jmliss Posted August 9, 2021 Posted August 9, 2021 With respect to the attached example, I believe there are two bugs. I created two alpha zones (alpha less than 255) in two spots. I am using 1.10.0 on Catalina. Bug 1, the merge Visible layer with normal blending seems darker then just looking at the alpha zones layer. The second bug is related to selecting alpha ranges. The three selection ranges seem to work on the merge visible layer, but not the alpha zones layer. In addition, I experienced similar bugs when using the channel mixer on the alpha channel. Alpha bugs.afphoto Quote
NotMyFault Posted August 9, 2021 Posted August 9, 2021 (edited) 1 hour ago, jmliss said: Bug 1, the merge Visible layer with normal blending seems darker then just looking at the alpha zones layer. I think this is by design and expected (not a bug), based on the blend formula. Ares = 1- (1-A1) * (1-A2) https://en.wikipedia.org/wiki/Alpha_compositing so from 114 you get 177 (using 114/255 to normalize range) 1 hour ago, jmliss said: The second bug is related to selecting alpha ranges. The three selection ranges seem to work on the merge visible layer, but not the alpha zones layer. The next topic is by design, too: If you want to create a selection from the mask layer (Curves Adjustment), you can select the adjustment layer (not the pixel layer) and use Select>Selection From Layer 1 hour ago, jmliss said: In addition, I experienced similar bugs when using the channel mixer on the alpha channel. Can you explain? Channel mixer works perfectly fine, but the resulting effects could be un-intuitive. Edited August 9, 2021 by NotMyFault Quote Mac mini M1 A2348 | MBP M3 Windows 11 - AMD Ryzen 9 5900x - 32 GB RAM - Nvidia GTX 1080 LG34WK950U-W, calibrated to DCI-P3 with LG Calibration Studio / Spider 5 | Dell 27“ 4K iPad Air Gen 5 (2022) A2589 Special interest into procedural texture filter, edit alpha channel, RGB/16 and RGB/32 color formats, stacking, finding root causes for misbehaving files, finding creative solutions for unsolvable tasks, finding bugs in Apps. I use iPad screenshots and videos even in the Desktop section of the forum when I expect no relevant difference.
jmliss Posted August 9, 2021 Author Posted August 9, 2021 @notmyfault - I appreciate your feedback, but I don't agree with you. Yes, I may be wrong, but in the example I posted from my intel iMAC running Catalina. When I get a chance I will try on my intel mac-mini running Big Sur and if I see a difference I will post accordingly. Seems you are on W10 and iPads, and I assume you did not see the behavior using my posted example file? 1) Even just selecting partial alpha transparency, on my MAC, it only selects the full transparent section, even when just selecting on the curves layer. But in my posted example, if I select on the merged layer, it does it correctly, where correctly is as per my expectations 2) From what I understand, the normal blend mode should be as follows (see below - copied from wikipedia) which seems to agree with the AP help page also copied below - 3) What I mean about channel mixer is that if I choose the alpha channel and say lower red, then the selection behavior is similar to using curves on alpha channel, which I contend as per my post, is buggy. 4) Last week Dave Straker (in affinity) also posted a bug pertaining to alpha channel manipulation. I actually found that bug and asked him to verify before he posted. Normal blend mode[edit] This is the standard blend mode which uses the top layer alone,[2] without mixing its colors with the layer beneath it:[example needed] where a is the value of a color channel in the underlying layer, and b is that of the corresponding channel of the upper layer. The result is most typically merged into the bottom layer using "simple" (b over a) alpha compositing (making the actual formula ), but other Porter-Duff operations are possible.[2] The compositing step results in the top layer's shape, as defined by its alpha channel, appearing over the bottom layer. From help page: Blend mode types Affinity Photo supports an impressive selection of different blend modes. The most commonly used blend modes are as follows: Normal—The default blend mode. The top pixels display over underlying pixels according to the level of top layer opacity. Quote
NotMyFault Posted August 9, 2021 Posted August 9, 2021 6 minutes ago, jmliss said: Seems you are on W10 and iPads, and I assume you did not see the behavior using my posted example file? You are right that i could not reproduce any "faulty" rendering. It would be great if you could take screenshots (or save "merge visible" layers), but please describe the exact context how they were created. If you get different results on MAC, this is certainly a bug. Never the less, to make the potential issue more clear and reproducible: Below i made 3 layers to compare side by side: 1. left: your "issue statement", blend of "Merge Visible" and "Alpha Zones" layers. Red is darker because alpha increases to 177 from 114 by alpha compositing. As this is the only area where i spotted the color changed described by you, my first answer relates to this area. If there is another color change, i misinterpreted your post. 2. middle: "Merge Visible" layer 3. right: "Alpha Zone" layer. 2 and 3 look absolutely identical, the info panel shows identical RGBA values (3 samplers positioned into the semi-transparent area) Quote Mac mini M1 A2348 | MBP M3 Windows 11 - AMD Ryzen 9 5900x - 32 GB RAM - Nvidia GTX 1080 LG34WK950U-W, calibrated to DCI-P3 with LG Calibration Studio / Spider 5 | Dell 27“ 4K iPad Air Gen 5 (2022) A2589 Special interest into procedural texture filter, edit alpha channel, RGB/16 and RGB/32 color formats, stacking, finding root causes for misbehaving files, finding creative solutions for unsolvable tasks, finding bugs in Apps. I use iPad screenshots and videos even in the Desktop section of the forum when I expect no relevant difference.
jmliss Posted August 9, 2021 Author Posted August 9, 2021 2 minutes ago, NotMyFault said: You are right that i could not reproduce any "faulty" rendering. It would be great if you could take screenshots (or save "merge visible" layers), but please describe the exact context how they were created. If you get different results on MAC, this is certainly a bug. Never the less, to make the potential issue more clear and reproducible: Below i made 3 layers to compare side by side: 1. left: your "issue statement", blend of "Merge Visible" and "Alpha Zones" layers. Red is darker because alpha increases to 177 from 114 by alpha compositing. As this is the only area where i spotted the color changed described by you, my first answer relates to this area. If there is another color change, i misinterpreted your post. 2. middle: "Merge Visible" layer 3. right: "Alpha Zone" layer. 2 and 3 look absolutely identical, the info panel shows identical RGBA values (3 samplers positioned into the semi-transparent area) Thank you again for responding. Firstly 1.10 behaves the same way on my Mac Mini, Big Sur. Perhaps the issue is the marching ants. I did another experiment of copying and pasting from the alpha zone layer and it seems what I expected worked, as does doing that in the merged visible layer. However, the marching ants in the alpha zone layer only seems to be around the fully transparent circle, thus that through me off, where it was around both circles in the merged layer. Yes, I realize there is some algorithm that AP uses to determine when to display marching ants. I will do some more experimenting. If I still have issues, I will include more screen shots. Thanks again 1 minute ago, NotMyFault said: You are right that i could not reproduce any "faulty" rendering. It would be great if you could take screenshots (or save "merge visible" layers), but please describe the exact context how they were created. If you get different results on MAC, this is certainly a bug. Never the less, to make the potential issue more clear and reproducible: Below i made 3 layers to compare side by side: 1. left: your "issue statement", blend of "Merge Visible" and "Alpha Zones" layers. Red is darker because alpha increases to 177 from 114 by alpha compositing. As this is the only area where i spotted the color changed described by you, my first answer relates to this area. If there is another color change, i misinterpreted your post. 2. middle: "Merge Visible" layer 3. right: "Alpha Zone" layer. 2 and 3 look absolutely identical, the info panel shows identical RGBA values (3 samplers positioned into the semi-transparent area) Quote
jmliss Posted August 9, 2021 Author Posted August 9, 2021 @NotMyFault - well I was wrong about the opacity blend modes and the documentation certainly for me was not clear. This video explains the formula and is perhaps what you were saying initially, so thanks again. Perhaps my only issue then, is the marching ants - thanks again for your help. Seems that Dave always has a video that addresses issues. NotMyFault 1 Quote
jmliss Posted August 12, 2021 Author Posted August 12, 2021 @notmyfault - seems their maybe a bug in the color and alpha range selection feature and Dave verified it for me. I will post the potential bug in another post. This is what was throwing me off. Thanks Quote
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.