Can Lightroom's NEW Denoise AI Replace DxO, Topaz, and ON1?
Вставка
- Опубліковано 18 кві 2023
- Lightroom Editing Companion - bio.site/alexarmitage
Denoise has likely made the most progress in a relatively short amount of time over the past couple years. Last year I used it on a few wild life images and basically came to the conclusion that you can shoot at extremely high ISOs without much worry. After testing quite a few programs, I settled on DxO PureRaw as my go to program. Within this video I used DxO PureRaw 3, Topaz Denoise AI, and ON1 NoiNoise AI 2022.
Lightroom just released it's own AI Denoise integration and the results will surprise you!
𝐒𝐨𝐜𝐢𝐚𝐥𝐬
Website: alexarmitage.com
Patreon: / alexarmitage
Discord Community: / discord
Instagram: / alexarmitage_
Vero: vero.co/alexarmitage
𝐆𝐞𝐚𝐫 (Affiliate Links)
𝐅𝐮𝐥𝐥 𝐆𝐞𝐚𝐫 𝐋𝐢𝐬𝐭 𝐚𝐧𝐝 𝐌𝐲 𝐎𝐩𝐢𝐧𝐢𝐨𝐧𝐬 : www.alexarmitage.com/gear
New Photography Camera: geni.us/Canon-R5
New Video Camera: geni.us/Canon-R7
Drone: geni.us/Mavic-2Pro
Old Video Camera: geni.us/FujiXT4Body
Old Photography Camera: geni.us/5DMark4
Camping 𝐆𝐞𝐚𝐫
Big Table: geni.us/campbigtable
Small Table: geni.us/campsmalltable
Camp Stove: geni.us/everestcampstove
Cutting Board: geni.us/campcuttingboard
Knife: geni.us/campknife
Skillet: geni.us/camplegendskillet
Fridge: geni.us/campfridge
Power/Battery: geni.us/campbattery
Camp Light: geni.us/camplight
Music (Typically) from Musicbed:
fm.pxf.io/c/3300666/1347628/1...
I want to clarify that this is strictly a noise reduction comparison. I didn't use things like Topaz Photo AI because I wanted to compare noise reduction (also haven't had great luck with its results). All of the programs here enhance detail/sharpening after reducing noise.
Interesting observations. I have to add that it is necessary to be transparent about the actual versions of software, camera, and camera firmware version in the comparison. And we have to distinguish two kinds of noise: (a) at reasonably normal light levels, especially in the below average EV zone, within the camera's contrast envelope (CE, this is the camera's Dynamic Range - DR - available in a single shot. The generally published DR is merely the camera's operating range and almost totally irrelevant to individual shots.) (b) At low light levels and with high ISO.
Your introduction with the video suggest you shot the image with a Canon R5 and while it has pretty high resolution, it still may have an anti-aliasing (AA) filter. I call that AA filter a "fuzzy filter" and this was introduced decades ago to make raw processing easier. In a Bayer architecture digital colour system, you shoot monochrome data elements into the raw file - there are no RGB pixels in it. In "raw processing" the missing GB are guessed with the monochrome R data elements, the missing RB are guessed with the monochrome G data elements, and the missing RG are guessed with the monochrome B data elements. That "raw processing" is computationally stressing, but at the time of its inception with low resolutions was not big deal. It just could not happen efficiently in camera without help, given the processing power of the time and its energy need, as well as the price for needed power. So the fuzzy filter disperses light travelling to the sensor and this reduces low light sensitivity, reduces colour space, reduces DR and CE, and reduces contour sharpness. But it helps wild-assed (but mathematically precise and repeatable) colour guessing, deals with jagged edges following from the sensor grid of squares and helps address banding at least a bit. As the fuzzy filter also adds to the vignetting problem of "digital", and this increases with increasing resolution, Nikon dropped it at above 24MP per full frame photosite density in the sensor (so it is absent in a 20 MP APS-C camera). My impression is that Canon wants to retain it.
The problem of the wild-assed colour guessing (deBayerisation) is that it generates colour noise and luminance noise beyond what we legitimately could attribute to the sensor (so most attributions are false, IMO). The problem with the demosaicking is that it may remove recognisable artefacts (like Moiré), but not the ones we call noise.
In this sense, almost all colour and luminance noise we see, so far, in our photos, especially of type (a), follows from failed raw processing. The digital photography does not want "us" to understand this. And its marketers have added to the general confusion by suggesting that a sensor has pixels (picture elements).
So, AA filter or not is an important distinguishing factor as Adobe Camera Raw (ACR) was originally conceived when all Bayer cameras had a fuzzy filter. Nikon eliminated the fuzzy filter from their D800E, some 10 years ago, and my suspicion is that Adobe never fully adapted ACR's raw processing algorithm to that (e.g. by having an alternative algorithm). Having more than one algorithm is not a weird expectation because the Bayer algorithm of wild-assed colour guessing does not work with Xtrans sensors. Also, with Bayer sensors, the colour-guessing works reasonably in the middle of the raw image, but not along the edges. So we need either a separate edge algorithm, or a sensor with more rows and columns outside the "effective" ones that are hardware help to make Bayer raw processing easier with a single algorithm.
In the case of Canon, I have read remarks in the Interwebs that they may actually modify their raw data so as to generate a Bayer raw file that generates less noise. To me that is a rumour, and it is unqualified in the light of (a) and (b) above.
As image quality (IQ) totally depends on these kind of details, we need to be transparent about them.
Today's version of Lightroom Classic has features that yesterday's version did not have. An improvement of an algorithm may work well with one camera and its firmware version, not another.
For the case of (a), my experience was until ACR 15.3 that Topaz did the best job. I'm not pointing to Topaz DeNoise AI here, but actually got convinced by Topaz Gigapixel AI to buy the suite, not the other apps in the suite. Its detail retrieval in the deBayerisation and demosaicking that are raw processing is phenomenal. In my genres of photography. It may be less good with night sky photos - which I generally don't do.
Having discussed raw processing, we have to know about Adobe that all raw processing is done in ACR. Both in Lightroom (Classic) and in Photoshop. LrC masks this in its user interface, but Adobe informs us about the ACR "plug-in" in their website. Ps is transparent about doing raw processing but making an excursion to the ACR plug-in. When ACR is done, Ps works with the result from ACR - and in LrC it works precisely like that. In LrC, the internal workspace (memory) holds ACR's result as processed into the ProPhoto colour space. This means that what we see on our monitors/displays is a compressed rendition in order to render in the sRGB colour space. This is not the same as bits-per-channel depth, though and in Ps we can render ACR's result as 16 bits per channel or even 32 bits per channel. The difference is visible in the compressed rendition on our monitors. If we have a great, calibrated, monitor that is also calibrated for ambient light. And assuming it has no dynamic contrast (implemented by variable backlighting of the actual LCD screen). We should also know that "14 bits depth"in our raw files reference the monochrome data elements before raw processing. A ProPhoto colour space rendition at 32 bits per channel is a 100% AI-imagination of what the subject may have been like.
In the specific case of the comparison here, the banding/striping artefact in the Topaz rendition suggests a sensor scan issue in the Canon's raw file to be raw processed into clear visibility.
I did not hear you say anything about the different choices you have with Topaz Denoise like Clear, Standard, Low light, RAW, Severe noise, etc where you have different sliders to control your results. There are noticeable differences in results according to the option you select. I'm not saying that Topaz is better or as good as Lightroom, I'm just saying that you kind of cut short your test of Topaz.
Nice one, I'm 100% with you on this one, I've been going through a bunch of shots from a very low light indoor shoot and was really struggling with the manual noise reduction lightroom had before the update, most shots were really soft and honesty was a bit worried about what the client would say, anyhow I saw the updated ai noise tool and picked my absolute worst image I could find and was completely blown away at how good the new reduction is, it went from a throw out shot to a great shot 🎉 This has honestly saved this whole shoot and I've used it on most of the images with 90% perfect results.. Absolute game changer for me and I don't normally use those words...ever. Thanks for the excellent comparison it's great to see it compared like this..
Great comparison video. well done.
Great job Alex ! I was ready to buy Topaz Photo AI. Glad I waited also.
Thanks for doing the work. Very, very helpful to know how LR compares to the others. (What a relief, actually.)
Great review Alex! I for one have been using Topaz DeNoise for several years but never the RAW setting. It's performed well for me but after seeing these results I'm going to revisit some of my shots and try out the new Lightroom tool. Looks impressive to say the least.
Great video Alex. Brings it to life. My own "testing" felt like the LR denoise was good, but its great to see an expert doing it properly!
Wait whos the expert here? Haha :)
Helpful comparison. Thanks.
Thanks so much for putting this out quickly, great comparison!
Thanks Adam!
I’ve watched a few of these videos, once I saw you made one I knew it’d be the best and it, of course, it was. Thank you!
Too kind. Thanks buddy
Great intro reel!
I have been a huge fan of Topaz products! I use them daily! I’m anxious to try this improved function in Lightroom! Thanks for tipping us off on this!
Would love to know your thoughts Randy!
Very good review. Most accurate youtube review so far!!!
Agreed, I tried it on some of my atro images from my f/4 lense and it’s impressive. It’s so good I can’t convince myself to upgrade to a faster lense. I can’t afford one anyways but now it’s not as big of a deal. Super impressed, thank you Lightroom and adobe!
Subbed for this!🎉
Dear Alex. A week ago I did for myself this same check with Topaz, DxO and CameraRaw 15.3. I ended up with the exact same conclusions that You made!!! Thx.
I’m really happy with this addition into LR, it’s a welcome addition
Excellent comparison - 0% filler in this 15 min vid.
Thanks for the test. I've been holding off buying DxO thinking Adobe cant afford to stay behind forever and they have a lot of resources. Maybe PureRaw still is in the lead, but Adobe seem to be at least close enough to make any PureRaw advantage marginal. I'll watch some other reviews too and see if there are differences between the results depending on camera types.
Great presentation, Alex. I have been considering the purchase of a Denoise program but won't do so. I will continue to use Lightroom Classic as my primary editing tool going forward! Thanks so much for the great analysis!
Thank you!
so jealous Alex, i wish i was in your shoes!!! keep up the traveling and showing us great locations
Thanks so much Tim!
Thank you very much for the comparisons. As a Lightroom user I had planned to buy separate software to denoise and so using money on it. That money stays in my pocket now.
Great comparison Alex! Thanks for taking the time. Very detailed, so to speak! I also like the results. I’m going to use it right away. One thing I might like in the future is the ability to denoise an area, shadows for example, without denoising an entire image. Do you think this is worthwhile sometimes? Thanks. Geoff
Not sure if that would be useful but I do know other programs allow you to mask certain areas for sharpening?
Bravo !
I think most of these comparisons aren’t fair at all. When adding sharpening, you often introduce artifacts and other oddities. You only add sharpening in LR near the very end of the comparison and then, don’t really adjust any of the sharpening in the other programs at all.
You can’t compare photos being fully auto adjusted to just denoised photos. It’s not apples to apples.
I’d like to see you take the same photo, spend 10 minutes on each, adjusting sharpness and denoise with all programs and see what the outcome is.
Hi Alex,
I was going to purchase Dxo pure raw 3 after my trial which finishes in next 2 days, Adobe new Denoise is way better than any other programme as per your comparison , has forced me to stick with adobe Denoise which will get better and better as time goes. Thanks you got new subscriber.
Welcome and thank you!
I was looking at Lightroom's Denoise on my own images and I thought I was going crazy but I saw the same thing. For DXO pure raw though, you have to turn off sharpening and then images are a lot closer (but I still think LR looks more natural and I actually reverted to using the PR2 algorithm because of the PR3 artifacts). I don't own ON1 so I can't comment on that one. Topaz needs a lot of manual tweaking but you can get close too - it's not automatic though. What's impressive with LR is that so far it seems to need zero manual intervention (and as you pointed out, it's included in the price, so that's nice).
Are you referring to the "lens softness" sharpening? Otherwise I didn't see anything related to sharpening in the settings.
thanks alex, time to time i add noise to my pictures :)
The hype is real!
Thank you for the video. Do you know if the Lightroom denoise Ai can be synced or pasted to several images at once?
Very good comparison, thanks. But as for sharpening, when I compare the raw file to the new dng file, the dng has more detail so I think maybe LR is adding some sharpening.
CaptureOne have "dehalo" slider and PERFECTLY dials back sharpening effect from DxO matching to that from Lr.
Having watched your video, I think that you may have had a preferred result in mind (but I absolutely think you have a right to your opinion).
I agree that it was only between light room and DXO. I thought the nighttime Aurora picture wasn't the right choice for an objective real world scenario of the average user... thousands of tiny stars and blotches of green fragmented light on night sky is a tough test for this kind of software... the details look like noise.
The bear picture was a fantastic choice. Yet, for some reason; even after sharpening, I can't see how you came to your preference. Look at the bears round ear, look at the salmon, light room also smoothed out a lot of the individual ripples in the fast flowing water. I almost thought the lightroom one was a long exposure. I might have also mentioned that you purchase DXO once and have it forever, whereas you must continue to pay a monthly subscription for Lightroom.
I'll have to try with some more images. My initial tests of a portrait taken at ISO 20,000 was that Adobe's denoise made areas of the skin and surrounding details look like clay. Me and a colleague both agreed DxO did a far better job. Not saying this is the case with all images but perhaps in that more extreme circumstance DxO still has its place.
After all these ai upgrades, i expected ai noise reduction ... and it is.
Something we've been waiting for. I don't use pure raw, but i think you can make som adjustments in Pure Raw. Maybe dxo photolab 6 could beat lightroom. Anyway very impressive. I'm glad i didn't upgrade to dxo photolab 6. I hope adobe will add this feature to lightroom mobile.
Adobe did a great job and not overcompensate noise reduction with AI sharpening. Just the right balance of what we needed.
You are right about Lightroom looking more pleasant in most of the examples, even compared to deep prime. But ... if you applied the same level of sharpness, I think artifacts are introduced, and the difference will be hard to tell. If the pleasant soft look could be turned into the same pleasant sharpness - it would be a miracle - because it probably will conflict with the laws for perceived optical sharpness. I can't back this up with my own tests, as I have left Adobe. The all over improvement though is fantastic and has been needed for a long time.
I definitely need more time with it and the correct workflow. I'm wondering if denoising in LR and sharpening elsewhere will be the go-to method.
I completely agree. The Lightroom version is really soft, and trying to get back any detail close to DXO will be impossible, even if you use something like Topaz AI sharpening.
I think Lighroom needs to find a better ballance. It does a great job on mostly homogeneous surfaces like a sky, but it should lower the noise reduction in areas with a lot of detail.
Thanks for an excellent video! Do you know if I can batch process multiple images in Lightroom with the new AI denoise feature?
Currently I dont think you can but I suspect it will be something we can do in the future if I had to guess
@@AlexArmitage Thanks!
I have used topaz and I'm using DXO and I think Lightroom is onto something here, all they need to do is add a slider to add sharpening texture clarity and Lightroom will definitely have them all beat
Short answer = YES!!!
Great Alex! Have actually been looking into Topaz and it's nice to know I can stay in Lightroom. Btw, what lavalier mic are you using? It's not on your gear list. I'm about to buy my first :)
Oh good call! I didn't realize its not on there. I'm using a 32bit system and the audio you hear in this video is actually from my NTG3, but I had a lav on in case it got too windy. I only use the lav for windy/distant situations so my use case is really narrow and I wouldn't suggest what I'm using for a main mic setup even though it works well.
@@AlexArmitage Thanks for replying Alex :) I was actually about to buy 32bit system from zoom because of not having to worry about clipping as much and since it would be cheaper than wireless. But ended up with RØDE wireless go ii single on sale plus their lav go. Realized I don’t have the time to match video and audio (barely have the time to record anything, working full time in office, thus no videos yet) so I rather take some clipping than fewer videos published. But if I was able to switch track I would definitely chose 32 bit.
Great comparison Alex, looks like Adobe has knocked it out the park, so glad I waited and haven't upgraded my Pure Raw
What percentage of nr did you use on lightroom?
I downloaded a trial of Topaz AI last week and was ready to pull the trigger. Glad I waited!! This new feature is absolutely going to bankrupt some of these software companies. And watch out for the price hike coming from Adobe I’m sure…
For DXO, i will never tick lenses collection and sharpening as I think it is way too much. I normally use it for noise reduction only and the rest I let Lr takes care of it.
I did try Lr AI deniose with my Milky Way photos with a really dark landscape. The result was not that great as there were so many artifacts on my photo. So I think Lr AI deniose might be good for some photos (As I can see from your video , ISO not too high) but not all of them.
Thanks for a good comparison.
I agree Lightroom is impressive. However, as a DXO user it has the significant advantages that it does the job much faster and can process files in batch mode, unless I'm missing something in lightroom.
My mistake, you can batch process them. Very impressive.
In my own experience, DeepPrime(XD) is indeed far better than Topaz or others, especially when it comes to fake details and artifacts. However, the issue also comes from PureRaw rather than PhotoLab, where you can lower oversharpening. Still, after trying I also think than LR is far better overall, with very few artifacts and a very natural look of processed images. Also, you don"t have typical DxO blurred parts surrounded by oversharpening areas : the processing is much more even across the frame with LR. Funny to try : add a bit of grain after removing noise and your pictures will look even sharper ;-)
Hi do you have to convert to dng can you use straight out of camera raw
I have sony a7iv
Not able to currently not compatible to my raw files
Try updating camera raw? Not sure why it's not working for you!
Can you use AI denoise with masks? Or is it only available as a universal option? Additionally, how do they stack up against one another in processing time/RAM usage?
Nice Presentation Alex. I only wish with LR it could be part of a non-destructive workflow, by being able to bring it into photoshop as a smart object.
I assume it could come in the future!
I was in the middle of a free trial for PureRAW 3, and was ready to drop the $130 for it. This LRC release could not have come at a better time.
Same! :)
Same here, even send Dxo a mail for maybe a little discount just for the plug in, as they do often discounts, and their answer was no discounts, if I wanted a discount I had to wait... Well Dxo, thank you for letting me wait! Cyaaaaaaaa
I tried Lightroom's version. It did as good a job as Topaz, but took 10 times as long to process it.
Lightroom Denoise only works on Bayer and Xtrans raw files. It tried it on some files from a monochrome camera that I’m renting for a week and it didn’t support the files. Converting to TIFF doesn’t help, but those are handled by Topaz.
Good morning Alex 🌈
Morning Linda!
I recently bought Topaz Denoise to improve grain on low light shots. LR new Denoise gives me much better results, I wish I hadn't bought Topaz's
DxO exposure "problem" comes from DxO files are linear and Lr does not apply curves on them like on files that Lr imports and debayer itself.
The time of processing remains important. I personally have not compared how long processing takes regarding Lightroom's new adjustment for denoise and Denoise AI. Then I have an older computer. My issue may not be a problem with much faster computers than what I use.
I didn't see any mention of how long this takes to process one photo in LR vs. the others. Right now, I'm "enhancing" an image using the AI de-noise in LR and it's telling me it will be about 29 minutes. I've tried this on a few images so far and I'm experiencing similar wait times. So right now, this is only good (at least in LR) for the occasional photo that needs it, rather than being able to apply it to an entire photo shoot. Maybe I'm missing something.
Something doesn't sound right, unless you are testing on a very old system. I just tested it on my current machine, an i7-12700 with an Nvidia 3080 GPU, and it was averaging around 5 SECONDS a photo to denoise. I also tried it on an older machine with an i7-6700k and an AMD 6600XT and that system was still averaging 18 seconds per photo. Now I was using ACR and not LR, but I think they work exactly the same way. Also, this is with a Canon 6D which is only 20 megapixels, but still.
I use Adobe Camera RAW in Bridge instead of Lightroom. Camera RAW has been updated with the same noise reduction feature. I find that Camera RAW and Lightroom function very similarly, though I just prefer Camera RAW.
Can I ask how long it takes to denoise in LR. For me it's around 6-7 mins, compared to Topaz under 1 min. The length of time it takes to denoise in LR is a deal breaker for me.
Mine is less than 10 seconds
@@AlexArmitage Thank you, after removing some clutter from the PC I'm down to 2 mins. I think I may have more stuff to clean out.
Did you turn off sharpening on pureraw 3?
Do you mean the lens softness setting?
@@AlexArmitage yes, because the optical module on dxo does sharpen.
And some users requested to turn off on pureraw, it's available now on version 3.
The feature is awesome. But as many things with Adobe, it comes a bit short. It's "meeh". As it only works for .RAW, and you don't have any flexibility in terms of workflow. The ux is a huge fail for me. It also feels like something should've been there from the beginning, and it arrived way too late to bother. Then again, it seems to be random which software does a great job on the whole photo. Everyone has issues. My pref is ON1, even if it doesn't do great in every situation that either. It's Quick and easy, simple, bang for the bucks, and I don't use sharpening with it If I don't have to. Most important; I can choose where in my workflow I want to use it, and I can use it whenever I want, for what file I want.
Is this on the mobile version
It is not
The UA-cam algorithm recommended this episode to watch. And it was interesting, however…
Watching your episode on UA-cam, and pixel peeping between the DxO version and the LR version it’s not a night and day difference.
You had to zoom in 300% to find subtle differences. Viewing at 100% or even viewing on a tablet or mobile device then it boils down to perceived perception and wanting to find wafer thin differences.
There is one major difference though, and it is the elephant in the room. It’s the price.
You either pay a one-off fee for a licence to use the software or you pay a rolling monthly fee for years to use the software to get virtually the same results.
Lightroom is just not cost effective when it is compared to DxO and you have to put examples under a microscope to tell the difference.
DXO PhotoLab I think it's a little better then DXO PureRaw because you can change the luminance.
DXO also restores loss color.
Only thing I don't like about DXO they don't support lens profile that is in your lens.
Haven't used PhotoLab yet. Ill give it a shot at some point
Will I ditch Topaz right now?- NO. While LR's DeNoise is an improvement is will not replace all the tools I have at my disposal. Topaz solved some of the noise problems I have earlier than LR. That why I have both. I got Topaz since LR didn't have the tool good enough. The right tool to use will depend on the photo, composition, type of photo(wildlife, landscape, astro, people). As usual the answer is depend on your photographic needs and it is up to the user to ultimately decide.
So very correct! Topaz wins.
@@jonasweiss5817 Not sayingTopaz wins. Just sayin that there are several tools out there. Better to learn for yourself the capabilities of any app. What seems to me is that by the time you learn and master one application, there is an update within a short time. Some times its an improvement and sometimes maybe no so much. I have noticed that processing times for both LR and Topaz are getting longer ( I shoot Nikon Z9 in RAW so the NEF files are large.) My MacBook is about 5 years old and thinking I need to upgrade. Oh all the new AI stuff is creeping into having the engines to process them in order to minimize post processing time. But as Alex notes the comparison is for noise reduction.
You should have compared it to Topaz Photo Ai as I don't think denoise is being developed any longer
Noice
Did you check the size of the DNG ? Lightroom is even bigger than Dx0 dx. And DNG from Adobe can't be enhanced for super resolution. DxO pure raw can.
BTW, both of them work only with RAW files.
Did a deep dive, using Topaz professionally and in detailed attention to its features. Topaz won. You can't compare them. LR will have to develop photo-contingent settings and options to even compete.
topaz is not winning at anything dont kid yourself
Hi Alex. Great video extolling the virtues of Adobe Lightroom's new AI noise reduction tool. Comparisons are really good and demonstrative regarding results. I do agree with you that it is a superb enhancement in an already great piece of software but, I have to say that like almost everyone else out there reviewing this great new tool, you have completely ignored how long it takes for the software to do the job. Not everyone has the latest and greatest hardware and may be frustrated by how long it takes to process huge RAW files. Would it be possible for you to show how long it takes for the different software to do the same task please? Quality of processing results verses usability and speed of performance may well be different priorities for some people.
Alan - it didn't even cross my mind the processing times for some of these. Sadly I don't have the means to test it but I'm wondering if we could get something working in my Discord to test different times
@@AlexArmitage Thanks for your reply Alex, much appreciated. Unfortunately, I also don't have the means to test in the manner that I suggest may be of interest to others in my position. Photography is a hobby of mine but I have not invested much in PC/Laptop hardware, in comparison to camera gear and image processing software. Software like Adobe Lightroom and Topaz AI have come with upgrades and enhancements, cheeper and worthwhile upgrades to existing but old laptops like SSD have been my mantra to date instead of raw processing power. Perhaps it is time I bit the bullet and bought an up-to-date PC/Laptop to take advantage of new software. As it is, my laptop isn't good enough to be upgraded to Windows 11. All I intended to say was that a review like the one you posted, should consider more than just the end result. After all, you do include other consideration(s) like the cost of rival and complementary third party software plugins.
I just did a quick test of DxO PhotoLab 6 DeepPrime XD (my current go to) vs Lightroom on one image, so it's limited but might give you some idea. Image is a 14 bit NEF from a Nikon D7500 (20.1MP) and they're currently stored on a traditional HDD although my system drive is an SSD.
DxO: 21s
Lightroom: 58s (50s estimated)
Both set to use GPU.
System specs:
CPU: Intel i5-7600K 3.8GHz (No OC)
GPU: nVidia GTX 1060 6GB (studio driver v. 531.61)
RAM: 16GB (2 x 8GB) DDR4
OS: Windows 10 Pro (21H2)
So almost 3 times longer to process. However, if I do everything in LrC instead of using both, I'm saving time switching between programs, especially since both take significantly longer if running together. My hardware is also pretty out of date, I've been running this PC since 2016 (upgrade of CPU, GPU and RAM in 2018).
Also worth mentioning that in the case of this test image, LrC's result was marginally preferable. Not a massive difference, but comparing side by side there was a little more detail retained. Lens corrections were enabled with both (DxO set to -0.5 lens sharpness, LrC set to 60 sharpness). I need to do some more testing still, but it looks like LrC may win out IMO. Especially since this is the initial release and I read that they're hoping to remove the need to create a separate DNG in future...
@@english_electric7125 wow, thanks for that. I have just arrived back home from Konstanz, Germany with a load of photos to process. As I may have mentioned, my current laptop is old and I should upgrade. I use a Nikon D850 so raw files are huge, 45MP. I do Landscape, Nature and, when possible, astrophotography. I'll try to do some meaningful comparison between Lr and Topaz AI Denoise tools and share when I can. I appreciate your feedback, thanks!
@@AlexArmitage FYI, while in Konstanz, Germany, I visited the Dornier Museum, Friedrichshafen, a must if you're ever in the region!
I knew that I just had to wait for Lightroom to come out with their own AI and not fall for all the Topaz Ai ads which are being pushed these days. Saved myself some money by being patient!
There is no way that these small companies can compete with the sheer size and resources of a large a company like Adobe has.
I came to the same conclusion, LR is superb and better than DxO or Topaz. But, unfortunately, it so slooooooow on my computer, much slower than the other two. Really hope this will improve. Did you notice that too?
Topaz: Wake up, the easy money has gone, time to get hungry again. Whilst you are at it notice that nobody offers AI blown highlight and clipped shadow detail recovery yet....
I would love if Lightroom decoded X-Trans files better though... :( the enhanced details feature makes it better but not perfect.
They all work great, so no. Use the one you like and have!
Agreed!
I'm no expert but I'm not convinced that not sharpening the LR image at all makes if a valid comparison.
Lovely compare, just please keep the examples still for a longer time, several seconds at the least. You throw them in split seconds drum-roll as you speak, and I had personally problem with identifying anything in such short periods of time.
I completely agree with you. It wasn’t until I was done I realized I should have stopped moving around so much. I suspect it’s because moving helps me narrate quickly and I just didn’t think about it.
Your test is not comparing apples with apples.
DXO PureRaw 3 must be used with lens sharpness off (or soft, in the worse case) Anything else is oversharpening and even adding an ugly countour around the edges.
LR sharpening must be added at the end for both.
Based on my tests done this way, DXO is still a bit better than LR.
On the other hand, LR can be used on some (not all) DNGs shot with my phone, while DXO can't do that, yet.
I tried it today and I am a Topaz Denoise user. I don't see that Adobe's noise reduction is up to the task of Topaz's. Topaz has a better interface, more options and better results.
Shows an image of the bears. Then: "Things to Pay attention to are the grass hear" 😂😂😂. Not sure how many people are obsessed with looking at the surrounding than the subject.
Have you met photographers? Haha :)
I just bought pureRAW and Topaz... So I need to refund all of them))
Lightroom can deliver as good image quality as good as gold. But 68 min (!!!!!) processing time for noise reduction from an ISO 6400 photo is not acceptable. Topaz AI delivers that in less than a minute.
Mine is 5 seconds?
All the Lightroom images are far softer. If you cranked up the sharpening in the Lightroom images or simply reduced the sharpening and noise reduction on the other software images I doubt they would look much different, but that’s just my opinion, I haven’t tried the comparison myself.
Agreed 100%. When adding sharpening, artifacts and noise are often introduced or re-introduced to a denoised photo. You can’t correctly compare a denoised and sharpened photo to an only denoised photo, then point out how the sharpened one has more noise and artifacts…. Of course it does.
To qualify this, I do agree with you about ON1 No Noise AI. I have the ON1 plug in for Capture One and often find it problematic due to artefacts and over sharpening even when the sharpening is reduced. I really do like the DXO DeepPrime results though and am happy to overlook minor flaws for the very acceptable level of detail that is retained, far better than ON1. I’ve never tried Topaz so I’m not qualified to comment. I don’t use Adobe LR and doubt I will due to having many Fuji files which simply render better in Capture One, but I’m happy to go with your experience of LR’s new AI Noise Reduction and accept that it’s excellent and probably the best for LR users. Thanks for the video.
14:00 - u don't have to pay extra for it. YES I have. Paying for crappy Adobe product forever, just because currently it happens they bought a great denoiser and managed to somehow integrated into Lr (mess) in a way that currently work. For the fact it is slightly better at some cases than DxO it does not mean that someone would like to deal with Adobe software and pay for it on top of other software he uses. You said nothing about speed or performance and the fact that DxO is batch process while Lr is manual file-per-file. Lightroom is certainly not a viable option for some.
too biased towards LR,clearly can see bears hair looks like its painted i also noticed that on other videos LR vs dxovs topaz
Apples and oranges, designed to get UA-cam views and really not relevant. Having a tiny window of opportunity at the Camera Raw ingest moment only is not even remotely equivalent to the cited working solutions.
you scroll around to much & too fast! slow down so i can see what your talking about
topaz is not even good, let alone the best. it was never good for NR, even for upscaleing. Too much artefacts and oversharpening, halos, smudges, ...and SLOOOW - as you said: too fiddly, one come nowhere using it
I tried it today and I am a Topaz Denoise user. I don't see that Adobe's noise reduction is up to the task of Topaz's. Topaz has a better interface, more options and better results.
I tried to make Topaz work and it complete and utterly failed because it was so inconsistent that you couldn't trust it to not remove detail in the middle of my wildlife subjects. DxO was equally unreliable (within the same image some areas looked good but others got obliterated because the AI deemed that detail to be noise. The new LR one when applied to the problematic images was consistent throughout the frame, it left some more noise in the background (which helps a lot regarding printing the image later because it prevents posterisation) but throughout the subject it was consistent...