1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Dx0 PureRAW Sofware Test

Discussion in 'AP Magazine Feedback & Suggestions' started by Learning, May 24, 2021.

  1. Learning

    Learning Ethelred the Ill-Named

    I liked the software test in the latest issue (cover date 29 May 2021). I would like to ask Andy Westlake if he tried passing the intermediate linear DNG files into editors other than Photoshop. My own interest is in how it might work with Affinity Photo and I can imagine some users of Skylum Luminar might also be interested.
  2. Andy Westlake

    Andy Westlake AP Staff

    Thanks, I'm pleased you enjoyed my review and found it useful.

    I'm afraid I didn't have time to test the DNG files from PureRAW with software other than Adobe. Your best bet really is to download the free trial version from www.dxo.com and try it out to see how well it works for you. After all, every user has their own needs and expectations. How much benefit you might get depends on the cameras and lenses you use, and the way you evaluate and output your images.
    Last edited: May 24, 2021
  3. Learning

    Learning Ethelred the Ill-Named

    Thanks Andy. I thought that I would ask before possibly wasting time on it. I will do as you suggest and do my own trial with PureRAW and Affinity.
    I should have something to report by tomorrow unless the weather improves.
    Andy Westlake likes this.
  4. PeteRob

    PeteRob Well-Known Member

    From what you wrote I think it should work with any other software that supports DNG.

    In comparing with Adobe did you turn on the enhance details (now called raw details) switch? This also uses an AI approach to reducing noise, improving edge definition and reducing artefacts. I think it also produces DNG output.

    At the moment there is a slightly different implementation in ACR compared to Lightroom but that I think that only applies to a resize algorithm which is something extra again.
  5. Andy Westlake

    Andy Westlake AP Staff

    Yes this is absolutely correct, PureRAW offers the some potential benefits regardless of the raw converter you use. But how much you gain in practice depends on how DxO's lens corrections and noise reduction compares to the program you normally use.

    I did quite a lot of work looking at how PureRaw compares to, and interacts with Adobe Super Resolution, which is effectively the latest iteration of Enhance Details. The two do different things; PureRaw is great at lens corrections and noise reduction, while Enhance Details / Super Resolution is impressive at interpreting fine pixel-level detail. But if you feed it noisy, soft images, all it really does is make larger noisy, soft images.

    This means that rather than the two being competing ways of doing the same thing, they actually complement each other quite nicely. DxO's AI in DeepPRiME will do a really nice job of cleaning up the raw data, then Adobe's AI detail enhancement goes a step further and gets the most out of DxO's DNGs.
  6. PeteRob

    PeteRob Well-Known Member

    Thanks Andy, I’d only played with the enhanced (raw) details in LR and once with the Super Resolution option in ACR (not yet in LR). They both lay claim to using AI which is why I though there may be parallels wrt noise reduction in raw details. Super Resolution is something else. I’ve got a feeling that Canon also have something similar for their raw files only but not with a DNG output. Now I have a mix of Fuji X and Canon I stick to LR for raw conversion.
  7. Learning

    Learning Ethelred the Ill-Named

    I have tried Dx0-PureRAW on decently exposed NEF files and then developed the the DNG files in Affinity to produce afphoto files.
    I also developed the NEF files in Affinity Photo to afphoto files. The files superficially looked similar and both needed adjustments to lift them. Looking closer at a magnified image the Dx0 route seemed to be more detailed.
    Next I used Nikon Studio to process the NEF files to 16bit tiff and then used Affinity photo to convert the tiff to afphoto. This gave a result that reduced the need to adjust exposure and curves very much. This was the easiest route to get the overall effect although for the best sharpness and freedom from noise the Dx0 route was better. I haven't printed any of the files. If I was making big prints I think that the Dx0 route would be beneficial.
    Finally there is a recently known problem of reading linear DNG into Affinity photo if the image contains areas in which the highlights are burnt out. The blown areas take on a violet/pink colour cast.
    I reproduced that fault on a couple of files which really weren't worth processing by anything. It could matter if only a small highlight triggered the effect and that the picture was salvageable.
    It seems that this effect only affects Affinity and looking at forum entries on both the Dx0 and Affinity sites I would say that the ball is in the Affinity half of the court.

    These opinions are based only on about 20 files which I had in a camera plus the two badly overexposed files which I had not bothered to delete.
    I might investigate further.
  8. Bazarchie

    Bazarchie Well-Known Member

    Did you like the results Pete?

    At present I am using Topaz Denoise AI which I find is very good when used with Lightroom.

    I will give DxO pure raw a trial.
  9. PeteRob

    PeteRob Well-Known Member

    I can’t fully recall. I think enhance details worked well with the 5D files but go to higher pixel densities then the benefits are less clear. I’ve always been happy with what LR does. However, if I go back to something ancient and reprocess it because I think it can be better, it is difficult to pick out if it is the better software available now or because I’ve had 12 years more practice at raw conversion. I started preparing some super-resolution versions of very hard crops. I got tangled in exporting edits from LR to ACR to use super resolution and then returning the interpolated DNG but I didn’t get as far as actually printing the results which is the way to test. Maybe I should revisit.
    Bazarchie likes this.
  10. Chester AP

    Chester AP Well-Known Member

    Having read the review, and the above, can I ask Andy another question?

    I currently use the Adobe RAW converter in my 2009 copy of Photoshop Elements 7 (my Pentax creates DNG RAW files so I have never been forced to buy a newer version). For prints up to 50 x 75 cm, how likely am I to notice any improvement if I run the DNG RAW file through DxO PureRAW before opening it with the RAW converter in PE7?

    For me, the main benefits would be reduced noise in dark areas and the ability to easily remove the annoying colour fringes that were never a problem with film (I have some old 24 x 36 inch Cibachrome prints that demonstrate this). I can reduce both of these 'manually' in PE7, but it involves some trial and error.
  11. PeteRob

    PeteRob Well-Known Member

    Too late to edit the last post. Not sure that this is in anyway useful. So the lifetime of this post might be short.

    The original is 900x900 pixel cropped from 4896 x 3264 and this was super-resolutioned and recropped to approximately the same proportion 1777x1777 out of 9792 x 6528. Both then sized to 800x800 to put on here.

    [​IMG]BV9R0681-Enhanced.jpg by Pete, on Flickr

    [​IMG]BV9R0681.jpg by Pete, on Flickr
  12. Andy Westlake

    Andy Westlake AP Staff

    Sorry, I only just spotted this. If you haven't already tried PureRAW, these are two things it should be good at, just as long as your camera and lenses have been profiled by DxO. So it's definitely worth downloading the trial version and giving it a go.
  13. Chester AP

    Chester AP Well-Known Member

    Thank you for the reply - I will investigate further. I have worked out how to reduce colour fringes in Elements, and the results can look quite good when printed, but it is not a quick automated process.
  14. Chester AP

    Chester AP Well-Known Member

    DxO Pure RAW testing with trial version.

    The software cannot be persuaded to attempt any lens correction for lenses not in the DxO database (I was expecting some generic options based on focal length, or perhaps low/medium/high, etc.). So the colour fringes on shots taken with my Tokina 80-400 remain, whilst those on my Tamron 18-250, Tamron 90 macro (20 + years old) and Sigma 10-20 are reduced. However, some shots appear to have suffered an adjustment to brightness (most noticeably darker shots with the 10-20 made lighter).

    I applied lens correction if available, and created a DNG file of each of the 3 types offered. I then opened each of these DNG files in PE7, made no changes of any kind to the DNG file, and then applied 25% sharpening at 1 pixel radius before saving as a JPG to compare with the original unprocessed JPG from the camera. So every DxO DNG files was processed in the same way when producing a JPG for saving.

    I then compared noise reduction results on these JPG files with the original unprocessed JPG files from the camera body. Some were old ones taken with a Pentax K10 (10 Mp) and some more recent with a K5 (16 Mp).

    The 'HQ' noise reduction appears to be no better than the original JPG, the 'Prime' better and the 'Deep Prime' better still. On my PC 'Deep Prime' takes about 4 minutes to process a 16 megapixel RAW file of 26-30 Mb in size.

    Buying the software will require some consideration. If it was £30 I would buy it now, but £115? It has failed me on the lens that has the worst colour fringes because it's not recognised by the software, and with more modern lenses image noise has not been a problem on images printed as large as 50 x 75 cm. The results of DxO colour fringe 'removal' look remarkably similar to results I have achieved 'manually' careful use of PE7's 'replace colour' facility (where I can even apply it to only a selected area of the image to reduce the risk of affecting similar colours in the rest of the shot).

    Has anybody else tried this software?
  15. Learning

    Learning Ethelred the Ill-Named

    Thanks again for drawing my attention to this software. It works. It is also not compatible with my present preferred workflow. I have uninstalled the free test software.
    For anyone using the Adobe ACR or Affinity raw engine I am sure that this DXO software may be useful. It is worth assessing.

    I am a Nikon user and do my raw processing with Nikon Studio. I previously used NX-D. Sure these are a bit clumsy but do an extremely good job of processing raw files to 16bit tiff. I then edit in Affinity Photo, a poor person's Photoshop, and irrespective of price is very good. Nikon is every bit as good, or even better, as DxO at processing NEF files although they don't do this super sharpening magic. Nikon do their own lens corrections as well as dXO.

    I don't consider my time wasted. It is good to know what is possible even if I do not have need to use it.
  16. Chester AP

    Chester AP Well-Known Member

    I have done some more testing, this time looking only at the original unprocessed RAW file and the Deep Prime RAW file, using the Windows 'Photos' software to view them. This converts them to JPG files for display, but hopefully does no further processing.

    The Deep Prime processing reduces colour fringing, although I have found an example where it occurs at the corner of an image on an out of focus area and there is no obvious change. Perhaps it only works on areas with clearly defined 'edges'. What I have noticed is the correction of barrel and pincushion distortion (subtle on a Sigma 10-20 and more obvious at the wider end of a Tamron 18-250, although the result in no better than I would get from doing it 'manually' in PE 7). With these there has sometimes been some change to colour balance, the intensity of some colours and contrast which I didn't expect.

    The Deep Prime RAW files are consistently between 2 and 3 times the size of the original RAW file, in one case resulting in a 41 Mb RAW file from a an old 10 megapixel DSLR.
  17. Bazarchie

    Bazarchie Well-Known Member

    I downloaded the free test version. I need more testing but I cannot see it being part of my workflow. Lightroom and Topaz Denoise when necessary works for me. Just wish I could understand PS.
  18. PeteRob

    PeteRob Well-Known Member

    I’d guess that is because the DNG files are close to Tiff format. 10 MP x 3 colours x 2 bytes* per colour adds up. If it’s “only” 41 MB some compression must be done.

    For an old camera this could be 12/16 x 2 if lossless compression is used.
  19. Chester AP

    Chester AP Well-Known Member

    Some examples (no option to compress RAW files when saved by the camera body):
    10Mp sensor - original DNG from camera was 16.1Mb, Prime DNG was 37.2Mb and Deep Prime was 37.0Mb
    16Mp sensor - original DNG from camera was 29.0Mb, Prime DNG was 76.7Mb and Deep Prime was 76.2Mb

    The result of all my testing is that I will wait and see what happens to the price.
  20. Chester AP

    Chester AP Well-Known Member

    I have now discovered that my old-model Sigma 17-70 (I believe introduced c. 2008 and replaced c. 2017) is also not supported by DxO. I was offered the chance to use the settings for a Sigma 18-50, which produced some optical corrections when used for processing. Yet my Tamron 90 mm, which is much older, is supported. However, the Sigma 17-70 has proved to be such a decent and versatile lens that I probably use it for at least half the shots I take.

    Does anybody know if DxO continue to add older lenses to their database?

Share This Page