On every other photo I see posted in social media where people used DxO PhotoLab, the photos are way oversharpened, almost smart phone level oversharpened. Likely because they leave the default “Lens Softness Correction” setting on them, which introduces tons of ringing and halos when too strong, which it is by default. This is sad because I think it may give DxO PhotoLab an undeserved bad reputation.
I also think it may give DeepPRIME a disadvantage in denoising comparisons by clueless Youtubers because they think that artifacts came from the denoising when it was infact the default over-sharpening.
I have changed Lens Softness Correction to -1 in my profile and I think it improves every photo. I rarely find myself wanting to change it. IMHO, this should be the default strength. Example:
Yes, exactly. I have the same feeling. The default value is too strong. It should be way less. I started noticing it too. It introduces artifacts, especially on the grass like in your example. It becomes almost white.
I fully agree that a default of +1 is usually too much. I like 0 with my Olympus and Panasonic lenses. I’ve seen others say that they prefer -1 for their lenses.
Gets my vote. Please be sure to vote for your own request!
FWIW, DxO received comments from users when they introduced this new default. Fortunately, it’s easy for us to make our own defaults (which I think we should be doing anyway). Similarly, I find that a denoise default of 40 is way too much except when ISO is very high or shadows are lifted extremely.
Same here. I always use 0 instead of +1 as starting point, and sometime lower it down to -1.
Really easy to modify in preset.
Nikon cameras and lenses here.
1 or zero is oversharpen… (marketing purpose?, for “wow” effect?)
This is nonsense because dxo (optical sharpness) is only supposed to correct optical defects, the low pass effect… (sharpness) which should therefore logically be non-adjustable because they are supposed to having made perfect measurements…
and should leave the management of accentuation only for “creative” purposes to the sharpness mask tool
Did I change a default preset at some point? I don’t recall doing it. This is what I see on a previously untouched photo.
But… I notice both examples shown in this thread feature foliage (or whatever you call dense wooded growth) which is the absolute best way to shine a light on over sharpening.
Try a picture of a ship or aircraft or car and you might judge it less harshly. My usual preset for “wildlife” (really anything natural) is +0.25 and for aircraft (a frequent topic for me) +1.00, though I’ve started to dial that back a bit lately.
Where I find the +0.25 becomes problematic is with things like fine grasses or other types of foliage that cover significant area.
You seem to have changed the default because the magic wand has no sparkles.
Yes, it’s most obvious in foilage but it’s obvious in other footage as well. Just look at the green mailbox at the left center on my example photos. On the middle one with default +1, it has a bright contour in the snow and a dark contour in the green. The same artifact would most likely had happened on a photograph of a plane. To me that looks awful. The bottom one with -1 setting has sharp contours that look natural. Maybe something like -0.9 or -0.8 would had looked even better, but -0.5 was too much and looked ugly and -1 is good enough. Without the filter it is much too blurry, but still better than the +1.
The aim for “Lens Softness Correction” really should be a look that is as natural as possible. Extra sharpening (for “pop” or whatever) should be added on demand by the user for artistic effect, not as a default setting.
To me, oversharpening is the photography equivalent of voice autotuning in music. I find that both effects almost always ruin the feel of the result and I’m flabbergasted that a lot of people don’t seem to notice, and some even prefer the artificial looking/sounding version.
At times, the original image isn’t big enough to print at its target resolution. Maybe it has been heavily cropped, or it was just an older camera that won’t print up to A2.
In these cases, I export at original size and then pass it to Topaz Photo AI, which handles resizing far more accurately, including any sharpening or de-blurring that might appear at the larger size.
Well yes, printing is a different thing. Suppose you are right about Topaz, which I never used.
Still up today I’m using Nik output sharpener with special settings to make up for the ‘ink jet bleeding’, which is something else.
sharpening for screen
Years ago, when running the homepage for a local photoclub, I came across all sorts of pics with no DxO OpticsPro or PhotoLab involved. And often enough I used “Adaptive Sharpening” from Nik output sharpener to fine tune the screen output. Anyone who has it available can check it out with a scale of 100%.
I think, the use of Lens Softness Correction (also) depends on the subject, the lighting – and the style of shooting.
While I’m not that concerned with landscapes, where everything is often in focus from front to back, I tend to work with a shallower depth of field. Then I don’t (or rather rarely) use global Contrast, Clear View Plus and the like, but apply a lot of things locally to guide the viewer’s eye.
But thank goodness everyone can customize the startup default to their liking.
I repeat myself but the sharpness correction which at dxo corresponds to precise scientific measurements… and should logically have a single non-adjustable value corresponding to the results of their measurement of the objective…
otherwise it’s like the competitors… looking at the screen at random, therefore subjective and no added value…
the adjustable “creative” sharpness should only be in the “sharpness mask” tool (also reserved for jpeg/tiff etc…)
It’s incomprehensible that a company that prides itself on making ultra-precise lens sharpness measurements with only one possible value should provide an adjustable tool…
the real sharpness is not the sharpening (like the others do) it is supposed to be natural and without the artifacts of the sharpening tools
clearly this is not the case and I am not the only one to see it.
the only thing I ask of such a measurement is its accuracy, its naturalness and the appropriate correction over the entire field of the lens.
Maybe if your lens does not deliver (and lot of do not - very expensive ones included - compare to what good sensors should provide), even a precise scientific aproch to improve it can’t give a unique perfect result and has to let user choose at some point.
What does not exist has to be created.
Just a thought.
yes:
but in reasonable proportions taking into account the dispersion of the components (lenses/assembly)
but the setting between -3 (zero accentuation) and +3 is a caricature
the new +1 is overemphasized with visible artifacts (as shown above): who wants that? a graphic designer?
not a photographer it seems to me
Indeed default settings are to high (since v5 or v6 I think), but it’s easy to change presets settings to get what’s intended.
I would add, I prefer parameters which allow to go to far, than parameters which don’t allow to go far enough to get intended result.
@zkarj dense “woody” images are definitely not very forgiving of over-sharpening, which is why it has been used in this topic to demonstrate the problem.
Yes indeed. The images used above were all exported as 100% JPG which I use for “experimental” purposes rather than 97% which I typically use!
I don’t use DxPL resizing at all, any resizing I do is done in FastStone which offers
I have used Lanczos2 in the past which adds sharpening so I started exporting with no DxPL sharpening at all but have now changed to Lanczos3.
DxPL offers
However, I am preparing images for viewing not printing (and then viewing) and use FSIV to create a resized library of images at 1920 x 1440, which sit on the NAS and can be accessed via phone, tablet or PC screens etc. .
Even with my MFT cameras that is not always easy to achieve and a little disappointing for someone used to taking photos with Bridge cameras.
Providing a photographer can achieve what they consider to be “correct” by selecting appropriate settings or disabling the feature completely. While others can experiment and “deviate” from the “pure” settings then surely everyone gets what they want.
DxO are creating a product to appeal to the widest possible audience, or at least they would be if they implemented a fraction of the things users have been asking for!
@JoPoV PL6 is the same as PL7 but this is PL5 when you double click on the slider
Yes. Questionable default setting choice. But at least the +1 (not 0) suggest it is already a bit more than what should generally be the best choice for people who want to get the best from their raws.
