This page is work in progress as I update it each time I buy an update. I intend to add example photographs and revise the text as I gain more experience and use DXO PureRaw 4, 5 and 6 with photographs taken in the past with older cameras and lenses. I also intend to explore how it works with UV and NIR photographs.
Introduction
I used years ago DxO Optics Pro 11 Elite, released in 2016. Although it did a good job with image processing, I found the interface cumbersome and after some time stopped using it. I was still using Lightroom at the time. While DxO Optics Pro 11 Elite was a full photography editor, the predecessor of DXO Photolab Pro, DXO PureRaw is a raw file converter, that is normally used in combination with Adobe Lightroom, Capture One or another editor.
Some time ago, I tried DXO PureRaw 4 on some photographs taken in low light using very high ISO, and the result impressed me. Using default settings I saw some artefacts and a completely smooth “texture” even in images captured at ISO 12800. Tweaking the settings by decreasing the “Luminance” (noise reduction) and “Force detail” (sharpening/reconstruction) I obtained results that to me looked very good. There is a caveat, the processing does create some artefacts in the image, such as edge halos, but these are at a very small scale and visible only at high magnification. Consequently, even if when printed in A4 size a photograph taken at ISO 12800 may be difficult to distinguish from one taken at ISO 200, I can usually see a difference when printed at A3 size or larger sizes.
Colour profiles
At the DXO web site the importance of the output being a linear DNG file is highlighted, but not well explained. A linear DNG file is produced without applying any colour profile, while retaining the embedded colour profile from the raw file, set in camera. (It is similar to what one has to use as input, for example, to Lumariver Profile Designer when creating an ICC colour profile from scratch.) As a consequence, the effect of applying the excellent colour profiles of Capture One or user generated custom colour profiles remains (nearly?) identical as when applied directly to the ORF files used as input to DXO PureRaw. I find this to be a huge advantage, as Capture One colour profiles are well known for their quality, and because for some of the light sources I use, I have created custom colour profiles.
Lens profiles
As a user of micro four-thirds cameras and lenses I have found Capture One lens profiles to be sometimes unavailable and sometimes when available, to result in bad corrections of colour fringing at certain magnifications. In contrast DXO lens profiles in DXO PureRaw make a consistently good job, at least based on my experience. In addition, DXO has profiles for all the MFT lenses I use, both from Olympus/OM-System and from Sigma. For lenses that can be used with teleconverters, DXO provides profiles for each lens plus teleconverter combination. There are profiles even for four-thirds lenses that can be used with an adapter on the OM-1 and other MFT cameras, some of which I have used in the past. Having good lens profiles that effectively correct automatically colour fringing and other aberrations saves time when editing large sets of photographs. For example, I have found that Capture One’s lens profile for the M.Zuiko 60mm 1:2.8 Macro lens when applied to photographs taken at close distances can make the colour fringing worse than without applying a profile. To fix this problem, a correction has to be computed based on the image, and this correction based on analysing the image takes some time and is not necessarily as good as that based on a well matched lens profile usable through the whole focusing range of a lens. The profile-based corrections can be disabled individually if one desires to preserve the “character” of the lens.
DXO mentions that their lens profiles take into account variation among camera sensors. Although not explained in detail, back side illuminated (BSI) sensors behave differently than front side illuminated sensors when light impings on them at an angle away from the normal, thus the same lens can differ in vignetting properties depending on the type of camera sensor (see page Notes on the OM-1 (digital) camera and links therein).
Processing time
Processing by DXO PureRaw is slow enough that for medium and large sets of raw files (> 50-100 photographs), processing files in the background or overnight is most convenient. The largest batch I have processed over night in DXO PureRaw 4 and 5 consisted in over 1000 raw files. This worked smoothly. In fact, loading the files can be slow in PureRaw 4 (but not in PureRaw 5) if other or the same processed files are already present in the destination folder. Although I have seen this described as a bug, what seems to be happening is that DXO PureRaw checks for each raw file being added to the batch processing queue if a matching processed version of the same file already exists in the destination folder. It seems to be doing this file by file. If there are already hundreds of DNG files in the destination folder and one is adding hundreds of files to the queue, the repeated search can take a very long time. If the destination folder does not yet exist or is empty, adding the files to the queue is extremely fast in both versions 4 and 5. In this respect version 6 performs similarly as version 5.
PureRaw provides two different algorithms for processing the photographs: DeepPRIME and DeepPRIME XD2s/XD (version 4) and DeepPRIME 3 and DeepPRIME XD2s/XD (version 5). With DeepPRIME XD2s/XD, the most effective in recovering detail and slower of them, it takes approximately 26 s per ORF (Olympus OM-1, 20 Mpix) raw file in my computer, using the GPU to 95% of its capacity but making little use of the CPU, thus making concurrent use of other software possible. Even Capture One uses the GPU only in short bursts during normal editing, relying mostly on the CPU, and with enough RAM available both remain normally responsive when used concurrently. Version 6 makes available the improved algorithm DeepPrime XD3 to all cameras. It is described as recovering more details, but the difference seems subtle.
File size
The uncompressed DNG files produced by PureRaw 4, 5, and 6 are much larger than the original ORF files. In addition, it is necessary to retain the RAW (ORF for Olympus/OM-System) to be able to take advantage of future enhancements to the algorithms. Thus, the most significant new feature in PureRaw 6 is a new lossy compression option for the output that brings the size of DNG files to that of the ORF files or even smaller. At reasonable print or screen size, and without heavy-handed editing the compressed DNG files are nearly indistinguishable from the uncompressed ones.
Performance with older cameras
As I have been saving images taken with Olympus cameras as ORF raw files since around 2015, I have many files available for testing PureRaw. The cameras are E-510, EM-1, EM-1 II, and OM-1. PureRaw includes profiles for all of them and for the lenses I have used with them. However, along the way as the sensors improved and cameras implemented higher ISO values, I have used higher ISO settings when taking photographs. Thus, I have no photographs from the E-510 taken at ISO > xxxx. I do have still the other three cameras, and can take new photographs as needed for a comparison.
OM-1 ISO 12800
I edited the same ORF raw file using two workflows. 1) Conversion from ORF to linear DNG in PureRaw, followed by edit in Capture One Pro, 2) Conversion from ORF and edit in Capture One Pro, trying my best at matching the result obtained in 1). With the versions of these software available ar the end of May 2025, the resulting images were very different ?@fig-OM1 photos a) and c). PureRaw removed noise preserving details, while in Capture One one had to choose one or the other. Even using masks, and different settings for noise reduction and sharpening for subject and background I was unable to obtain a comparable output image in c). OM System’s own free software removed noise very effectively while recovering less detail than PureRaw or Capture One, but providing a natural looking and artefact-free image e) with almost no effort (I failed to adjust the white balance to match the a) and c)). Almost one year later the situation drastically changed with the just released Capture One 16.8.0 with the new “enhanced denoise” feature producing image d) which is as noiseless as with the current DxO PureRaw 6.2 b). The edits in these two new images b) and d) are copied from a) and c), respectively, except that the enhanced denoise made it possible to increase the clarity and detail settings in d).
A crop from a photograph taken under low light at ISO 12800 with the OM-1 camera handheld. Five versions, using different software for raw conversion a) DXO PureRaw 4, b) DXO PureRaw 6, c) Capture One 16.6, d) Capture One 16.8, with enhanced denoise, c) OM Workspace, with AI noise reduction.
E-M1 Mk II
Olympus E-510 and E-M1 “Mk I”
This is still work in progress and I give below only some general impressions.
To perform well, PureRaw seems to depend on RAW files with relatively high resolution. My initial tests with ORF files from the E-510 four-thirds camera are not encouraging. The E-510 was released in 2007 and has a 10 Mpix sensor. With only 10 Mpix to start with, artefacts created by the detail recovery algorithm are much more visible than with files from the OM-1.
With photographs taken with the E-M1 (“Mk I”) camera from 2013 with a 16 Mpix sensor noise reduction is usable at moderately high ISO, but results are not as good with newer cameras with more uniform and featureless noise partterns.
With photographs taken with the E-M1 Mk II camera from 2016 with a 20 Mpix sensor PureRaw works nearly as effectively as with photographs taken with the OM-1 (“Mk I”) camera from 2022 with 20 Mpx “BSI” sensor.
Although the sensors in the E-M1 Mk II and the OM-1 have the same resolution, they differ significantly in the properties of the noise. The EM-1 II at very high ISO is affected by banding and patchy noise while the noise in the OM-1 (and OM-1 II) is pattern-free and appears like random noise. This even distribution of noise seems to make noise reduction algorithms more effective and the recovery of details less prone to generate artefacts.
Conclusion
In my case, using the M.Zuiko 300mm 1:4.0 with a MC-20 2x teleconverter, DXO PureRaw allows me to take reasonably good-quality photographs of birds in low light conditions or in flight. It is quite an amazing feat of technology from both OM-System and DXO what makes it possible to use a 1200 mm-equivalent 1:8.0 lens handheld to take photographs in the dense shade of trees or when the sun is low in the sky or under dark overcast sky conditions, still obtaining high quality images at a rate of 50 frames per second.
Capture One 16.8, released in June 2026, incorporates a new noise reduction approach (“Enhanced Denoise”), based on image analysis. As it tends to be the case with other functions in Capture One, the user interface and approach are centred on usability and simplicity. I have just started using it. My first impression is that while noise reduction is as good as in PureRaw 6, detail recovery is both weaker and much less prone to generate artefacts. Having this feature well integrated into the editing workflow within a single program is a big advantage. I will keep using both Capture One and PureRaw at least for some time. Adobe has also improved noise reduction in Lightroom, however, I stopped using Lightroom several years ago.