Home · Register · Join Upload & Sell

Moderated by: Fred Miranda
Username  

  New fredmiranda.com Mobile Site
  New Feature: SMS Notification alert
  New Feature: Buy & Sell Watchlist
  

FM Forums | Post-processing & Printing | Join Upload & Sell

1       2       3              15              17       18       19       end
  

Lightroom Classic AI Denoise (12.3)

  
 
Charlie San
Offline
• • •
Upload & Sell: On
p.16 #1 · p.16 #1 · Lightroom Classic AI Denoise (12.3)


Re Topaz again, I forgot about B&W Effects! But a weird thing happened yesterday evening after my remarks. I found and opened an old image and tried to open Denoise. I couldn’t apply the changes I made, it said I was on a trial version and to buy the app, one I have been using extensively as I work on some old film images! A glitch or is big brother watching. I could log into Topaz but had to change my password. I’ll try it again today, hopefully just a glitch.


Jun 14, 2023 at 08:01 AM
Zenon Char
Offline
• • • • • •
Upload & Sell: Off
p.16 #2 · p.16 #2 · Lightroom Classic AI Denoise (12.3)


Charlie San wrote:
Re Topaz again, I forgot about B&W Effects! But a weird thing happened yesterday evening after my remarks. I found and opened an old image and tried to open Denoise. I couldn’t apply the changes I made, it said I was on a trial version and to buy the app, one I have been using extensively as I work on some old film images! A glitch or is big brother watching. I could log into Topaz but had to change my password. I’ll try it again today, hopefully just a glitch.


You most likely need to enter the license key code.



Jun 14, 2023 at 09:15 AM
aKansasKid
Offline

Upload & Sell: Off
p.16 #3 · p.16 #3 · Lightroom Classic AI Denoise (12.3)


Charlie San wrote:
Re Topaz again, I forgot about B&W Effects! But a weird thing happened yesterday evening after my remarks. I found and opened an old image and tried to open Denoise. I couldn’t apply the changes I made, it said I was on a trial version and to buy the app, one I have been using extensively as I work on some old film images! A glitch or is big brother watching. I could log into Topaz but had to change my password. I’ll try it again today, hopefully just a glitch.


Same thing happened to me, and everyone else. Topaz admitted on their community forum yesterday that Sharpen and Denoise authenticate your login by connecting to their servers. They use AWS (Amazon Web Services), which had a major outage much of Tuesday. AWS was restored by 5 PM CDT, and logins worked after that.

(This also means you can't use Sharpen or Denoise without an internet connection - previously undisclosed. )



Jun 14, 2023 at 10:43 AM
Zenon Char
Offline
• • • • • •
Upload & Sell: Off
p.16 #4 · p.16 #4 · Lightroom Classic AI Denoise (12.3)




aKansasKid wrote:
Same thing happened to me, and everyone else. Topaz admitted on their community forum yesterday that Sharpen and Denoise authenticate your login by connecting to their servers. They use AWS (Amazon Web Services), which had a major outage much of Tuesday. AWS was restored by 5 PM CDT, and logins worked after that.

(This also means you can't use Sharpen or Denoise without an internet connection - previously undisclosed. )


So much for owning it.



Jun 14, 2023 at 04:33 PM
schlotz
Offline
• • • • •
Upload & Sell: On
p.16 #5 · p.16 #5 · Lightroom Classic AI Denoise (12.3)


If the "undisclosed" is accurate, that would be a major issue not having been disclosed at time of purchase.


Jun 15, 2023 at 08:19 AM
Blueshound
Offline
• • •
Upload & Sell: Off
p.16 #6 · p.16 #6 · Lightroom Classic AI Denoise (12.3)


Should be easy to test, simply turn off network / internet access, boot up one or more of the indicated Topaz apps and see what happens.

I've just tried that, turned off internet access, and directly booted Topaz Denoise AI. It came up normally, and I was able to browse to a suitable low light image, and load it. The Denoise app functions seemed to work as normal, and I was able to both denoise and sharpen a RAW file, and save it back to the original folder as a .dng file.

Checking the folder in Bridge shows the denoised and sharpened file in .dng format.

Did I miss something?

Edit: I'm referring to Denoise AI, not older apps. I've also just tried calling Topaz Denoise AI as a plugin to PS CC with WiFi and hence internet access turned off, and that also worked as normal.

Brian

Edited on Jun 15, 2023 at 08:53 AM · View previous versions



Jun 15, 2023 at 08:41 AM
jhapeman
Offline
• • • •
Upload & Sell: On
p.16 #7 · p.16 #7 · Lightroom Classic AI Denoise (12.3)


Blueshound wrote:
Should be easy to test, simply turn off network / internet access, boot up one or more of the indicated Topaz apps and see what happens.

I've just tried that, turned off internet access, and directly booted Topaz Denoise. It came up normally, and I was able to browse to a suitable low light image, and load it. The Denoise app functions seemed to work as normal, and I was able to both denoise and sharpen a RAW file, and save it back to the original folder as a .dng file.

Checking the folder in Bridge shows the denoised and sharpened file
...Show more

I think if you haven't logged in for a while it might need to ping the server and validate your license is active? That would be a pretty typical approach.



Jun 15, 2023 at 08:52 AM
aKansasKid
Offline

Upload & Sell: Off
p.16 #8 · p.16 #8 · Lightroom Classic AI Denoise (12.3)


jhapeman wrote:
I think if you haven't logged in for a while it might need to ping the server and validate your license is active? That would be a pretty typical approach.


From "ida.topazlabs" on Topaz Community forum June 13:

"That is a good point and to be transparent with you there’s a bug in the older apps where they have to talk to the server every startup.

Essentially, you are not being logged out, the app just doesn’t know that the version you have exists without our server being active."

Also from ida that day:

"Newer apps like Topaz Video AI and Topaz Photo AI are only affected if your last login was more than one month ago or your license is expired"



Jun 15, 2023 at 10:30 AM
exdeejjjaaaa
Offline
• • •
Upload & Sell: Off
p.16 #9 · p.16 #9 · Lightroom Classic AI Denoise (12.3)


mcbroomf wrote:
Interested in Tensor cores


And why are you sure that on PC/Win platform software like ACR NR or DxO PL DeepPrimeXD are using tensor cores at all ? unlike Apple Mac/OSX platform there is probably no standard way to use them like neural engine in M1/M2, more so w/ other GPU vendors like AMD and Intel present ... AMD GPUs do not have anything like tensor cores at all - no ? so what is the point for software developers to bother at this stage




Jun 18, 2023 at 09:44 AM
mcbroomf
Offline
• • • • •
Upload & Sell: Off
p.16 #10 · p.16 #10 · Lightroom Classic AI Denoise (12.3)


I have no idea about DXO and have never mentioned it. But Eric Chan wrote that for Win Adobe Denoise uses Nvidia Tensor cores. This is the only reason I added them to the table. He made no mention of AMD or Intel GPUs which is why I didn't include them.

Not sure what you mean by the last question ...



Jun 18, 2023 at 09:54 AM
 


Search in Used Dept. 

mcbroomf
Offline
• • • • •
Upload & Sell: Off
p.16 #11 · p.16 #11 · Lightroom Classic AI Denoise (12.3)


I should also add that on the Mac side while Eric Chan wrote that Denoise uses Neural core, that Adobe has since admitted that it does not due to a problem on Apple's side. No mention of the specifics mind you. That is why I added GPU cores to the Mac table for Denoise.


Jun 18, 2023 at 09:58 AM
exdeejjjaaaa
Offline
• • •
Upload & Sell: Off
p.16 #12 · p.16 #12 · Lightroom Classic AI Denoise (12.3)


mcbroomf wrote:
But Eric Chan wrote that for Win Adobe Denoise uses Nvidia Tensor cores.


you are referring to this line = "Finally, we built our machine learning models to take full advantage of the latest platform technologies, including NVIDIA’s TensorCores and the Apple Neural Engine. Using these technologies enables our models to run faster on modern hardware."

but it does not directly sound that ACR/LR code actually uses it __to apply__ NR right now ...

in any case a good thing will be to compare nVidia GPU w/ tensor cores vs AMD GPU (that do not have tensor cores) based on similar OpenCL test results ( https://browser.geekbench.com/opencl-benchmarks ) and see how those 2 GPUs compare in Adobe's NR ( then nVidia while having similar OpenCL benchmark must be a very noticeably faster than AMD )



Jun 18, 2023 at 10:12 AM
jhapeman
Offline
• • • •
Upload & Sell: On
p.16 #13 · p.16 #13 · Lightroom Classic AI Denoise (12.3)


exdeejjjaaaa wrote:
And why are you sure that on PC/Win platform software like ACR NR or DxO PL DeepPrimeXD are using tensor cores at all ? unlike Apple Mac/OSX platform there is probably no standard way to use them like neural engine in M1/M2, more so w/ other GPU vendors like AMD and Intel present ... AMD GPUs do not have anything like tensor cores at all - no ? so what is the point for software developers to bother at this stage



Yes. The tensor cores are used just like Apple's neural engines to enhance certain tasks, among them things like machine learning model inference used in things like Adobe's Denoise and in DxO's Denoise. Nvidia has created a very easy framework for developers to leverage this; it's not a small part of their success in this space. Most of the vendors have set up APIs to make it a lot more transparent to take advantage of these cores, but it's still early days overall. DxO has been using the Nvidia Tensor cores since PL5, BTW.

It's one reason AMD has lagged; they are behind the curve in this area. They were supposedly going to introduce tensor cores in the RDNA3 generation but then chose not to.



Jun 18, 2023 at 10:21 AM
mcbroomf
Offline
• • • • •
Upload & Sell: Off
p.16 #14 · p.16 #14 · Lightroom Classic AI Denoise (12.3)


exdeejjjaaaa wrote:
you are referring to this line = "Finally, we built our machine learning models to take full advantage of the latest platform technologies, including NVIDIA’s TensorCores and the Apple Neural Engine. Using these technologies enables our models to run faster on modern hardware."

but it does not directly sound that ACR/LR code actually uses it __to apply__ NR right now ...

in any case a good thing will be to compare nVidia GPU w/ tensor cores vs AMD GPU (that do not have tensor cores) based on similar OpenCL test results ( https://browser.geekbench.com/opencl-benchmarks ) and see how those 2 GPUs compare
...Show more

Chan's whole article is about Denoise and includes this comment;

"Denoise uses machine learning to interpolate those patterns and remove noise at the same time. That is, our models are designed and trained to perform both demosaicing and denoising in a single step."

If you'd like to add AMD GPU data and other parameters to the table let me know and I'll send you the excel file.




Jun 18, 2023 at 10:30 AM
jhapeman
Offline
• • • •
Upload & Sell: On
p.16 #15 · p.16 #15 · Lightroom Classic AI Denoise (12.3)


exdeejjjaaaa wrote:
you are referring to this line = "Finally, we built our machine learning models to take full advantage of the latest platform technologies, including NVIDIA’s TensorCores and the Apple Neural Engine. Using these technologies enables our models to run faster on modern hardware."

but it does not directly sound that ACR/LR code actually uses it __to apply__ NR right now ...

in any case a good thing will be to compare nVidia GPU w/ tensor cores vs AMD GPU (that do not have tensor cores) based on similar OpenCL test results ( https://browser.geekbench.com/opencl-benchmarks ) and see how those 2 GPUs compare
...Show more

You are incorrect. He is directly stating that they are using those models run on the tensor cores. It's very clearly supported by the data @mcbroomf has already shared--as the number of tensor cores goes up, the performance increases. It's not linear, but on the other hand older cards without tensor cores perform *very* poorly.

AMD has implemented a different approach in their cards that lets them handle some of these types of calculations without having tensor cores but it isn't able to keep up, as the bar charts I showed demonstrate. It's not a huge difference by any means, but it is a difference. I don't have 79000X series card to test, as the 3D rendering software we use performs much better on Nvidia cards, so its not worth me buying one, but it would be interesting to see how that card competes.

One of the conclusions from some of this is that if you can get even a 30x0 series Nvidia card into a PC, you'll get most of the benefits of the AI functions at a very reasonable price and with quite good performance. At the same time, for these photographic tasks the Apple GPUs compete with the higher-end 40x0 cards, which is impressive for a first-generation product. This is particularly the case as we know that Adobe still can't leverage the full power of the ANE due to the bug in MacOS Ventura, so there's an anticipated performance boost coming at some point soon.

The OpenCL benchmark would be a poor choice here because there's not enough in there to directly compare some of the newer tasks being done in the real world and because it's easy for vendors to code to benchmarks to help sell their products. Much better to just directly compare how both Nvidia and AMD cards run these tasks.



Jun 18, 2023 at 10:32 AM
exdeejjjaaaa
Offline
• • •
Upload & Sell: Off
p.16 #16 · p.16 #16 · Lightroom Classic AI Denoise (12.3)


mcbroomf wrote:
I have no idea about DXO


DxO PL6 has an option for Mac/OSX to select whether you are using Neural Enginer -or- just regular GPU cores - there you can actually compare what is the gain and be sure what you are using - that is always a good thing for a use... on PC/Win side - no such thing, unfortunately ...

PS: as a side note DxO is ~1.5+ times faster than Adobe if testing batch conversion of the same raw files in my case ( 40 x A7R4 raw files = 4 per minute /raw -> linear DNG/ in ACR Enhance NR and 6.5 per minute /raw -> linear DNG/ in DxO PL6 DeepPrimeXD )




Jun 18, 2023 at 10:42 AM
jhapeman
Offline
• • • •
Upload & Sell: On
p.16 #17 · p.16 #17 · Lightroom Classic AI Denoise (12.3)


exdeejjjaaaa wrote:
DxO PL6 has an option for Mac/OSX to select whether you are using Neural Enginer -or- just regular GPU cores - there you can actually compare what is the gain and be sure what you are using - that is always a good thing for a use... on PC/Win side - no such thing, unfortunately ...

PS: as a side note DxO is ~1.5+ times faster than Adobe if testing batch conversion of the same raw files in my case ( 40 x A7R4 raw files = 4 per minute /raw -> linear DNG/ in ACR Enhance NR and 6.5 per
...Show more

What are you doing in the batch conversion? In my experience Adobe seems very slow on the DNG creation side.



Jun 18, 2023 at 10:47 AM
rscheffler
Offline
• • • • • •
Upload & Sell: Off
p.16 #18 · p.16 #18 · Lightroom Classic AI Denoise (12.3)


Well, late to the party and finally updated LRC from somewhere before 12.3 to 12.4 to take advantage of the AI Denoise goodness.

Tried this 24MP ISO 6400 file with the subject requiring about another +1 push. First is Denoise on left at 67% and right is traditional LR luminance NR at 50 (sharpening at default, not disabled). Second is the same Denoise settings as the first, and at right is luminance NR disabled. Quite the difference. Love that the Denoise version is not the mush of traditional NR and can be tweaked further. May pull it back more or add some grain to tame the waxy plastic look. Would be tempting to run this on all files ISO 3200 and higher (Canon R6/R6II), but may reserve it for 6400+. This example, the DNG ended up ~100MB vs. ~33MB of the original.

Is everyone just keeping the new DNGs despite the size? For some events I'm dealing with hundreds of images and it all adds up. Might just keep the originals with their XMP adjustments/settings sidecars and run through Denoise again if there is a need to revisit/reproces.






Bought DXO PL6 on Black Friday but hadn't bothered installing it, in part due to the hassle of leaving the LR environment on a case by case basis. Being able to do this within LR makes this so much more useful.



Jun 18, 2023 at 05:48 PM
mcbroomf
Offline
• • • • •
Upload & Sell: Off
p.16 #19 · p.16 #19 · Lightroom Classic AI Denoise (12.3)


I've kept Denoise dngs for the time bring but I won't for files that are exported to jpgs and only destined for the web. For print files I keep psbs or single layer tiffs so I will probably kill the dngs for those.

I've only used a setting higher than 50 for abstracts where I've wanted all noise gone and kind of looking for a plastic effect. I keep a little noise and/or add some texture. I zero out all details sliders before denoise and (re)sharpen after denoise. I think I may test grain instead of or with texture.



Jun 18, 2023 at 06:07 PM
rdeloe
Offline
• • • •
Upload & Sell: Off
p.16 #20 · p.16 #20 · Lightroom Classic AI Denoise (12.3)


rscheffler wrote:
Well, late to the party and finally updated LRC from somewhere before 12.3 to 12.4 to take advantage of the AI Denoise goodness.

Tried this 24MP ISO 6400 file with the subject requiring about another +1 push. First is Denoise on left at 67% and right is traditional LR luminance NR at 50 (sharpening at default, not disabled). Second is the same Denoise settings as the first, and at right is luminance NR disabled. Quite the difference. Love that the Denoise version is not the mush of traditional NR and can be tweaked further. May pull it back more or
...Show more

Gotta love the University of Waterloo PhD robes and hat!



Jun 18, 2023 at 06:47 PM
1       2       3              15              17       18       19       end






FM Forums | Post-processing & Printing | Join Upload & Sell

1       2       3              15              17       18       19       end
    
 

You are not logged in. Login or Register

Username       Or Reset password



This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.