The only thing that is "baked in" to the "enhanced" DNG file that LR creates is the effect of the NR. Sharpening, lighting, etc slider will be in the same positions when the DNG opens and can be adjusted as desired.
 
Back
Top