Graphical glitch in Kyoukaisen-jou no Horizon

[daniel229]

Since unknownbrackets@4b978ed

before

[unknownbrackets]

And the before is how it ought to look, right? What blending mode is it using?

-[Unknown]

[daniel229]

It should be like this

saw two blending mode

[unknownbrackets]

And, if you add return false; after bool ShouldUseShaderBlending() {, it works like before, right? What if you add return false; under case GE_SRCBLEND_DOUBLESRCALPHA:?

It almost seems like it's drawing the wrong texture or wrong part of it now... weird.

-[Unknown]

[daniel229]

either return false in bool ShouldUseShaderBlending() { or case GE_SRCBLEND_DOUBLESRCALPHA: it make the screen like before.

[daniel229]

another screens compared

[unknownbrackets]

What if you replace "vec3(v.a * 2.0)" (both) with "clamp(vec3(v.a * 2.0), 0.0, 1.0)"? What if you use "vec3(v.a)"?

-[Unknown]

[daniel229]

with "clamp(vec3(v.a * 2.0), 0.0, 1.0)" or with "vec3(v.a)" both like this

[unknownbrackets]

Does it change at 1x render? It seems like even using shader blending screws up the texture. I guess it must be drawing twice to draw the texture, and then the white of the bg or something? Maybe the issue is becaues these overlap and are combined into a single drawcall...

But, no, they'd have different textures... hmm.

-[Unknown]

[daniel229]

No difference between 1x and others

[daniel229]

fixed in #6350

[unknownbrackets]

Cool, I think it was drawing with overlap or something, which the blitting method doesn't support.

-[Unknown]