NotMyFault

This is to be expected. adjustments do not get converted to give visual identical results when the color space or color profile is converted. most adjustment work „simplistically“ on color values, e.g. curves, levels etc. when you convert to a wide color profile, the color values in pixel layers change, so the relation between input / output will change. if you change color profile from rgb to cmyk or lab, some adjustments simply stop working e.g. invert gives useless results in cmyk.

You may try this. A simple channel mixer adjustment can do the trick to show areas exceeding given limits.

I spent a few hours testing lots of combinations, and documented my results in the forum. Your screenshots show that you did not follow the tutorial exactly, and deviated in many relevant aspects: tutorial says exactly two layers, pixel on bottom, mask on top. you used a group of vector shapes, and placed a rectangle with channels mixer below. This is not to blame you, just an observation and my conclusions. It is too time consuming for me to reverse engineer your edit steps and guesstimate your intentions. Please try it again on your own, following exactly the steps as described. From older discussions i know that many new Affinity users perceive this learning curve as too steep, and i understand this. If your time is worth more than 200 bucks per year for Adobe, make a wise decision.

The trigger condition is simple: If mask is full black: TGA export is broken (mask is ignored, export is fully opaque). If mask is not fully black, but is only non-black where the pixel area is black: The exported TGA does not allow to recover RGB pixels. all RGB values get zeroed out. You may have any number of black mask pixels but you need at least one single pixel which is both not black in RGB channels and not black in alpha channel. Then the exported TGA file can be fully recovered in RGB channels. Tested on Mac with Photo V2. You milage may vary on iPad or V1

not for TGA export on iPad and a fully black mask. It is a specific edge case.

To be precise: specifically if all / majority of pixels are alpha zero. I still need to figure out when it toggles from correct (as before) to unusable. This has not changed. V1 and V2 behave identical. I only never checked for this specific case before you raised the issue.

When using a specific method for channel packing (see link below), export is wrong for TGA format if mask is fully or almost black. The test file has a mask with only 2 white pixels upper left / lower right. Problem 1: When exporting to TGA, the mask gets ignored and the TGA is fully opaque, despite the preview rendering. When changing export from TGA to PNG, export is representing the rendering. Image 1: Preview rendering for TGA wrong, check with navigator Panel Image 2: Image rendering correct for PNG (same file) Image 3: Solo / Isolation Mode of Mask Problem 2: If you re-import the TGA or PNG back gain into Photo, the RGB channels are not recoverable, but zeroed out. If you fill the mask with a white to black diagonal gradient, all TGB pixels are recoverable by filling the alpha channel after opening. almost empty mask export bug.afphoto

on Mac the behaviour changed as below: The tutorial works as before, when having a mask which is not fully black, e.g. using a black to white gradient. If the mask is full black (all zero), the exported file gets wrong: either the zero mask is ignored (despite correct rendering of preview), and the exported file is 100% opaque, or the export is not recoverable (RGB channels zeroed out). For your specific case requiring alpha to be zero for the complete canvas: this specific edge case is currently not possible with Affinity apps.

I checked again the tutorial on iPad with Photo V1 and V2. Unfortunately it is no longer working as before. Will test on desktop and update.

If you want alpha to be zero, just add an empty mask layer on top. the mask needs to be on top, otherwise it will get ignored.

So it is not about pixel selection, but about selecting layers in layer stack. i doubt that invert in layer stack will be usefull, as affinity treats parent / child layers or layer hierarchies very special. if you want do delete all except the selected layers, copy / create new from clipboard could do the trick in simple cases. Unless Affinity says “clipboard not suitable for new document” for unknown reasons. In complex cases inverting is not well defined

I don't understand the layer stack. the channels mixer in your screenshot is nested to a rectangle - and above there is a group of layers. Normally the alpha channel is a pixel layer (when created by the tutorial). any chance you can upload the actual file? and what editing do you want to achieve?

yes. During Export you can just check "rasterized", no need to do it seperately in Designer (which is possible)

While possible with many caveats in theory, this does not work well in Photo. While using Develop Persona: you can save presets for many of the sliders, but you must save several of them (no grouping possible), and Affinity often does not apply a preset even if it is shows as selected in UI (old bug). While in Photo Persona: You can save presets for many adjustment layers You can reord macros of edit steps. But this is far behind Adobe from the functionality and hampered by too many unfixed bugs. If you need to mass-edit Photos, look elsewhere.

I tried but was unable to reproduce on iPad, Mac, and Windows 10 PC (see signature for specs). As MEB could, it does not matter. But out of curiosity i would really like to see a screenshot what "white patches" appear for you.

And i see you have a blend range active on one mask. Bit depth changes may change blending results as rounding of values near zero or near 1 may have big impact

I see only 2 unsharp mask layers, but nothing else so the canvas is completely empty. Can you provide more details, including screenshots showing the before / after state? Did you attach the correct file?

Probably the simplest way, working in all 3 apps on all OS platforms since day 1 (V1, V2) Have the photo available as file on disk create the 9 rectangles (any shape will do, e.g. puzzle pieces) select all rectangles in layer stack use fill, set mode to bitmap, choose photo from disk

You can: use a rectangle shape filled white, and nest it to masking position of the image (over the thumbnail). rasterize and trim.

The actual afdesign file (maybe reduced to the relevant layers) would really help to reproduce. Any chance you can upload?

Can you show us the exact steps which are working on Mac? I did not fully get what you try to achieve, and how this step (probably embedded in additional steps not yet described) play into the target. The transform panel on iPad is more restricted wrt to formulas, it does only accept numbers and not the usual input variables like w or h. When working with numbers only, it has the same capabilities of adding / subtracting etc.

Photo is definitely the better tool for this task.

There is a workaround. You can activate rulers, grid, guides, or create helper objects to give you visual references. If required: use the pen tool to create a line marking the reference in the source image. Then use the transform panel, it shows the angle. Copy this value. Then select the image, paste in the value with opposite sign.

But probably this bug

I have a weak memory about a similar issue (but not related to export persona), about 2 years ago there were multiple reports but it was fixed