dallvr wrote:
LR Classic Denoise vs. DxO DeepPRIMEXD test on lion at 12,800 ISO shot just before nightfall.
A couple of notes, I processed both on unedited Sony raw files. The Adobe version stacked the processed file with the original raw. The DxO doesn't do that. When I applied my previous edits (including cropping) to the Adobe version, it looked identical to my originally processed version except with less noise. But the DxO had color shifts compared to my original version. I re-edited the DxO version to match my original version. In addition the DxO version did not copy over my keywords to the new dng. I have been working with their tech support ever since version 3 came out to find out why this is happening. So the workflow has more steps with DxO.
You can judge for yourself which you find more appealing.
DXO looks a bit over sharpened, you can see it in the hair ends and muzzle of Lion
This is not say DXO DeepPRIMEXD is inferior to Adobe's new AI noise, because its a excellent product, its just a matter of tweaking the sliders, perhaps somewhere between theses two sample images you posted.. Trail & Error
I believe after working in DXO DeepPRIMEXD for some time, you will find a sweet spot for the sliders, depending on each image
IndyFab wrote:
DXO looks a bit over sharpened, you can see it in the hair ends and muzzle of Lion
This is not say DXO DeepPRIMEXD is inferior to Adobe's new AI noise, because its a excellent product, its just a matter of tweaking the sliders, perhaps somewhere between theses two sample images you posted.. Trail & Error
I believe after working in DXO DeepPRIMEXD for some time, you will find a sweet spot for the sliders, depending on each image
Beautiful capture !!
Thanks. I just played around with these tools a bit more, using darker photos shot at 20k ISO of a leopard on rocks. Interestingly, even when I dialed down the sharpening on DeepPRIMEXD to soft, the results are significantly sharper with DeepPRIME compared to Adobe. The raw files are sharp even though the noise is atrocious. The Adobe Denoised ones look mushy compared to DeepPRIME. So experimentation will be necessary for different situations. I am delighted that there are more tools out there to reduce noise.
dallvr wrote:
LR Classic Denoise vs. DxO DeepPRIMEXD test on lion at 12,800 ISO shot just before nightfall.
A couple of notes, I processed both on unedited Sony raw files. The Adobe version stacked the processed file with the original raw. The DxO doesn't do that. When I applied my previous edits (including cropping) to the Adobe version, it looked identical to my originally processed version except with less noise. But the DxO had color shifts compared to my original version.
The DXO DNG files are almost transparent compared to the original raw files. At least in my experience. But you should never copy paste a white balance from the raw file to the DNG file. For whatever reason, the same white balance (visually) leads to différent values in the blue/yellow and green/magenta sliders.
The initial white balance stays of course correct when you convert the file, but it doesn't display the same values.
Can anyone with an M1/M2 MacBook Pro comment on how these new tools are pushing CPU/GPU use? Wondering if I need to consider 64gb RAM in my next laptop when I upgrade my underspec'ed M1 Pro I got with a lowly 16gb RAM.
Ayoul wrote:
The DXO DNG files are almost transparent compared to the original raw files. At least in my experience. But you should never copy paste a white balance from the raw file to the DNG file. For whatever reason, the same white balance (visually) leads to différent values in the blue/yellow and green/magenta sliders.
The initial white balance stays of course correct when you convert the file, but it doesn't display the same values.
I have noticed this color shift in the DxO processed files as well, but not on every file. I usually do the denoising on raw files that I have taken back to default settings, but keep a virtual copy with my processed settings so that I can sync them back to the new processed files. Sometimes there is no color change with DxO, but quite often there is, with the processed files looking more yellow and somewhat more magenta. I have not seen this happen with the Adobe denoise, at least not so far.
dallvr wrote:
DXO looks a bit over sharpened, you can see it in the hair ends and muzzle of Lion
This is not say DXO DeepPRIMEXD is inferior to Adobe's new AI noise, because its a excellent product, its just a matter of tweaking the sliders, perhaps somewhere between theses two sample images you posted.. Trail & Error
I believe after working in DXO DeepPRIMEXD for some time, you will find a sweet spot for the sliders, depending on each image
Beautiful capture !!
I think that DXO’s demosaicing and lens corrections produce crisper looking images than Adobe’s, even without denoise.
Goodrich wrote:
I think that DXO’s demosaicing and lens corrections produce crisper looking images than Adobe’s, even without denoise.
DXO does use a different approach. DPP has DLO which is true digital deconvolution. Not sure if DXO does that but they do specialize in adjusting for each lens. I think Adobe uses a generic method.
Also for lens sharpening. They offer USM but that is only there as a back up in case a lens is not on their list. DXO has put a lot of time and effort into this.
It does look crisper but to me Adobe looks more natural. One thing that has not been discussed in depth is how much more you can now push Texture, Sharpening and Detail without introducing artifacts. Not sure about others but I have been putting in a lot of time messing around with those. My workflow has changed.
The Detail slider is probably the least used and least understood adjustment. Least understood because there is so little information on it. While subtle you can see the difference.
Zenon Char wrote:
The Detail slider is probably the least used and least understood adjustment. Least understood because there is so little information on it. While subtle you can see the difference.
Several years ago a member of Adobe's technical staff had explained it on Luminous Landscape. The Detail slider represents weighting between Unsharp mask and Deconvolution. Beyond 50% Deconvolution is predominant.
Rajan Parrikar wrote:
Several years ago a member of Adobe's technical staff had explained it on Luminous Landscape. The Detail slider represents weighting between Unsharp mask and Deconvolution. Beyond 50% Deconvolution is predominant.
I actually learned that a LL and asked at the Lightroom Queens site to confirm. 0-50 is for halo suppression. 50-100 is for deconvolution but you lose halo suppression.
I created ISO Adaptive presets for each camera. At 100 ISO Detail is set to 75. As the ISO goes up the Detail goes down proportionately. That will remain as I won't put all files through Denoise. For the ones that go through Denoise everything has changed. I can push everything more.
Need for speed. Denoise is by far the most advanced of the three Enhance features and makes very intensive use of the GPU. For best performance, use a GPU with a large amount of memory, ideally at least 8 GB. On macOS, prefer an Apple silicon machine with lots of memory. On Windows, use GPUs with ML acceleration hardware, such as NVIDIA RTX with TensorCores. A faster GPU means faster results.
hoodlum90 wrote:
On macOS, prefer an Apple silicon machine with lots of memory.
On M1 and above, memory will probably help some, but so far, I haven't seen many folks with times faster than my M1 Pro 16gb laptop gets. Windows boxes are the ones struggling with the new Denoise the most.
Need for speed. Denoise is by far the most advanced of the three Enhance features and makes very intensive use of the GPU. For best performance, use a GPU with a large amount of memory, ideally at least 8 GB. On macOS, prefer an Apple silicon machine with lots of memory. On Windows, use GPUs with ML acceleration hardware, such as NVIDIA RTX with TensorCores. A faster GPU means faster results.
The minimum GPU requirement of 8GB for Denoise in LR hasn’t been updated on that page. The last update to that page was on Mar 3rd, before the 12.3 release.
Well today I removed Photo AI off my OS as well. All that is left is Sharpen AI which I don't use very often. As long as I have supported cameras. I think someone here suggested I check my account. I did see a support renewal package for $99 for Photo AI and the 3 apps. Those 3 apps may not be listed for sale or support by October.
There is no way I'll be shelling out $99 for the package or less for an individual app. It would have to be very reasonably priced. I predict they won't offer support for the individual apps individually if they are still there.
So when should we start bugging Adobe for Sharpen AI? Don't really need it as you can do well in LrC, especially with these new clean files. More for de-blurring for those times when you need it.
Zenon Char wrote:
Well today I removed Photo AI off my OS as well. All that is left is Sharpen AI which I don't use very often. As long as I have supported cameras. I think someone here suggested I check my account. I did see a support renewal package for $99 for Photo AI and the 3 apps. Those 3 apps may not be listed for sale or support by October.
There is no way I'll be shelling out $99 for the package or less for an individual app. It would have to be very reasonably priced. I predict they won't offer support for the individual apps individually if they are still there.
So when should we start bugging Adobe for Sharpen AI? Don't really need it as you can do well in LrC, especially with these new clean files. More for de-blurring for those times when you need it.
I find Adobe NR AI to be superior to Topaz (and OnOne and DxO) in NR abilities.
I do a lot of photography where motion blur is an issue - panning shots of fast moving subjects like race cars, airplanes, unlimited hydroplanes and the like.
The only thing Adobe does not do for me that Topaz and OnOne can do is in reducing motion blur. So I will continue to use Topaz Sharpen AI until Adobe gets a competing product.
I think we should start bugging Adobe now for a product in this market space - Adobe Sharpen AI.
I find Adobe NR AI to be superior to Topaz (and OnOne and DxO) in NR abilities.
I do a lot of photography where motion blur is an issue - panning shots of fast moving subjects like race cars, airplanes, unlimited hydroplanes and the like.
The only thing Adobe does not do for me that Topaz and OnOne can do is in reducing motion blur. So I will continue to use Topaz Sharpen AI until Adobe gets a competing product.
I think we should start bugging Adobe now for a product in this market space - Adobe Sharpen AI....Show more →
After all these years of testing and deciding which 3rd party to use. I won't miss that. It feels like a weight has been lifted off my shoulders. Think of all extra beer money by not buying/updating apps as well.
Messing around a bit more with it. Adobe, etc are saying not make any adjustments before sending it to Denoise. Unlike DXO LrC adjustments are retained in Denoise. I created a preset to take everything to zero before using it. A few times I forgot and thought it didn't make much difference. As long as you avoid multiple edits and masking. However I watched a video last night that showed it does make a difference. If you leave sharpening at the default of 40 and there is noise, it is sharpened. That winds up going to Denoise. The presenter showed the difference.
I also have a post Denoise preset that puts sharpening back which is a little higher than 40 and several other settings for the starting point.
It only took my system 34 days to process a single image. I liked the results but I won't be using this new LR feature as it stands right now.
If anyone has any suggestions on how to improve the speed performance I'm all ears. I'd love to stay inside LR for most of my processing, but as of now, DXO PR3 is my workflow of choice for NR work.
Are there any LR "preferences" tweeks we can make to improve performance?
I'm running this for my processing system.
CPU: AMD Ryzen 9 3900XT 12-Core, 24threads 3.8GHz
GPU: GeForce RTX 2060 6 GB
Memory: 64 GB
Win 10 Pro
lighthound wrote:
It only took my system 34 days to process a single image. I liked the results but I won't be using this new LR feature as it stands right now.
If anyone has any suggestions on how to improve the speed performance I'm all ears. I'd love to stay inside LR for most of my processing, but as of now, DXO PR3 is my workflow of choice for NR work.
Are there any LR "preferences" tweeks we can make to improve performance?
I'm running this for my processing system.
CPU: AMD Ryzen 9 3900XT 12-Core, 24threads 3.8GHz
GPU: GeForce RTX 2060 6 GB
Memory: 64 GB
Win 10 Pro ...Show more →
A little hard to tell how severe your issue is with the hyperbole, but...
Edit > Preferences > Performance tab > Make sure Use GPU is selected for image processing and export. Update your Nvidia drivers.
Don't really need it as you can do well in LrC, especially with these new clean files. More for de-blurring for those times when you need it.