Just realised my mistake and changed the post. Sorry. The different sizes threw me.
itās OK - I posted where exactly the bug is in DxO code ā¦
DSC03530.ARW (13,9 MB)
Here it is. I took the photo of a copy paper in the garden with normal daylight.
Did you apply DeepPrime noise reduction?
The screen viewer displays the result of DXOās basic High Quality noise reduction. The controls for HQ do include a āMaizeā slider. Does this impact your results?
If you use DeepPrime noise reduction then this doesnāt have a āMaizeā slider as presumably the demosaic engine doesnāt need one.
I have exported jpgās from your file from DXO and C1 and canāt see the issue you show on your screenshots.
If your computer is capable of handling using DeepPrime I always recommend using the superior demosaic option available with DeepPrime.
It is interesting that older correction methods offer more options. I also find the following link interesting on the subject
Maze Slider in DXO
Do I understand correctly that the use of āMazeā is one way of dealing with the symptoms themselves and the other way is to integrate the problem solution fundamentally into the software?
I donāt know if it helps with your issue but the more advanced DeepPrime demosaic doesnāt appear to need a āMazeā slider as it appears to deal with your raw file as far as I can see, Similarly with C1.
Personally, unless your computer is not powerful enough to use DeepPrime , I canāt understand why anyone uses the old and inferior demosaicing methods. Lightroom updates its demosaic engine from time to time and I never used an older demosaic engine after a new improved version was introduced. Why would you?
following the thread ā¦ (checked with PL7.2 / Win)
with āMazeā at default
with āMazeā at custom setting
@gserim
the lens / focal length shows one hell of a light falloff
(apart from the āpicturesqueā sharpness)
and thanks for the link Maze Slider in DXO
Very strange anyway.
So without noise reduction result should be :
And this problem would be a noise problem ?
Thank you !
There is a difference when you look at the jpg export with the Prime / Maze variant compared to DeepPrime. The DeepPrime export shows errors in this area compared to the export with Prime / Maze. There should be differences in the processing.
True, the lens has its weaknesses, but I can compensate for them quite well in PL. Itās just a utility lens for point and shot. I also use the NEX-5 with a pinhole lens, so the parameters are different again. But thatās another topic :-).
it has nothing to do with any NR and no NR ( DeepPrime or XD ) will fix the maize unless DxO will fix the code to account for green channels disbalance for the camera model
this is OPās 2nd raw with āno correctionā, followed by DeepPrimeXD @ 40
the issue is that you can NOT see , try harder
if āmazeā slider equalize green channels disbalance ( or / less chances / switch to another demosaick altogether that is green channels disbalance maze-proof ) then it makes a difference ā¦ but it is a of little consolation in general ( and if āmazeā slider fights it in a different manner by blurring details then it is even worse )
PS: basically DxO might need to move āmazeā slider in a separate tool ( like āmoireā slider ) because it has nothing to do w/ any noise reduction really -OR- ( donāt they claim to actually test every camera model ? ) like other raw converters (ACR/LR, C1) just always fix the green channels disbalance behind the scenes for all camera models where it is necessary ā¦
I missed that. Is that still necessary? Where can I report the bug?
OK, cannot file a bug in the forum ā¦
go to DxO website , find support section