Why hasn’t the new lens distortion algorithm from the latest version of PhotoLab been applied to PureRAW?
What new lens distortion algorithm?
New distortion correction model PhotoLab 10.2 changes the way lens distortion is corrected. The new model handles rectilinear and fisheye lenses alike and fixes several long-standing issues near the image borders. For most lenses the visual result is very close to the previous model. • Images opened for the first time in PhotoLab 10.2 use the new model automatically. • Images already edited with an earlier version keep the previous model, so their rendering is unchanged. • A new “Use legacy Distortion correction” option in the Distortion palette lets you switch any image back to the previous model, or compare both models on the same image. • DxO presets now use the new model. Your own presets containing a Distortion correction are switched to it when applied.
Oh! Thank you for pointing this out! I see now that this is mentioned prominently in the PhotoLab 10.2 release notes. I wish I knew what this actually does. I knew that the 10.2 release is for ViewPoint 8 integration and some routine updates. A new distortion correction model doesn’t seem to be tied to ViewPoint - so I imagine it will make its way into PureRAW soon, much as the improvements to Lens Softness Compensation appeared first in one product and were soon added to the other as well.
Sometimes DxO informs us of these plans, but sometimes one needs to contact the support team to find out. As I recall, DxO made its intentions about lens softness compensation a lot more clear. I’m surprised that this new enhancement is appearing so quietly.
If an image comes with distortion correction from an earlier version, PL10.2 leaves it as is. To apply the new correction, uncheck the respective box in the distortion correction tool.
Comparing an export from an image and its VC (one set to old, one to new DC) does not show much difference if most settings remain at default values. With unconstrained ratio in crop and DC, images export to slightly different ratios and differences show up e.g. in Photoshop (layer and subtract) no matter what the ratio was set to.
Depending on the quality of the images, one can see a change of bulge when the classic distortion correction is switched on and off. This might also be due to changes necessary to display images of slightly different image width (or hight).
…only DxO could tell, but they’ll not do so here - or elsewhere - I presume.
And if we assume that distortion correction was perfect so far, it must be perfect now too, in a slightly different way. My guesstimation is that differences are mostly irrelevant to average human perception. Therefore, the question remains: why bother with a new distortion correction at all? Maybe the programming is simpler and more robust. maybe it opens the door to some new tech coming from camera companies (e.g. embedded correction instructions) etc.
I also assume that introducing the new DC with the new version of PR wasn’t possible due to some development delays and was therefore postponed to be introduced in PL first.
because it’s ‘new’, which means they can shout about it as a ‘new feature’ when it does reach PureRAW, which I’d guess will be the next major upgrade next spring. Of course DxO might introduce it sooner than that,
The dxo.com page is falling behind again. It still promotes PL 10.1, not 10.2. So maybe DxO is preparing a 10.2 ad package that describes the updates in more detail? The user guide already mentions the new distortion correction model but doesn’t explain it. I get the impression that it’s technically an improvement but not one that will often be obvious. One can still get pleasing results with the older model.
I’ve noticed before that some of my ultra-wide-angle photos have imperfectly-corrected distortion. DxO couldn’t improve it in the past, but now I’ll have to test those RAW images with PL10.2 and VP8. I haven’t started a trial yet. Might do so on Monday when I start a week of vacation. I’m glad DxO introduced this now.
The new distortion model uses different, and more accurate, mathematics to correct the distortion. I cannot explain the precise mathematics, but I believe it’s rooted in how they store the distortion model and thus how they calculate the transformation of each pixel.