Exploring How DxO PureRAW 5 Deals with Purple Fringing
We wanted to get back to exploring how third-party solutions like
DxO PureRAW 5 deal with
purple fringing in raw data.
The reason we look at third-party solutions in the first place is that
ColorPerfect uses point operations and point operations only. All pixels are processed individually, each governed by the same physical scaffold and rules, but without any interdependency on other pixels. Adjacent pixels play no role in the computational result for any given pixel.
That, in turn, makes the elimination of phenomena such as
chromatic aberration impossible, because we simply do not know whether a pixel should be purple by its own right or whether we are seeing an
edge effect. To resolve that, we would need a completely different approach to image processing within the plug-in. And again, if that turns out to be an area in which others already excel, why even bother redirecting our resources towards it — especially since this would require a tremendous development effort.
Anyway, I will post a screenshot of the
PureRAW 5 user interface as I had it when I made the attached DNGs, using their
new model specializing in Fuji X-Trans imaging. These are RAW images that originate from Fuji’s proprietary sensor design rather than from a standard Bayer color filter array.
When it comes to
lens and image corrections, PureRAW offers four main options. We can turn the entire lens correction system
off. We can turn
lens distortion correction on, and once we do, we can select from three possible output types:
- The image can be cropped to the original aspect ratio.
- It can be cropped to the maximum rectangle possible.
- Or we can choose to retain the complete image area.
That last option essentially yields a square image in which the four corners are filled with real image data, while a sort of “cushion effect” appears along the edges — an arched black region where no data exist. This happens because the distortion correction remaps pixels to where they would be
if there were no lens distortion, which in turn produces a non-square distribution of the pixels our square sensor originally captured.
This kind of output can be particularly interesting if we are targeting
alternative aspect ratios, such as 16:9 or even 2.35:1, thinking in filmic terms.
Now, the
fourth setting — turning the
entire lens distortion system
off — has far bigger implications for all images captured on modern cameras whose lenses are designed around
mandatory digital de-distortion.
This may sound odd at first, but it’s an important point. I wrote about it in more detail in an article discussing
shortcomings in the current and previous generations of the DNG standard, and I might as well include that text here. It isn’t published on the website yet, but you can all read it here for context, because it will give you a clearer impression of what is actually going on — both with
modern lens de-distortion in general and with
why the Adobe DNG format is no longer fully fit as an archival medium in this new world.
https://www.colorperfect.com/photo-arch ... t-changes/
That is even in 10 languages already

. So, read that first, and then keep the following in mind:
when you
turn off the lens de-distortion system in
PureRAW 5, it
completely removes all
originally embedded lens correction data that the camera manufacturer had written into the metadata.
In other words, you will end up with an image that is actually
worse off than if you had simply processed the same Fuji RAF RAW file through
MakeTIFF using
Adobe DNG Converter’s interpolation path — rather than
LibRaw’s interpolation path — because PureRAW, in this configuration, discards information that would otherwise have been preserved.
Addendum
I should probably also note that in making the four DNG examples included above, I kept both
Vignetting Correction and
Chromatic Aberration Correction enabled throughout all three examples that have the
Lens Correction System turned on.
I find that the vignetting correction, while it seems to work out quite well, makes direct comparisons to the original
MakeTIFF-derived data somewhat difficult. The
Chromatic Aberration Correction definitely improves matters — yet it does not completely eliminate
purple fringing. You can still see traces of it, for instance, in the lower right corner of the example image.