DeepPrime and Prime noise reduction from Photolab will only give preview of a small section of the image. To see the full effect you would need to export the file and then do the comparison.
Yes, unfortunately if you don't know how the software works it makes comparisons difficult. Perhaps Peter can revisit the comparison when he understands how it works.
I worked on a set of Sony A9 images shot at 20,000 ISO. I used DxO and Topaz. Before DxO I always used Topaz Denoise AI with outstanding results on high ISO Olympus images. Since DxO Deep Prime was released there is a new king of the jungle. With the 20,000 ISO images Topaz left colour noise and artifacts on some edges. Deep Prime gave images that looked as if they had been shot at base ISO, [images were of a night soccer game under lights]. I have gone into Preferences in DxO and turned all presets off,I experience no unwanted image manipulation . I export a DNG with only the optical and Deep Prime results to Lr for comparison with the Topaz images. I still use Topaz for ISO up to 3200 and Gigapixel resize after cropping my Oly images,I resize them back to 5184x3888 px , they always resize sharper than the cropped image. Keep up the good work Peter and stay covid free.
Peter, thanks . I think your conclusion is correct - I have both Topaz DeNoise AI and DxO PL4 and shoot Olympus. I think you actually need to compare the raw images and not DNGs. Also, I don't think you were correct that you tuned off all the PL4 default adjustments. Click on the "Apply Presets" in the upper right corner and a drop down menu will you to select "No Corrections". I find DxO to be superior to Adobe Camera Raw as a raw file processor - it really is great although there is a learning curve. Also, you can begin your work in PL4 and then export it as a tif and finish working in Photoshop and or any of the Topaz plugins. Another bonus is that you also get Nik inside PL4. I have the "Elite" version of PL4 - not sure if that applies to what you showed. Have you investigated the PL4 "Clear View"? In summary, the lens correction, sharpness, noise reduction, and color controls all are superior to Adobe Camera Raw (also Lightroom). Thanks again.
Ken Weiss: Do you prefer the raw file conversion of DxO to Olympus Workspace with ORF files or are you using DxO raw conversion for the convenience/time-saving?
@@PinnaclePete I seldom use the Olympus Workspace. I find DxO PL4 raw converter and lens corrections just wonderful. Sometimes I do turn off all the defaults settings in DxO except lens correction and then do everything manually.
@@weisskm Thanks Ken. You mentioned that you can send a TIF file from PL4 to Photoshop if you wish. I guess there are still a few things (maybe a lot of things) Photoshop can do that DxO can't? (I've never any of the Adobe software products)
Thanks Peter! I use Topaz Denoise AI and my workflow: Adobe Lr -> TIFF -> Topaz Denoise AI -> JPEG. In Topaz I rarely use noise reduction above 10 points and sharpness above 15 - typically - 8 and 12. In Lr I turn OFF both Noise reduction and Sharpness and use Topaz for both these processing. First of all, I don't have color or exposure shift in image after Topaz processing. Second, Topaz has some other drawbacks you haven't mentioned: it makes areas out of focus (DoF) a bit sharper trying to recover contours and details which are naturally blurred by the optics. Sometimes Topaz spoils bokeh making it more contrast. I think artifacts are unavoidable if AI applied to any problem solving, artifacts are intrinsic to AI. But if we set noise reduction and sharpness in Olympus camera high the result in camera JPEG will be much pooper in regard of natural rendering of the subject comparing to Topaz. I have large archive of JPEGs shot by E-PL2/3 cameras 8-9 years ago and the default settings in these cameras are heavy noise reduction and high sharpness - the noise really almost out but textures and micro details too - faces have plastic like skin. Topaz Denoise AI performs much much better.
Yeah, for some reason Topaz Denoise AI has quite bad RAW conversion performance and works much better with TIFF. On the other hand DxO Photolab has always had excellent RAW conversion. So to get the best results with Denoise AI it's best to import other than RAW.
I am a little bit confused - Prime and DeepPrime in DxO PL must exported to get the results. You did not answer any of those questions in the comment, but form the way you compared the different outcomes you must have exported them. I have used Topaz Denoise for many years but the integration via tiff is awesome, must be at the end of a process, and the results are mixed depending on the motif, faces may become very exaggerated or even spooky. Now I am testing PureRaw vs Photolab for my Workflow with Capture One and again the results are mixed, some very brilliant, some unnatural like Topaz depending on the motif.
It does not make sense to export the demosaiced Raw file as DNG from LR and open it in DXO for best Raw conversion, sharpening and noise reduction. Moreover you see the DeepPrime effect only in the small window. Please check the manual...
I'm using the Topaz DeNoise AI a lot and I have never experiened any change in exposure as you got in the video. I always use it as a plugin in lightroom. Perhaps that is the difference?
The same. I use Lr+Topaz, I denoise JPEG files exported from Lr in TIFF format and then Topaz produces JPEG file and don't see any difference in color, exposure between TIFF and denoised resulting JPEG file.
I use it as a Standalone and and only have change in brightness wenn I set the Hook for the „very dark pictures“. I‘m still very happy with the results and the speed. But you have to set the hook for the GPU performance. It‘s so much faster on my Mac!
Peter, I've tried DxO on a bottom end laptop (dual core, 4Gb RAM integrated graphics card and 2.3GHz) . The Prime and DeepPrime are only applied to the whole image on export and very good but.... HQ and Prime export reasonably quick. However exporting a DeepPrime worked image took over 10min on a laptop without a substantial spec. It'll need a well spec'd PC/Laptop for numerous DeepPrime work.
I have DXO PL4 and I have been truly amazed. I have used DXO for my editing since Optics Pro 8 and PhotoLab 4 was a bigger general leap than Optics Pro 9 Elite to Optics Pro 10 Elite was when Prime NR was introduced. The DeepPRIME NR is so good that I have started using my G9 up to 6400 ISO and have far less hesitation taking out my old GX1 when I need something small.
I think if you were prepared to experiment a bit more with the parameters in Topaz Denoise then you might find it works better for some images than if you just use automatic settings. I tend to turn off the sharpen slider and do sharpening separately with Topaz Sharpen AI. For instance, on the gull picture, I am sure you could improve on the feather detail by doing the two operations separately. For most of my images which require denoising, I find Topaz Denoise works like magic. I acknowledge that Topaz software can be somewhat slow (especially Sharpen - Denoise is usually pretty quick), but the most recent versions have sped up a lot since earlier ones.
@@ForsgardPeter Your name sounds more Danish than Swedish, Peter, but the old East Denmark, Skåne, Halland and Blekinge, is currently occupied by the Swedes, so maybe your name origins from that area. Good content btw.
Dude, Photolab doesn't give you a preview of prime or DeepPrime on the image, only on the loupe within the tool. Was bizarre that you were saying it looked better when the preview doesn't change. Also, the Photolab image is darker because it automatically applies a basic profile to the image, you need to turn that off if you want to do comparisons like this.
I can't understand the logic employed in your tests. Topaz does Denoise and sharpening the combination of which impacts the apparent result. With DXO you turned off the lens sharpness control?? The default settings in DXO are designed to leave some noise, you varied the Topaz settings but appeared to leave DXO at default?
Using Topaz denoise AI software with only the remove noise not set to 0 give me very fast and excellent results on straight Olympus raw files. (corrected typo)
I checked out the demo and imported a couple of images and promptly ordered it with the Black Friday deal! Now I have the best of both worlds with Topaz being able to denoise files that Photo lab 4 does not support! 🙏🏾
Is it better to start editing your photos with Lightroom (to change the white balance, exposure, contrast etc...) then export to TIFF/DNG to Topaze AI/DXO to reduce the noise, or do the opposite: start by removing the noise via Topaze AI/ DXO by importing my RAW files, then export to TIFF/DNG to Lightroom for creative editing? Thank you
IMHO, and others may disagree, it is better to follow (at least somewhat) an order to your post-processing adjustments. I follow the "Cambridge in Colour" suggested order in post processing. www.cambridgeincolour.com/tutorials/digital-photo-editing-workflow.htm If LR is your program of choice, I would start my processing there, export to DxO/Topaz for NR, then either finish your processing in DxO or if using Topaz DeNoise, send a TIF back to Lightroom and complete your processing there.
@@ForsgardPeter If you process in LR and export as DNG then the demosaicing has been carried out in LR. A DNG is a container format, it can contain a raw file but it can also contain demosaiced data, that is a linear DNG. A linear DNG is basically a tif file without the white point being assigned. This is why colour changes occur with different software as they treat the linear DNG as a raw file and apply their own individual default process to the file which includes white balance. You can avoid this by using a tif file, which does have a white point applied, data integrity is the same at 16 bits. This can be confusing as many people think that hdr or panos created in LR means that they have a total raw workflow. Hdr or panos in LR are linear DNGs. DXO DeepPrime's noise reduction and chromatic aberration corrections are performed on the mosaiced data unlike every other raw converter and therefore can only be done with a raw (mosaic) file.
In my own tests I find that for normal ISO images up to around ISO1600, Topaz Denoise Ai’s Ai Clear function does a better job at a kind of pre-sharpen on raw files than anything else. But for my higher ISO 800 and above astrophotography raw files, I think DXO4 prime noise does a more natural looking cleanup, and other astrophotographers I know have felt the same way about that. It looks more believable with a more natural subtle grain. I haven’t compared it to some high ISO sports work yet, which I need to do. Remember when there used to be sports photography? Before the pandemic? LoL.
Thanks for the info. Unfortunately I have not had a chance to test the noise reduction in astrophotography yet. I have waited for clear skies, but I guess I have to go back to my old images.
@@ForsgardPeter I often find that the Topaz AI is not aggressive enough for my severely noisy photos - the ones I'd really want to denoise. The detail recovery option lets me push the noise reduction much further than I'd otherwise be able to in my normal processing app.
Be sure to check if your camera is supported. After downloading the trial version, I was saddened to find out my cameras were not supported. These are multiple months old, so I was very surprised. BTW, Topaz DeNoiseAi supports their raw files. Just bed sure before spending your hard earned money.
Great video Peter ! I've been using DxO for many years and this new version is far better than its predecessors. However, there is still one thing were DxO is not so good, that is sharpening (this can be seen with the first image test). And good sharpening is very important for an Olympus user, because of high resolution : RAW files of high resolution images need heavy sharpening. I use Topaz Denoise for that, because Denoise has also a very good sharpening tool. Speaking of noise reduction, you're right : I would say that DxO deep prime is very near Topaz denoise, sometimes better. Again with high res images, DxO is not the best tool when you have to use Photoshop layers for masking artefacts of moving objects. So I use DxO for every single shots, but for high res shots, the workflow of Lightroom, Photoshop, and Topaz plugin is better. Another thing is that Lightroom offers integrated HDR and Panorama fusion which work well. So I'm happy using these three tools : Lightroom+Photoshop, Topaz Denosie and DxO. One can not entirely replace the others.
One very important limitation to note is that DxO DeepPrime only works with original raw files from supported cameras. I just learned this the hard way by spending over an hour trying to convert JPG to DNG with Topaz JPEG to RAW v2.2.1 but each time PhotoLab 4 would complain that it "cannot load image data". So if you're thinking to use this app to denoise your smartphone or low-end camera photos, you're out of luck :-( If someone figures out another way to convert JPG to a compatible DNG raw format, please share with the rest of us!
For myself the DXO products works better but i have to turn down the sharpening a bit to get the look I like. The topaz product is too slow in my 2013 Mac. It may work better for others with more capable hardware
Considering one is a full editing program and the other one a standalone and then looking at the prices, DxO has more "jam" to it in my opinion. Topaz solution is still seen as one of the best, but the DxO features for also masking the noise reduction and then doing other adjustments that DxO also takes into account, DxO goes out as the winner here for me. Topaz software is not worse just because there´s a contender now, but their pricing for all their programs adds up so fast, that I´d always take the whole package over Topaz solution. The difference is so negligible that I care more about the customisation options and manual control than a 1-3% more "effective" solution, especially when the lead changes from image to image. I can always sharpen it afterwards again or change the color - which I then can´t do with saving in between and thus applying some compression again when using Topaz.
Thank you Peter, I have Photolab 4 and yes the new Prime is excellent.I was wondering how it compares to Topaz as many friends use Topaz. now I know, thank you. By the way I find DXO Photolab great for processing Olympus high resolution images (have E-M5 mkII) and it enables me to get very good images at full RAW resolution without having to resize down to the JPEG file resolution. May be you can try this and see what results you get with Olympus high resolution processed in Photolab 4. Thank you again.
I did a test I downloaded from DPreview E-M1iii jpeg and raw file and the Panasonic S1 jpeg and raw file at ISO 6,400 and the E-M1iii with Topaz DeNoise was almost as good at the Panasonic S1 with noise reduction. It was kind of like with full frame cameras when they use the same sensor and in two years they release a newer model with a little bit of improvement.
The review image in PL is only HQ. You have export to see Prime and DeepPrime results so actually not faster. That said although I own Topaz Denoise I never use it because DXO is jist simply unbeatable when it comes to the final result
I can't get very excited about all of the enthusiasm for noise reduction. With all of the technological advancements in photography over the last decade, photographers who have a need for extreme noise reduction can readily achieve what they need. Use larger sensors, native ISOs and proper exposure. I do not take my photos inside a witch's hat. My MFT equipment and my skills as a photographer allow me to produce images that are sharp as they need to be and pleasing to the eye of the beholder. For most large negative spaces in my photos, the viewer who is not pixel peeping will never be distracted by (quelle horror!) noise.
For low light wildlife photography advanced NR in post is still very much needed. You often need to use reasonably fast shutter (1/500s and above, up to ~1/2500s) to freeze the movement. And even if you have the money to get something like FF 400mm f/2,8 you might not get close enogh with that reach or need somewhat more DoF.
That's interesting question why Adobe lag behind in noise reduction? I don't think they are incapable. Traditional noise reduction by different linear and non-linear filtering (median filtering, for instance) is indiscriminate towards image's subject and therefore does not produce surprising artifacts. AI noise reduction is trained on real subjects and discriminate on subjects. Year ago I encountered very weird artifacts produced by Topaz Denoise AI like literally "black hole" in the image. These possible artifacts are completely unpredictable for the vendor itself. Adobe has very huge customer base and need to carefully read their EULA. Could be legal cases from customers facing artifacts? Having such great number of customers it'd better think twice. Non-predictable issues is the main fetter on adoption of AI in all areas, not only in image editors.
Peter, thanks a lot! It is quite interesting results. I were always thinking - Topaz Noise Reduction is sort of a gold standard for NR but looks like we have got a very good contender. Need to check it out! Thanks again! Very useful video.
It's because he used dng which does not have a white point defined. The software does that on opening the file, different software different white point applied. Does not happen with tif or jpg which have white points assigned.
Absolutely terrible review. I thought I was suffering bandwidth lag. There is a major difference between the two software programs; one is a complete editing application whilst the other is designed to deal with a single feature of the editing process. To a great extent, DxO is designed to replace or be an alternative to Lightroom whilst Topaz (in its various forms) is designed to complement it.
DeepPrime and Prime noise reduction from Photolab will only give preview of a small section of the image. To see the full effect you would need to export the file and then do the comparison.
Yes, unfortunately if you don't know how the software works it makes comparisons difficult. Perhaps Peter can revisit the comparison when he understands how it works.
I worked on a set of Sony A9 images shot at 20,000 ISO. I used DxO and Topaz. Before DxO I always used Topaz Denoise AI with outstanding results on high ISO Olympus images. Since DxO Deep Prime was released there is a new king of the jungle. With the 20,000 ISO images Topaz left colour noise and artifacts on some edges. Deep Prime gave images that looked as if they had been shot at base ISO, [images were of a night soccer game under lights]. I have gone into Preferences in DxO and turned all presets off,I experience no unwanted image manipulation . I export a DNG with only the optical and Deep Prime results to Lr for comparison with the Topaz images. I still use Topaz for ISO up to 3200 and Gigapixel resize after cropping my Oly images,I resize them back to 5184x3888 px , they always resize sharper than the cropped image. Keep up the good work Peter and stay covid free.
Thanks Peter, great job again. Your findings are exactly the same as mine. An excellent video to share in the different Olympus groups in facebook.
Many thanks!
Peter, thanks . I think your conclusion is correct - I have both Topaz DeNoise AI and DxO PL4 and shoot Olympus. I think you actually need to compare the raw images and not DNGs. Also, I don't think you were correct that you tuned off all the PL4 default adjustments. Click on the "Apply Presets" in the upper right corner and a drop down menu will you to select "No Corrections". I find DxO to be superior to Adobe Camera Raw as a raw file processor - it really is great although there is a learning curve. Also, you can begin your work in PL4 and then export it as a tif and finish working in Photoshop and or any of the Topaz plugins. Another bonus is that you also get Nik inside PL4. I have the "Elite" version of PL4 - not sure if that applies to what you showed. Have you investigated the PL4 "Clear View"? In summary, the lens correction, sharpness, noise reduction, and color controls all are superior to Adobe Camera Raw (also Lightroom). Thanks again.
Ken Weiss: Do you prefer the raw file conversion of DxO to Olympus Workspace with ORF files or are you using DxO raw conversion for the convenience/time-saving?
@@PinnaclePete I seldom use the Olympus Workspace. I find DxO PL4 raw converter and lens corrections just wonderful. Sometimes I do turn off all the defaults settings in DxO except lens correction and then do everything manually.
@@weisskm Thanks Ken. You mentioned that you can send a TIF file from PL4 to Photoshop if you wish. I guess there are still a few things (maybe a lot of things) Photoshop can do that DxO can't? (I've never any of the Adobe software products)
Thanks Peter! I use Topaz Denoise AI and my workflow: Adobe Lr -> TIFF -> Topaz Denoise AI -> JPEG. In Topaz I rarely use noise reduction above 10 points and sharpness above 15 - typically - 8 and 12. In Lr I turn OFF both Noise reduction and Sharpness and use Topaz for both these processing. First of all, I don't have color or exposure shift in image after Topaz processing. Second, Topaz has some other drawbacks you haven't mentioned: it makes areas out of focus (DoF) a bit sharper trying to recover contours and details which are naturally blurred by the optics. Sometimes Topaz spoils bokeh making it more contrast. I think artifacts are unavoidable if AI applied to any problem solving, artifacts are intrinsic to AI. But if we set noise reduction and sharpness in Olympus camera high the result in camera JPEG will be much pooper in regard of natural rendering of the subject comparing to Topaz. I have large archive of JPEGs shot by E-PL2/3 cameras 8-9 years ago and the default settings in these cameras are heavy noise reduction and high sharpness - the noise really almost out but textures and micro details too - faces have plastic like skin. Topaz Denoise AI performs much much better.
Yeah, for some reason Topaz Denoise AI has quite bad RAW conversion performance and works much better with TIFF. On the other hand DxO Photolab has always had excellent RAW conversion. So to get the best results with Denoise AI it's best to import other than RAW.
I am a little bit confused - Prime and DeepPrime in DxO PL must exported to get the results. You did not answer any of those questions in the comment, but form the way you compared the different outcomes you must have exported them.
I have used Topaz Denoise for many years but the integration via tiff is awesome, must be at the end of a process, and the results are mixed depending on the motif, faces may become very exaggerated or even spooky. Now I am testing PureRaw vs Photolab for my Workflow with Capture One and again the results are mixed, some very brilliant, some unnatural like Topaz depending on the motif.
It does not make sense to export the demosaiced Raw file as DNG from LR and open it in DXO for best Raw conversion, sharpening and noise reduction. Moreover you see the DeepPrime effect only in the small window. Please check the manual...
I'm using the Topaz DeNoise AI a lot and I have never experiened any change in exposure as you got in the video. I always use it as a plugin in lightroom. Perhaps that is the difference?
The same. I use Lr+Topaz, I denoise JPEG files exported from Lr in TIFF format and then Topaz produces JPEG file and don't see any difference in color, exposure between TIFF and denoised resulting JPEG file.
I use it as a plug-in in Affinity Photo and do see some changes in exposure and colour but nothing like as severe. Stay safe and well.
Most likely it is the way I used it as. stand alone software.
I use it as a Standalone and and only have change in brightness wenn I set the Hook for the „very dark pictures“. I‘m still very happy with the results and the speed. But you have to set the hook for the GPU performance. It‘s so much faster on my Mac!
Peter, I've tried DxO on a bottom end laptop (dual core, 4Gb RAM integrated graphics card and 2.3GHz) . The Prime and DeepPrime are only applied to the whole image on export and very good but.... HQ and Prime export reasonably quick. However exporting a DeepPrime worked image took over 10min on a laptop without a substantial spec. It'll need a well spec'd PC/Laptop for numerous DeepPrime work.
Excellent comparison, Peter. Thank you.
Many thanks!
I have DXO PL4 and I have been truly amazed. I have used DXO for my editing since Optics Pro 8 and PhotoLab 4 was a bigger general leap than Optics Pro 9 Elite to Optics Pro 10 Elite was when Prime NR was introduced. The DeepPRIME NR is so good that I have started using my G9 up to 6400 ISO and have far less hesitation taking out my old GX1 when I need something small.
I think if you were prepared to experiment a bit more with the parameters in Topaz Denoise then you might find it works better for some images than if you just use automatic settings. I tend to turn off the sharpen slider and do sharpening separately with Topaz Sharpen AI. For instance, on the gull picture, I am sure you could improve on the feather detail by doing the two operations separately. For most of my images which require denoising, I find Topaz Denoise works like magic. I acknowledge that Topaz software can be somewhat slow (especially Sharpen - Denoise is usually pretty quick), but the most recent versions have sped up a lot since earlier ones.
Hvordan kommer en finne til at hedde Peter Forsgård?
Ancestors are from Sweden and Finland used to be part of Sweden.
@@ForsgardPeter Your name sounds more Danish than Swedish, Peter, but the old East Denmark, Skåne, Halland and Blekinge, is currently occupied by the Swedes, so maybe your name origins from that area. Good content btw.
Dude, Photolab doesn't give you a preview of prime or DeepPrime on the image, only on the loupe within the tool. Was bizarre that you were saying it looked better when the preview doesn't change. Also, the Photolab image is darker because it automatically applies a basic profile to the image, you need to turn that off if you want to do comparisons like this.
DxO seems to apply a different color profile to the DNG file. Maybe try exporting in TIFF
Yes it does a bit. But I like the slightly warmer colors. It can then made a little cooler if needed.
I can't understand the logic employed in your tests. Topaz does Denoise and sharpening the combination of which impacts the apparent result. With DXO you turned off the lens sharpness control??
The default settings in DXO are designed to leave some noise, you varied the Topaz settings but appeared to leave DXO at default?
Using Topaz denoise AI software with only the remove noise not set to 0 give me very fast and excellent results on straight Olympus raw files.
(corrected typo)
I use both and I find that they both work about as good.
Could you do a comparison between these 2 software and the denoise tool in Luminar 4 ?
Yes I could. Lets if can do that in the near future.
I checked out the demo and imported a couple of images and promptly ordered it with the Black Friday deal! Now I have the best of both worlds with Topaz being able to denoise files that Photo lab 4 does not support! 🙏🏾
Is it better to start editing your photos with Lightroom (to change the white balance, exposure, contrast etc...) then export to TIFF/DNG to Topaze AI/DXO to reduce the noise, or do the opposite: start by removing the noise via Topaze AI/ DXO by importing my RAW files, then export to TIFF/DNG to Lightroom for creative editing?
Thank you
IMHO, and others may disagree, it is better to follow (at least somewhat) an order to your post-processing adjustments. I follow the "Cambridge in Colour" suggested order in post processing. www.cambridgeincolour.com/tutorials/digital-photo-editing-workflow.htm
If LR is your program of choice, I would start my processing there, export to DxO/Topaz for NR, then either finish your processing in DxO or if using Topaz DeNoise, send a TIF back to Lightroom and complete your processing there.
That is a good workflow.
@@ForsgardPeter If you process in LR and export as DNG then the demosaicing has been carried out in LR. A DNG is a container format, it can contain a raw file but it can also contain demosaiced data, that is a linear DNG. A linear DNG is basically a tif file without the white point being assigned. This is why colour changes occur with different software as they treat the linear DNG as a raw file and apply their own individual default process to the file which includes white balance. You can avoid this by using a tif file, which does have a white point applied, data integrity is the same at 16 bits. This can be confusing as many people think that hdr or panos created in LR means that they have a total raw workflow. Hdr or panos in LR are linear DNGs.
DXO DeepPrime's noise reduction and chromatic aberration corrections are performed on the mosaiced data unlike every other raw converter and therefore can only be done with a raw (mosaic) file.
why not the same white balance ? Excellent video by the way :)
Thanks.
A raw file or a linear dng file does not have a white point assigned. Therefore the software assigns whatever white point it uses by default.
DXO denoise works only on raw while topaz denoise can work on jpeg，，right？
Yes. I should have said that!
brilliant, as always
In my own tests I find that for normal ISO images up to around ISO1600, Topaz Denoise Ai’s Ai Clear function does a better job at a kind of pre-sharpen on raw files than anything else. But for my higher ISO 800 and above astrophotography raw files, I think DXO4 prime noise does a more natural looking cleanup, and other astrophotographers I know have felt the same way about that. It looks more believable with a more natural subtle grain. I haven’t compared it to some high ISO sports work yet, which I need to do. Remember when there used to be sports photography? Before the pandemic? LoL.
Thanks for the info. Unfortunately I have not had a chance to test the noise reduction in astrophotography yet. I have waited for clear skies, but I guess I have to go back to my old images.
Hi Peter, did you not use any detail recovery in Topaz? That was the most attractive feature of the app for me.
To be honest not in all images. Maybe I should check that option too.
@@ForsgardPeter I often find that the Topaz AI is not aggressive enough for my severely noisy photos - the ones I'd really want to denoise. The detail recovery option lets me push the noise reduction much further than I'd otherwise be able to in my normal processing app.
Good analysis.
Fin on a whiskey tour of Scotland = photos with screwed up exposure the next day ;-)
Thank you for the wonderful tutorial, very useful as it happens several times to have photos in these conditions, greetings Fabio Torino
Glad it was helpful!
Be sure to check if your camera is supported. After downloading the trial version, I was saddened to find out my cameras were not supported. These are multiple months old, so I was very surprised. BTW, Topaz DeNoiseAi supports their raw files. Just bed sure before spending your hard earned money.
Good points.
Great video Peter ! I've been using DxO for many years and this new version is far better than its predecessors. However, there is still one thing were DxO is not so good, that is sharpening (this can be seen with the first image test). And good sharpening is very important for an Olympus user, because of high resolution : RAW files of high resolution images need heavy sharpening. I use Topaz Denoise for that, because Denoise has also a very good sharpening tool. Speaking of noise reduction, you're right : I would say that DxO deep prime is very near Topaz denoise, sometimes better. Again with high res images, DxO is not the best tool when you have to use Photoshop layers for masking artefacts of moving objects. So I use DxO for every single shots, but for high res shots, the workflow of Lightroom, Photoshop, and Topaz plugin is better. Another thing is that Lightroom offers integrated HDR and Panorama fusion which work well. So I'm happy using these three tools : Lightroom+Photoshop, Topaz Denosie and DxO. One can not entirely replace the others.
One very important limitation to note is that DxO DeepPrime only works with original raw files from supported cameras. I just learned this the hard way by spending over an hour trying to convert JPG to DNG with Topaz JPEG to RAW v2.2.1 but each time PhotoLab 4 would complain that it "cannot load image data". So if you're thinking to use this app to denoise your smartphone or low-end camera photos, you're out of luck :-( If someone figures out another way to convert JPG to a compatible DNG raw format, please share with the rest of us!
True and should have mentioned it.
For myself the DXO products works better but i have to turn down the sharpening a bit to get the look I like. The topaz product is too slow in my 2013 Mac. It may work better for others with more capable hardware
Considering one is a full editing program and the other one a standalone and then looking at the prices, DxO has more "jam" to it in my opinion. Topaz solution is still seen as one of the best, but the DxO features for also masking the noise reduction and then doing other adjustments that DxO also takes into account, DxO goes out as the winner here for me.
Topaz software is not worse just because there´s a contender now, but their pricing for all their programs adds up so fast, that I´d always take the whole package over Topaz solution. The difference is so negligible that I care more about the customisation options and manual control than a 1-3% more "effective" solution, especially when the lead changes from image to image.
I can always sharpen it afterwards again or change the color - which I then can´t do with saving in between and thus applying some compression again when using Topaz.
Thank you Peter, I have Photolab 4 and yes the new Prime is excellent.I was wondering how it compares to Topaz as many friends use Topaz. now I know, thank you. By the way I find DXO Photolab great for processing Olympus high resolution images (have E-M5 mkII) and it enables me to get very good images at full RAW resolution without having to resize down to the JPEG file resolution. May be you can try this and see what results you get with Olympus high resolution processed in Photolab 4. Thank you again.
Good idea about the high res files and DxO. I will give it a try.
I did a test I downloaded from DPreview E-M1iii jpeg and raw file and the Panasonic S1 jpeg and raw file at ISO 6,400 and the E-M1iii with Topaz DeNoise was almost as good at the Panasonic S1 with noise reduction. It was kind of like with full frame cameras when they use the same sensor and in two years they release a newer model with a little bit of improvement.
The review image in PL is only HQ. You have export to see Prime and DeepPrime results so actually not faster. That said although I own Topaz Denoise I never use it because DXO is jist simply unbeatable when it comes to the final result
I can't get very excited about all of the enthusiasm for noise reduction. With all of the technological advancements in photography over the last decade, photographers who have a need for extreme noise reduction can readily achieve what they need. Use larger sensors, native ISOs and proper exposure. I do not take my photos inside a witch's hat. My MFT equipment and my skills as a photographer allow me to produce images that are sharp as they need to be and pleasing to the eye of the beholder. For most large negative spaces in my photos, the viewer who is not pixel peeping will never be distracted by (quelle horror!) noise.
For low light wildlife photography advanced NR in post is still very much needed. You often need to use reasonably fast shutter (1/500s and above, up to ~1/2500s) to freeze the movement. And even if you have the money to get something like FF 400mm f/2,8 you might not get close enogh with that reach or need somewhat more DoF.
I wonder why Adobe is not working on a better Noise Reduction for Lightroom
May be they think that AI noise reduction is controversial.
That's interesting question why Adobe lag behind in noise reduction? I don't think they are incapable. Traditional noise reduction by different linear and non-linear filtering (median filtering, for instance) is indiscriminate towards image's subject and therefore does not produce surprising artifacts. AI noise reduction is trained on real subjects and discriminate on subjects. Year ago I encountered very weird artifacts produced by Topaz Denoise AI like literally "black hole" in the image. These possible artifacts are completely unpredictable for the vendor itself. Adobe has very huge customer base and need to carefully read their EULA. Could be legal cases from customers facing artifacts? Having such great number of customers it'd better think twice. Non-predictable issues is the main fetter on adoption of AI in all areas, not only in image editors.
Peter, thanks a lot! It is quite interesting results. I were always thinking - Topaz Noise Reduction is sort of a gold standard for NR but looks like we have got a very good contender. Need to check it out! Thanks again! Very useful video.
Topaz DeNoise Ai has been, but DxO PhotoLab 4 and its Deep Prime seems to be better in most cases.
DeNoise AI was sharpening your photo. Photolab 4 did not. If you set photolab to 200 in sharpening I think there will be a very different result.
Topaz denoise NEVER lightens my pics like it does yours. I think you need to check your settings.
It's because he used dng which does not have a white point defined. The software does that on opening the file, different software different white point applied. Does not happen with tif or jpg which have white points assigned.
TOPAZ DENOISE.
Imforvative
Absolutely terrible review. I thought I was suffering bandwidth lag. There is a major difference between the two software programs; one is a complete editing application whilst the other is designed to deal with a single feature of the editing process. To a great extent, DxO is designed to replace or be an alternative to Lightroom whilst Topaz (in its various forms) is designed to complement it.