Zyl

Yep, that was it. I have to admit though I don't really understand why the background is subject to special treatment like so. This is good to know! I'm not really sure what are some use cases to want to blend in sRGB specifically. I imagine an overwhelming amount of hobbyists, and even some professionals, aren't even aware that getting this wrong can negatively affect their results. Even if they notice pixelated lines or such, they might just shrug it off as consequence of too low DPI, not knowing that something can be done about it. As a hobbyist, I can only attest to having learned about it by chance, and that thus I strongly feel that defaulting to anything other than blending in linear RGB is undesirable. Anyways, I got everything indistinguishably close to what Photoshop produces (the exact values are off by up to 2 in the 8 bit export, but that doesn't matter) and it's looking good, so I call this a closed case. Thanks, all!

Thanks for the detailed response. Unfortunately, I cannot reproduce the behavior you describe. Here is what I do: It's still blended in sRGB, even after export, despite having chosen 32 bit color format as you advised. What is wrong?

So... can we or can we not currently do this?

Here is a comparison which makes the advantage of blending in linear RGB quite apparent. Affinity: Photoshop: In case you have a high DPI display, I follow up with the same but at 2x resolution, with nearest neighbor interpolation. Affinity: Photoshop: The keen eye will be able to tell that the edges of the shape are more jagged in Affinity Designer's output. The cause of this outcome is purely mathematical. Blending in sRGB violates conservation of energy.

For comparison, the following is the result of the same in Photoshop with blending using gamma enabled: Each channel assumes a value of 188, which is the nearest sRGB representation for a pixel with 50% white and 50% black coverage in byte-encoding.

I made a simple 16x16 pixel file with a black vector triangle. I export this to PNG using the default settings, and inspect the result in MS Paint. The channel values of the center diagonal of grey pixels are 127, revealing that either the document has been saved in linear RGB or colors have been blended in sRGB color space. This matches what is visible while viewing and editing the project within Affinity Designer with "Pixel View mode" enabled: I want color information to be saved in sRGB. More precisely, I want all color blending operations to happen in linear RGB, and color display and export to happen in sRGB. Embedding the correct color profile into the file is not sufficient, as converting every channel to byte is already a highly lossy operation. How can I achieve this?

When I combine two overlapping curves with Add the overlapping parts get clipped away. How can I just dump the nodes of one curve into the other?