Distortion Type value not retained in display

#1

I see there’s a recently released new revision of PL but I generally wait a bit for others to find the new problems before I install. So I haven’t installed it yet.
My version/revision is v2.1.0 build 23440 Elite Windows 10
Camera: Lumix FZ1000

It looked as though lens correction for the focal length I used might be incorrect according to the results from different photo processing software (in any case, I happened to like the proportions of the other software’s output). So I changed to manual distortion correction in PL and got close to matching the other software’s output by changing the pincushion to 41. “Keep aspect ration” is enabled. Closed PL and reopened it later (almost certain I had closed and then reopened – but it might have been left open).

When reopened, I saw that the pincushion setting was 41, but then I changed the distortion type to look at the other types – which were displaying values of ‘0’. When I returned to pincushion type, it too now displayed ‘0’ instead of ‘41’. But the photo distortion didn’t change from when I first opened it with ‘41’ displayed.

I wanted to record all distortion type values by reviewing them. But it seems the values revert to 0 when switching between distortion types rather than reloading the values which had been set.

Apparently, the software understands what the manual settings are without displaying them properly (if a new distortion type is selected). And because the values I selected don’t get retained in the DOP file, it’s essentially impossible to record my choices unless I remember to write them down.

Has this been corrected in the latest release? If not, it would be nice if you could remedy that so there’s a way to review and record the chosen settings.

Can I also ask this here?.. While I’m at it, I’m suspicious of whether the lens correction (at certain focal lengths) for my fixed Lumix zoom lens is even close when looking at some photos. I see that I can change the focal length in PL, but when I tried that (both in Auto and Manual correction) nothing seemed to change. Is that normal? And, if I find that the lens correction is way off, is there a file I can edit to somehow force a particular correction dependent on the focal lenght? I know to shoot a grid pattern of parallel and vertical lines to see if the proper correction is being applied.

Thank you.

0 Likes

(Svetlana G.) #2

Hello @BBT,

  • Well, actually it’s the expected behavior. And if you want to see different distortion types for comparison you can work on the virtual copies. In the meantime we can ask @fdeitos if your request is possible for the implementation.
  • Could you, please, provide us with a sample to have a look? You can upload it here under your name upload.dxo.com and let me know when ready.

Thank you
Regards,
Svetlana G.

0 Likes

#3

I’ve taken some photos of grid patterns, and although I can see that I need some better grid images to better verify that auto-correction is applied properly, it does look as though DxO is getting the auto-correction done properly. I’ll report back if I see a problem once I get better grid images. I reported concern in my original post because other software which supposedly uses auto lens correction created a different image (which I actually thought had the proportions more correct, but couldn’t confirm).

Also, as long as I have lens correction set to ‘Auto’, I DO see slight correction differences as I change the ‘Focusing Distance’ slider setting.

But concerning the “expected behavior” of having the user’s intensity value be reset to ‘0’ when selecting a different distortion type… seems it shouldn’t be expected to behave that way by the user. If I enter a value for a distortion type, I’d expect that same value to return back once I’ve selected that same distortion type. As I had wondered before I first posted about this, apparently each distortion type is considered to be a mutually exclusive setting (2 of the 3 types will always be ‘0’).

Perhaps it’s not unusual for photo software to change a user setting when selecting a different attribute to inspect (I haven’t used many different photo processors), but I haven’t run into much software that I recall which blew away a user setting just because the user wanted to examine a different attribute. I do understand why it’s working the way it is presently, though – being that each distortion type is mutually exclusive of the others apparently. But definitely it’s not intuitive that previous settings won’t hold and will be lost.

Although I appreciate the difficulty in solving this, perhaps the easiest solution might be to ask the user upon attempting to choose a different distortion type whether it’s okay to reset their present distortion intensity value back to zero if it was set to any value other than zero (and in the same dialog, inform the user that only one of the three types can be a value other than zero – that only one distortion type can be applied at a time).

Thanks for your prompt reply to my original questions.

0 Likes

(Svetlana G.) #4

Hello,

Thank you for the detailed feedback.

  • It would be good if you find the image where you think the correction is not ok and provide us with the original + sidecar in order we could investigate the issue.

Regards,
Svetlana G.

0 Likes

#5

I uploaded the following (using ?BBT as support#):
Original RAW
DxO PL sidecar
JPG output from my other photo processor
JPG output from DxO PL with pincushion set to 41 to approximately match the other photo processor output
JPG output from DxO PL with auto lens correction

I’ve noticed at least one other that seems more appropriate when output from my other photo processor rather than from DxO PL. I had to change to manual in PL, except I see I had used fisheye 18 which got it very close to the other processor’s lens correction which I preferred (model looks slimmer and more naturally proportioned to my eye – this model is tall and thin in real life).

I had edited the other processing software’s lens correction file to include lens parameters I copied from a file I downloaded from Lensfun (the default software package didn’t include lens correction values for my camera). Because I did this for the my other software, wondering if I can do something like that for DxO too.

And I uploaded just the JPGs of another shot with a model laying on a rock. The filenames explain the process (auto lens correction and fisheye 18 which matches the other software output provided). You can see the wider shouldered one doesn’t seem quite as correct as the narrow shoulders. And, again, my other software creates the narrower shoulders. I didn’t include the original for this one because I feel a little uncomfortable passing the original of his one around for security sake.

I realize there’s a chance that I just like the proportions better of the manually adjusted ones even though the auto might be more correct (but auto doesn’t agree with my other software output).

It would be appreciated if you’re able to determine what the issue is using these photos. Thank you.

0 Likes

#6

And I just uploaded two photos of the same building.
One is with no zoom @ 9.12mm focal length
The other with some zoom @ 26.28mm focal length

Both taken standing in the identical location.
Both using DxO PL auto lens correction mode (both @100 intensity)
Both output to JPG using my same default Export to Disk settings
I cropped the no-zoom to align over the zoomed photo.
The center of cropped portion I’m providing is essentially from the center of the full frame.

Note: the no-zoom photo has got the same width but it is squashed in height (like I thought same as with the models)

But of course, for now, I can’t be certain which is the correct one – but I strongly suspect the 9.12mm is not being corrected properly by PL.

0 Likes

#7

And, uploaded 3 more JPG of a grid structure that I believe has square openings (but would have to go back to measure to be sure). One is focal length 9.12 and this seems squashed vertically, whereas the other two focal lengths of 15.26 and 24.66mm are essentially identical and openings measure close to square (cropped to align with the 9.12).

0 Likes

(Svetlana G.) #8

Good morning,

Thank you for the images. We’ll investigate them.

P.S. You should not worry about the security, we use the images provided by the users only internally while doing the investigation.

Regards,
Svetlana G.

0 Likes

#9

Any advice for how to quickly (hopefully automatically) achieve the proper lens distortion correction?
It appears that PL has something significant wrong with at least one focal length.
Other software gets it more correct so I don’t think it’s my specific camera that has an issue.
If you’d like to continue this by email, that’s fine.

0 Likes

(Svetlana G.) #10

Hello,

The image data team confirmed that there is an issue with the images you provided and they will work on the fix. In the meantime there are 2 ways: correct Distortion manually for each image or create a preset and apply it.

Regards,
Svetlana G.

0 Likes

#11

These questions concern the issue you’ve verified about Photo Lab’s incorrect lens correction for some focal lengths of the built-in zoom lens for bridge camera Lumix FZ1000.

If it isn’t already remedied in this latest release, are you able to say when that might happen?
And, when remedied, how will I know that it’s remedied aside from some kind of test of my own?
Will any remedy only be applied to photos that have not been opened yet in PL?
What about photos I’ve edited but didn’t manually correct the distortion because I didn’t notice it was wrong? The crop I did for some might change if remedy is automatically applied
Did you find that there are other faulty distortion corrections aside from @9.12mm? (I ask because it might be that I’m not noticing something that technically should be manually corrected at other at focal positions of my zoom lens).
Above what focal length can I assume PL’s correction is done appropriately?
Also, I imagine there’s some interpolation between correction points. So I can’t be sure (unless you tell me) which focal lengths should get manual correction (to perhaps a lessor degree as I move away from 9.12mm). At 9.12mm, if I set Manual Correction Barrel 100, that’s all that’s needed apparently. But I wonder about other focal lengths, and of course I’d rather not have to remember to do this manually.

Is there a way I can do something like I did with AfterShot Pro where I was able to find a way to use correction values from LensFun – is there a file I can edit? I’m an engineer by the way, software and computer literate. I’ve found the user.config file but that looks like it deals essentially only with the UI. I’m guessing the lens correction is held in the database.dll files unfortunately.

Please see my Reply following this post to see XML that seemed to work acceptably for AfterShot 3 Pro (altho not perfectly identical to PL correcton [at focal lengths where it seems to work properly in PL]).

Thank you for your help with this.

0 Likes

#12

I couldn’t post the XML directly so I’ll attempt again…
?xml version=“1.0” encoding=“UTF-8”?
LenseDb
Lens skip=“false” raw=“true”
ModelStandard/Model
Mountpanasonicfz1000/Mount
FocalLength max=“146” min=“9.1”/
Aperture0/Aperture
CropMultiplier2.73/CropMultiplier
Converter factor=“1” detected=“false”/
CorrectionCoefficients
RadialDistortion b="-0.05809" focal=“9.1” a=“0.01858” c="-0.06581"/
RadialDistortion b="-0.0582258" focal=“9.12” a=“0.0185912” c="-0.0652072"/
RadialDistortion b="-0.06488" focal=“10.1” a=“0.01914” c="-0.03567"/
RadialDistortion b="-0.047" focal=“11.1” a=“0.007” c=“0.005”/
RadialDistortion b="-0.0477" focal=“12.6” a=“0.01159” c="-0.00043"/
RadialDistortion b="-0.0259411" focal=“15.76” a=“0.00504993” c="-0.000187357"/
RadialDistortion b="-0.00914" focal=“18.2” a=“0” c=“0”/
RadialDistortion b="-0.00224" focal=“26.3” a=“0” c=“0”/
RadialDistortion b=“0.00035” focal=“65.8” a=“0” c=“0”/
RadialDistortion b=“0.00049” focal=“109.3” a=“0” c=“0”/
RadialDistortion b="-0.00305" focal=“146” a=“0” c=“0”/
Vignetting focal=“9.12” strength=“0.5” radius=“50”/
Vignetting focal=“11.1” strength=“0.5” radius=“50”/
Vignetting focal=“15.76” strength=“0.22” radius=“73”/
Vignetting focal=“146” strength=“0.5” radius=“50”/
ChromaticAberration r_c=“0.11” focal=“9.12” b_y=“0.08”/
ChromaticAberration r_c=“0.1” focal=“11.1” b_y=“0.1”/
ChromaticAberration r_c=“0.09” focal=“15.76” b_y=“0.09”/
ChromaticAberration r_c=“0.09” focal=“146” b_y=“0.09”/
/CorrectionCoefficients
/Lens
/LenseDb

0 Likes

(Svetlana G.) #13

Hello @BBT,

I have sent the request to the image data team. As soon as I’ve got a reply I’ll let you know.

Regards,
Svetlana G.

1 Like