Does Lightroom's Sharpening & Noise Reduction STINK?

[@AnthonyMorganti]

In this tutorial, I demo and compare Lightroom's sharpening and noise reduction to Topaz Labs Photo AI to Topaz Labs Denoise AI. Note that I should have, in the video, added some texture and clarity to the Green Heron although, I will say, doing it later, although it helped sharpen up the bird, it still was no where near as good a the other two images. Get my newsletter! anthonymorganti.substack.com/subscribe Check out the Topaz Labs Software here: bit.ly/3cDqa5J Currently, I do not have a personal promo code to share. Here is a video I previously did comparing Photo AI to DeNoise AI: kzbin.info/www/bejne/mpDSaoCrotuohbM Here is ANOTHER video I did previously comparing Photo AI to Denoise AI: kzbin.info/www/bejne/qojJh5dsqbSkb6M *I am an affiliate for all of the companies listed and will earn a commission if you purchase anything. Please read my Code Of Ethics Statement: onlinephotographytraining.com/code-of-ethics/ Here is my list of recommended software: wp.me/P9QUvD-ozx Here is a list of my current cameras, lenses, etc.: wp.me/P9QUvD-ozG Please help support my KZbin channel - consider purchasing my Lightroom Presets: www.anthonymorganti.com/ Help me help others learn photography. You can quickly offer your support here, where I receive 100% of your kind gift: ko-fi.com/anthonymorganti You can change the default amount to the amount you want to donate.

[@eric.p.merlin3537]

thx Anthony, I use DxO PhotoLab ELITE v6 myself, which i prefer .. Still, thx for your video

[@DaveReasonsSr]

WOW!!!!!!! I have owned Topaz Denoise AI for a while and had never used it. I tried it out on an image of a Sandhill Crane that was shot during a heavy fog. It was the ultimate test and Denoise AI did one heck of a job. Thanks for making this tutorial!!!!!!!!

[@stevejarvis186]

Hi Anthony. I only recently signed up to get your free newsletter and I found the first one I received about how you got into photography from a very young boy very inspiring. I look forward to reading more of these.

[@tomlitteral5738]

Thanks!

[@AnthonyMorganti]

Thank you Tom!

[@warrenlloyd7532]

Interesting video Anthony. Just wondering what the ISO setting was for that image? I've found for up to about ISO 800 Lightroom works quite well, but above this I definitely use DeNoise AI

[@stretchini]

Thanks again Anthony, I hope LR develops a decent Noise/Sharpen combo. I use Topaz and ON1

[@perryelder5045]

When you added global sharping with LR why didn't you use masking, that way you would not have to worry about adding nosie to the background, I think LR works pretty good for these as long as there is not a lot of noise and the subject is tack sharp, otherwise the AI plug-in work better, really liked your last couple of videos goes to show you have to pixel pep to really see the difference

[@redfordcole]

Adjusting the masking slider, along with the ones for radius and detail in the sharpening section is definitely part of my workflow for any images that need a lot of work. Not sure how the results would compare with the AI versions, as you’re still, in this case, trying to make an out of focus image in focus, but only using the basic sharpening and not taking advantage of the more advanced adjustments doesn’t make sense to me for the sake of this comparison. The other thing I’d add is that it’s very helpful to hit the option key, or alt or whatever on PC, when doing the sharpener mask adjustment, as you can focus on details, like the feathers in this case, that you are trying to preserve or emphasize. Bird’s beak also looked a little over baked in one of the AI images, but still an impressive improvement, just curious what they’d have looked like if the full sharpening panel had been utilized.

[@eartho]

Lr/ACR has always had the industry worst sharpening algo. It's the only sharpening i know of that creates details that don't exist and also can change the apparent exposure of an image when pushed too hard. And believe it or not, the algo was even worse 15 years ago before i spent months proving to the Adobe teams just how bad it was. They did little to fix it though, other than changing the amount of effect, but the underlying algo remains the same to this day.

[@christophmunch4796]

LR desperately needs an update to its sharpening and denoising algorithms so that they are on a 2023-worthy level instead of 2005!

[@ThirdEyeAngel]

Especially if they are going to charge us for a subscription instead of selling us the software -_-

[@fixitright9709]

I often wondered why programs by Adobe like Lightroom seem to have done nothing at all to improve their noise reduction and sharpening tools, I mean like never! You would think they own stock in topaz maybe?

[@BillFerris]

Interesting. The green heron is clearly out of focus. DeNoise AI does an impressive job of manufacturing sharpness from an image I'd normally delete or not bother processing. With the advances Adobe has made in using AI to demosaic images, I've gotta believe they're working on sharpening algorithms capable of repairing a missed focus shot. But as of now, they're far behind in this area. As others have mentioned, I find the masking slider essential to my sharpening and noise reduction workflow in Lightroom. I find a combination of masking and amount that totals 100-110 often produces a good result. Then, I'll use masks to apply sharpening and noise reduction selectively. Thanks again for sharing your workflow.

[@TrailHiker52]

Thank you for the video, Topaz has changed the way I take photos with Denoise being a big part of that transformation. I noticed the more "painterly" look of LR on it's own vs. Topaz. One quick thing on the pronunciation. I thought it was pronounced "Dee-noise" as un-noise vs. the short e of "De-noise". I may be wrong though. Thanks again, great video.

[@mirrorlessreflections]

Hi Anthony, this is interesting. However I have a method I use quite often, and not necessarily just on noisy photos. I generally mask the subject, then adjust as required to enhance/sharpen etc. Then I create a second mask for the background and take the texture, clarity and sometimes the dehaze down. It is usually also best to adjust the exposure in this mask, I then usually end up with a nice soft noiseless background. Granted this doesn't always work for all images but it's just another work around, plus t's dead easy to save it as a preset!

[@leniehulse1621]

I believe Anthony has done a video on this same method!

[@mirrorlessreflections]

@@leniehulse1621 Argh! Missed that, thanks.

[@stevecockbain8649]

Thanks Anthony, interesting and educational, as usual. You mention a newsletter but I have been subscribed to your channel for a couple of years now and not ever received a newsletter. Have I missed something somewhere?

[@kurtissutley1485]

There have been times I couldn't reduce sharpening enough in Photo AI to keep from over sharpening. In those cases I went back to Lightroom and added a little sharpening globally. It's rare but it's also very easy to over sharpen in Photo AI. That's my experience.

[@Wrecked01]

How about compared to DXO Photolab 6? I usually use PL6 to denoise and then import the DNG into LR.

[@davidromano2421]

Nice video. The sharpening and noise reduction is one of the reasons why I left lightroom. I'm normally at iso 100 so noise is not much of an issue at all, but sharpening is. In fact, I typically apply zero luminance reduction, only color reduction. In capture one, the masking is worlds better than lightroom. The capture one sharpening is not best in class either, but it is good, and with the superb masking tools, I can apply sharpening only where I want it. I do still use DXO for noisy images but it's rare that I need to. Maybe I'm different from a lot of people having done so much with film and darkroom, but I really don't mind noise that much as long as it is nice looking, and also not color noise. Color noise I have zero tolerance for. Luminance noise is OK.

[@ХристоРъцев]

Same here, C1 is way better than LR in noise reduction and sharpening. I usually go for global settings and mask rarely. One thing that still keeps LR up is maybe the interface is more friendly and you get more adjustment faster through the sliders compared to C1. Also I think catalogues in C1 are complicated.

[@waratah5422]

Lightroom does menu things with a small hard drive foot print, the other two are made for sharpening with large file size & $

[@86BBUB]

Every one of my shots, no matter the ISO, goes through DeNoise now. I love the way it sharpens images, more than Sharpen AI.

[@KaosK9]

Its great to see that the devs still have not addressed the issue of Photo AI obliterating fine detail. A comparison of 10:15 of DeNoise AI to 10:25 of Photo AI clearly shows how PAI destroys the fine detail in the feathers behind the eye. Sad to see so many Photo Influencers propping this product up when it still doesnt even match the quality of the original DeNoise AI. DeNoise is still a superior product.

[@BobG-eh5fc]

Curious if Photoshop sharpening would be better, especially all the different techniques in PS.

[@tonysun2051]

Not sure what happened: tried to change settings to save image as DNG after processing RAW file in Topaz Denoise AI, but when clicking SAVE IMAGE, however TIFF file saved back to Lightroom, and worse is that edited version still looks too bad noises in contrast to preview denoised file in Topaz DeNoise. Any idea, Anthony. Need your help please.

[@Suhailkhan53]

Great video

[@kbruff2010]

Thank you for this information

[@BobG-eh5fc]

Thank you great overview. Maybe one day Adobe will address this gap, but for now Topaz does.

[@jamesshrouds4130]

How does sharpening in PS compare to the three you show in the video?

[@billferreira7263]

PS and Lightroom use the same underlying software for sharpening, the differences come down to the masking capabilities for which PS is more capable.

[@craigpiferphotography]

I know that I could look for myself, but have you done a comparison of Topaz vs OnOne? I did trials for both, plus DXO, and ended up pulling the trigger for OnOne. It seemed to do a better job on the photo that I had in front of me, and it was considerably less expensive than Topaz on black Friday. I'm just not seeing much out there that really gives a good comparison.

[@billferreira7263]

After owning the Lightroom, ON1 and Topaz tools long enough to develop an opinion: For Noise Reduction: Lightroom is years behind; ON1 NoNoise AI and Topaz DeNoise AI are outstanding. But the batch processing in ON1 makes the process effortless while Topaz is painful. For Sharpening: Topaz wins easily. ON1 has released the first version of their AI sharpening tool and it is comparable to Lightroom. Just as it took Topaz a couple of years to get the kinks worked out of its tool suite, I expect it to take a bit for ON1 to get there with AI sharpening .

[@craigpiferphotography]

@@billferreira7263 thanks for the comment. I do like that ON1 put the sharpening in NoNoise, even if it's still a work in progress. My biggest complaint is that the plugins are all separate, meaning that you have to do one, then do the next one on the resulting file and so on. At least Topaz made a move to eliminate that with Photo AI so that it's only 1 roundtrip back to LR.

[@billferreira7263]

The latest ON1, both the stand-alone/plug-in NoNoise AI and Photo Raw, have merged noise reduction and sharpening into a single tool. But for my Fuji raw images, I'm getting poor results from ON1's AI sharpening. Other than for the basics, I'm still struggling with ON1 Photo Raw. Because the conversion cannot handle Lightroom's Transform adjustments, migrating my 30k image library would be a nightmare.

[@Bigtbone205]

Seems like most reviews of topaz use pics of birds. Seems their ai is trained on bird photos but when used in general low light street photography I can’t see the benefit over standard Lightroom,photoshop or affinity.

[@glennwebster6344]

My preference for using LR is to not do global adjustments , but do it separately on subject and background. I am not sure after removing all the noise globally that you would be able to bring back the detail. The other think I have been doing is to make the background a little less saturated, and blur it even more. This makes the subject stand out even more. Is there an advantage to doing the adjustments globally? Also I think that the busy heron can tolerate more noise in order to get more sharpness.

[@rhiwderinraytube]

I have abandoned Lightroom entirely in favour of ON1 Photo Raw. I hate LR’s annual subscription scheme for a start. Anyway, there are several applications now on the market which are better, faster and bring out better updates than LR. Sharpening and Denoise are cases in point.

[@jpdj2715]

Does Lightroom's Sharpening & Noise Reduction STINK? That's a valid question. The Mudbricks (*) probably wouldn't understand this though, and think the question is invalid. Because sharpening and noise processing are done in a plugin to Lightroom called "Camera Raw" (ACR). "Lightroom" (LrC for "Classic") helps us build catalogs and offers non-destructive modification of how our images are rendered. It also now offers some modifications that we could call "Photoshop (Ps) Light" but this too is implemented in a non-destructive way. Anthony knows this, but I point to it because otherwise the Mudbricks - programmer techies and their spreadsheet managers - will not understand our feedback. The culprit that REALLY STINKS is Adobe Camera Raw (ACR). The Mudbricks discuss ACR as plugin to LrC. So LrC does all those things I mentioned except "raw processing". ACR is a plugin into LrC but this is not visible in the user interface of LrC. In Ps however, ACR being a plugin is extremely visible. In the Ps context ACR may seem to be able to do things we cannot do in LrC, but that's due to Ps. We, users that can choose with our feet, have to postulate that all noise in our digital images that is within the dynamic range of our cameras, actually is "FAILED RAW PROCESSING". As the Mudbricks have used my money, yours too, of the past decade to develop new apps doing the same, but on another platform, no doubt in new code streams and code maintenance streams, thus raising the cost (L) in their P&L, they have not made a priority of improving raw processing, apparently. As subscribers of a decade or so, we have not been paid our dividends and if we held "real" (sic) shares, at the next shareholder meeting we should propose to have all bonuses stripped away from the past 10 years - as we have been cheated on. The problem between a corporation and its leaders on one side and the "real" shareholders on the other is that nobody has a problem as long as the $$ numbers are made. That's not a problem for a day trader, but a serious investor with a longer term perspective may end up with a loss when the attitude of not enough innovation starts to cause loss of sales. If we collectively as - I would argue - the most important stakeholders do not become vocal, then nothing is going to be innovated. So, for example, while Nikon had their own "free" raw processing software for PC/Mac in the past, they stopped that at some point when they started collaborating with the Mudbricks. Well, Nikon rebuilt their raw processing app and started releasing it again. That, to me, is a "writing on the wall". We have to understand that all straight out of camera (SOOC) JPEG images are the result of in-camera raw processing, and that non-raw movie formats are like JPEG images that have had a second level of data compression applied for frame sequences. So while our cameras do all that raw processing and subsequent data compression in real time, exporting a raw image to JPEG in ACR (LrC) relatively speaking takes ages. Talk about "rendering a video from raw" ... . And our favorite "in post" software may actually not be as good as the in-camera version. The reason we shoot raw is because we want to influence how the raw processing is done, before the lossy data compression. Well, if you shoot the Ken Rockwell way (of interwebs and YT fame - he never shoots raw, he says), you might argue here that some camera brands actually offer you different profiles for the in-camera raw processing, and then there are cameras that allow you to create and apply, in camera, your self-defined profiles. If we look at the DxO Mark company's work, we can see that what they call the best "sensors" (they mean "cameras") have a color space from 14 bits raw that does not exceed 27 bits. Even with 16 bits cameras. How come? Well, the 14 are monochrome bits in each data element that represents the EV, measured by one photosite that is filtered down to one color, either red, or green, or blue. Raw processing must turn a single photosite's 14 Red bits into RGB, and do that with the Green and Blue ones too. The raw file is pure color noise. Ugly AF - and that does not mean autofocus here. While the camera knows how to do the raw processing, our problem is that we may want to process our images in one way today and a completely different one tomorrow. That's why WE WANT RAW even when we often do not really need it. This channel - Anthony's - very well illustrates why. Also, as the camera would give us 24 bits pixels in JPEG (8 bits red, 8 bits green, 8 bits blue per pixel, but stored with lossy compression) we still can get 27 bits (9+9+9 = 8 times better and no lossy compression) from our 14 bits monochrome raw data elements. What about DeNoise AI, Sharpen AI, Gigapixel AI (Topaz's holy trinity now amended with the fourth musketeer "Photo AI"), or DxO's PureRAW, or ... ? These work as plugins to LrC in a way that makes these plugins actually complementary to ACR. Bluntly put, these 3rd party apps offer seriously improved noise processing and sharpening that either fix ACR's already made failures, or manipulate a raw image in a way that ACR gives a better result. One of the [camera brand] rumor sites hypothesizes that some model of their favorite brand may pre-process a raw file so as to have less noise later (i.e. when coming out of ACR) and the complaint is that it is not perfect. But, to me, is a "writing on the wall". When the Mudbricks ask us if we could recommend to friends, relatives, other businesses, their apps, then my answer is, NO, not as long as I need a 3rd party app to fix your failures. These questions generally come once a year, some time before your subscription needs renewal. Well ahead of that moment to not think about the money you are going to spend. And the question only serves to make you accept their price yet another time a couple months later. Well, we have to be more critical. Running Adobe on MS Windows, I need a very powerful workstation and a migration to Linux might cut that in half while slashing the prices at all levels. I cannot run LrC and Ps in a stable way on a Windows/Intel notebook with 16GB RAM and discrete GPU with 4GB video RAM - when processing images from my 46MP camera with two 4K displays attached. That notebook takes 100W max power, by the way and with two displays and a few peripherals would burn through 1 kWh every three hours. So this stinks. My two beast workstations sized for LrC and Ps with two 4K displays burn through some 3,000 kWh per year. 750W "Titanium" efficiency class PSU each. Usually with only one display each. Considering a computer running a Linux operating system generally needs half the resources of a Windows PC, I would go to Linux today rather than tomorrow if the apps I need were available at it. But the Linux world focuses on spending (wasting) effort on distros rather than on compatibility and drivers, rather than simplifying the UI by removing different ways to do the same management tasks. This Linux world has "free" raw processing software too but the UI is a mess and I still need Ps and other tools. An Apple computer? I hate paying even more for a brand name than I already do, especially a totally closed system. I lived with it for 10 years in the 90s, have hated Windows for its deep-rooted issues, but closed and limited - meh. Arrogance and vanity. Yes, ACR stinks. (*) a mudbrick is also called an adobe.

[@billferreira7263]

So much work for an inferior result. For my work flow all my raw images are batch processed by ON1 NoNoise AI with the noise defaults but sharpening disabled and saved as DNGs. I find that 100% of the images are improved, even those shot at the lowest ISOs. The raws and DNGs are imported into Lightroom. For those shots needing sharpening, I use the Topaz DeNoise AI on the DNG. ON1 and Topaz noise reduction is comparable but the ON1 batch process takes all of the work out of the process. On1 is just getting started with AI sharpening and at least for Fuji X-trans images (X-T3 and X-T1), the results are not very good.

[@Malaise05]

I usually like your KZbins but I think this is a biased comparison. You have clearly not used global sharpening to its best in LR. You could get much better fine tuning by using the mask in the Detail panel. As others have also said, you have started with an image not properly focused on the bird’s eye and unnecessarily shot at a high ISO thus adding more noise.

[@wellwhatthen10101]

Lightroom is lacking behind the denoise

[@waratah5422]

Auto correct! Many

[@davidlawrence7540]

I would reject this image as it isn't sharp in the first place! You can't, how ever hard you try sharpen something that isn't sharp! Why such a high ISO, no need stationary bird! Get everything right in camera, no need for all this noise reduction software! Where has all the camera craft gone, just fire away and shove through denoise!!! Perhaps it's KZbin making it look blurry!!

[@DennisMathias]

LR stinks :-)..at least as far as noise/sharpen goes. AI Photo really is amazing.

[@LaughingLion4Ever]

Yes it stinks.

[@HansvanMeteren]

I don't like this. Stink?! What a word!! It works pretty well except when pixel peeping

[@JV-qb6ib]

Great job Anthony!

[@TC_Conner]

Photo AI is definitely better than the other two in this instance Anthony. But I'd be willing to bet a Yankee dime (Google it) that it actually depends on each photo and if you took a different photo with similar needsd and did this same comparison you'd get a new winner. 😎

[@lawrencerayburn5172]

Thanks!

[@AnthonyMorganti]

Thank you Lawrence!