Adjustment bug - would someone see if they can reproduce?
- Open an image
- Click on the ‘Local’ tab
- In the default adjustment, invert the mask so the adjustment applies to the full image. I also labeled it ‘1st Adjustment’
- Leave the settings at default (-1 exposure, everything else 0)
- Click the button on the adjustment title to turn the adjustment on and off
- Note that the adjustment works as expected, the image darkening when the adjustment is on and lightening when it is off
- Double-click on ‘Exposure’ to set it to 0
- Click the button on the adjustment title to turn the adjustment on and off
- Note that the adjustment works as expected, the image does not change
- Add a 2nd adjustment and invert its mask; leave the exposure set to -1, I labeled this ‘2nd Adjustment’
- Enable and disable both adjustments noting that they both behave as expected, the 1st adjustment has no effect on the image and the 2nd adjustment darkens it
- Drag the 2nd adjustment above the 1st
- Turn off both adjustments, then turn on the 2nd. The image darkens as expected. Now turn on the 1st adjustment – the image lightens even though the 1st adjustment is set to do nothing and indeed did nothing until I added the 2nd adjustment
-
Yep. Same thing is happening to me. However, the order of the adjustments I made might have been inverse. My stack was:
2nd Adjustment
1st Adjustment
And then when you said "drag the 2nd Adjustment above the 1st, I took that to mean "in front of the 1st" and my stack became"
1st Adjustment
2nd Adjustment
But the end result was the same. Turning on and off the 1st Adjustment indeed did cause it to lighten the photo back to normal.
0 -
Leroy,
On my pc this works as I expect - ie my results conflict with yours
0 -
Just wanted to add some screen snapshots. The 1st one is with 1st Adjustment disabled. The 2nd image is with it enabled. The 3rd is of the same environment as the 2nd image, just showing you that the 2nd adjustment has a -1 exposure.
0 -
Kevin, thanks for verifying the issue. I can reproduce the issue without dragging the adjustments too.
Peter - thanks for trying. I don't understand why your results differ. Are you using PR 2019.2? What version of Windows?
0 -
Leroy,
I used a jpg to test
- Win 10 pro
- i5 8500
- SSD for images, NVME for scratch and cache
- PR 2019.2
- NVIDIA GeForce GTX 1050 Ti
0 -
Leroy,
I used a 14-bit Nikon NEF raw file to test. Windows 10 Home
0 -
This happens on Mac as well and is "broken as designed".
In 2019 Local Adjustments were re-implemented so that every local adjustment is based on the basic image. So in the example if the top level adjustment has all sliders set to zero this means that the local adjustment is effectively a copy of the basic raw image disregarding any previous adjustments. As the mask is set to apply this adjustment to the whole image it overrides any previous local adjustments such as the one that darkens the image.
This means that if you want to apply local adjustments within local adjustments top level adjustments have to be set up incorporating the changes from previous adjustments. E.G. if an area has been lightened by +1 and then you want to darken some parts of this area by -0.5 then the second adjustment has to be +0.5 not -0.5 as it would have been in 2018 when local adjustments took any lower overlapping adjustments into account.
Lightroom works as 2018 did, so why in promoting On1 as a Lightroom replacement they decided to change the behaviour I have not idea. As far as I am aware none of the publicity about re-implementation of local adjustments in 2019 to work on the Raw data mentioned this change in behaviour.
I believe that there have been a number of complaints about the change and a future upgrade will introduce a way to select how multiple local adjustments interact. No idea when this will happen.
0 -
Thanks for the explanation. It sounds like the top adjustment overrides everything below it except where it is masked out, is that correct? I'm going to have to play with this and see if I can understand how adjustments interact.
Don't changes in the software like this change images that were created with the old behavior?
0 -
Correct, if local adjustments overlap then where they do the overlapped area gets the adjustments in the top level only, with sliders set to zero meaning that this setting is not changed from the basic developed image, not left as it was as a result of the lower level adjustments in the overlapping area.
I think that somehow 2019 realises that it is working on an image first processed in 2018 and behaves as it did in 2018, or at least in terms of opening the image in the first place. Not sure what happens if a new local adjustment is then added.
Regardless of whether it changes earlier images or not it has meant a change in workflow. Some courses I viewed made significant use of overlapping local adjustments with finer adjustments within a larger local adjustment.
0 -
David, I ran into this issue while (trying to) follow a tutorial video. The new behavior (seems to me to) make adjustments less flexible and powerful. But their must be an advantage to doing things this way or presumably they wouldn't have done it (I hope).
If you don't mind my asking, how do you have such insight into PR internals? Or did you figure this out by trial?0 -
As I joined On1Plus I had access to the 2019 beta program and ran into this problem when exploring that. Like you I though that it was a bug and reported this and followed various discussions. I don't have any particular insight into PR internals but I spent my working life involved in the provision of a wide range of computer services in UK university environment, so lots of insights into programs not doing what one would expect and figuring out what it had decided to do instead, if not exactly how it came to be doing this.
Eventually On1 did indicate that this was a designed change which seems to be linked with local adjustments being based on Raw data. There was a Redit discussion with Dan from On1 where he said that there had been a lot of feedback about the change and that they were planning to introduce a switch to allow users to select the behaviour that they wanted. I have not seen any formal notification that this will actually happen, and when.
I agree that the change doesn't seem to make any sense from usability point of view, even if there are technical advantages in current implementation. In addition the new implementation has also changed the behaviour of temperature adjustments so that even the smallest allowed step introduces a significant colour shift.
Effects still work as one would expect so one way forward is to use those for local adjustments but need several filters to cover the range of adjustments in local adjustments and then have to fiddle around copying and pasting masks since no way to group filters together beyond getting into layers.
0 -
Thanks, David.
One of the fascinating features of PR is the way effects filters stack, each with its own mask, mode and sliders to allow fine control of the filter on the image. I'm glad we'll have the option to stack adjustments in a similar fashion again.
Has there been any discussion of PR's changing an image's colors when it opens the (non-raw) image? I mentioned this issue in another thread in this forum (https://tinyurl.com/y2ubjkey) but haven't received any comments.
0 -
Sorry Leroy, nothing to add about the colour issues. Not something I have come across in my photo processing which doesn't do the sort of things that you are describing.
0 -
Thanks, David.
I don't think this is something photographers and artists would ever notice unless they were doing something very specific. Support got back to me on this yesterday reporting no complaints and I made the same point in my reply.
Even though this isn't something that PR's target audience will worry about it is, IMHO, a fundamental problem. Editing software (expect for raw files) should not change the value of a single pixel merely by opening the file. For some applications this is a deal-breaker (not for me any more).
I've picked up Affinity Photo and am learning with it (while awaiting the next PR update). I'm not giving up on PR - they're very different tools with different strengths and weaknesses.
0
Please sign in to leave a comment.
Comments
14 comments