abcdaniel Posted November 23, 2020 Share Posted November 23, 2020 (edited) Lowering luminosity with an HSL adjustment layer on pale colored pixels might saturate the pixels, instead of making them darker. This is true where any of the RGB values is maxed out, ie at 255. I discovered this with a sky that was slightly blown out. The example image shows blown out reflections of spotlights on the ice in a hockey game. A slight adjustment of luminosity makes the border of the blown out area saturate, in RGB/CMYK base colors. This can be reproduced by purposely filling areas of a pixel layer with colors like 255 255 249 and then applying HSL and lower the luminosity. If you push down brightness far in a brightness/contrast adjustment layer, you will get a similar effect. I consider this a seriuos bug. Lots of source photos have blown out details by necessity, like the hockey rink spotlight reflections in the image. Combine it with jpeg blocking and the luminosity slider is unusable. Edited November 24, 2020 by abcdaniel Added similarities to brightness/contrast and version number 1.8.6 Quote Link to comment Share on other sites More sharing options...
abcdaniel Posted November 25, 2020 Author Share Posted November 25, 2020 This bug was also reproduced on a friends mac running Affinity Photo 1.8.3 and on my iPad, with the latest version of Affinity Photo for iPad. Quote Link to comment Share on other sites More sharing options...
lepr Posted November 25, 2020 Share Posted November 25, 2020 5 hours ago, abcdaniel said: This bug was also reproduced on a friends mac running Affinity Photo 1.8.3 and on my iPad, with the latest version of Affinity Photo for iPad. HSL Shift Adjustment had a useful L control in AP 1.6.7. When it changed in 1.7.x to the crap it is now, I posted a bug report and a Serif representative replied that the HSL Shift Adjustment was deliberately changed to make it match "another app", but that app was not named. 32 bpc mode still uses the old HSL Shift code. Quote Link to comment Share on other sites More sharing options...
abcdaniel Posted November 26, 2020 Author Share Posted November 26, 2020 My user scenario: I teach at a university. We did not get access to Adobe licenses this semester, and Affinity is much more benign, so now we're using Affinity for remote learning. Thanks Affinity! I'm teaching 60 students this semester, half of which have used Adobe before. That's 60 new Affinity users. I've been ok about switching to Affinity, because I use it privately in parallell to Adobe, I am optimistic about Affinity, and I wanted to stress test it. If there are these kinds of bugs, HSL being one of the adjustments we depend on, Affinity is a no go after the pandemic , and we switch back to Adobe. The students already familiar with Adobe will be disappointed and frustrated, and so will I. Quote Link to comment Share on other sites More sharing options...
Staff Chris B Posted November 27, 2020 Staff Share Posted November 27, 2020 Hey abcdaniel, We changed it so it behaves more like luminosity should i.e. match the perception of human vision. However, colours shouldn't saturate more if we drag the slider below 0 so I think there is indeed an issue here. I will feed this back to the developers. Quote How to format a bug report | Learning Resources | List of V2 FAQs | YouTube Tutorials Link to comment Share on other sites More sharing options...
abcdaniel Posted November 30, 2020 Author Share Posted November 30, 2020 Hey Chris B, thanks for the reply! I think HSL is a true and tested approach/algorithm to handle color adjustments. The implementation here, that is the bug, is spreading out the RGB values in pale colors when pushing down luminusity; then we get saturation. I believe the classical approach to lowering luminosity or brightness is to either keep the RGB-values at the same relative strength when lowering their values (basically subtracting the same value from each of the RGB) or compressing them to a new range, which will actually decrease their spread and thus the saturation. In my courses we talk a lot about perception. Do you have a reference or a link to how your approach fits better to human perception? If this bug isn't resolved, it would be a nice thing to bring into the courses. Quote Link to comment Share on other sites More sharing options...
Staff Chris B Posted December 1, 2020 Staff Share Posted December 1, 2020 On 11/30/2020 at 11:22 AM, abcdaniel said: Do you have a reference or a link to how your approach fits better to human perception? I have asked the developers. Just for the record, I think there is a bug with negative values for luminosity. I do not think they should saturate. Quote How to format a bug report | Learning Resources | List of V2 FAQs | YouTube Tutorials Link to comment Share on other sites More sharing options...
abcdaniel Posted December 2, 2020 Author Share Posted December 2, 2020 Thanks Chris B! We're talking in our course about pros and cons of different software, about perception and the tools we use to look at and create images. This discussion here is live, so it is interesting to put it in the teaching mix! Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.