One remark: This finding was based upon the HUD elements. I didn't check if the same would apply also e. g. for the DrawTLColumn, I will have a closer look there.

Edit: So the Translucent Column drawing works in alternative mode for Palette:

r_draw.c:

.. and since the BlendOverTinttab is assigned here (currently working like the alternative translucency), this part seems fine.

Just note, that if the above suggestion to swap it for Heretic is considered, this blend function assignment is impacted aswell.

😟 I'm out of ideas how to make it proper. TrueColor support for Heretic was implemented before Hexen, and it's normal (non-alt) translucency was picked after some joint testing, but looks likes I have missed alt-one.

Another problem is - we don't have a formulas to set a proper percent for opacity, so...

Probably easiest way will be to have a separate I_BlendOverAltTinttab function with different values?

😟 I'm out of ideas how to make it proper. TrueColor support for Heretic was implemented before Hexen, and it's normal (non-alt) translucency was picked after some joint testing, but looks likes I have missed alt-one.

I wouldn't say that the alternative one is missed, rather that it requires a swapped behavior to Hexen. Currently in truecolor Heretic, the alt one has too much opacity, while the normal one has too little. In terms of functions, here an example how it could be corrected:

i_truecolor.c:

// [crispy] TINTTAB blending emulation, used for Hexen
const uint32_t I_BlendOverHexTinttab (const uint32_t bg, const uint32_t fg)
{
    return I_BlendOver(bg, fg, 0x60); // 96 (38% opacity)
}

// [crispy] More opaque ("Alt") TINTTAB blending emulation
const uint32_t I_BlendOverHexAltTinttab (const uint32_t bg, const uint32_t fg)
{
    return I_BlendOver(bg, fg, 0x8E); // 142 (56% opacity)
}

// [crispy] TINTTAB blending emulation, used for Heretic
const uint32_t I_BlendOverHerTinttab (const uint32_t bg, const uint32_t fg)
{
    return I_BlendOver(bg, fg, 0x8E); // 96 (56% opacity)
}

// [crispy] Less opaque ("Alt") TINTTAB blending emulation
const uint32_t I_BlendOverHerAltTinttab (const uint32_t bg, const uint32_t fg)
{
    return I_BlendOver(bg, fg, 0x60); // 142 (38% opacity)
}

Another problem is - we don't have a formulas to set a proper percent for opacity, so...

Yeah, getting it 100% right might be another topic. A swap would be a start to get it closer to vanilla, in my opinion.

Probably easiest way will be to have a separate I_BlendOverAltTinttab function with different values?

This could be one way, like the abstract posted above. Another alternative could be to have global variables with the opacity being initialized differently based upon the game and used for calling the I_BlendOver functions.

// [crispy] TINTTAB blending emulation, used for Hexen and Heretic
const uint32_t I_BlendOverHexTinttab (const uint32_t bg, const uint32_t fg)
{
    return I_BlendOver(bg, fg, He_Normal_Opacity);
}

// [crispy] ("Alt") TINTTAB blending emulation, used for Hexen and Heretic
const uint32_t I_BlendOverHexAltTinttab (const uint32_t bg, const uint32_t fg)
{
    return I_BlendOver(bg, fg, He_Alt_Opacity); 
}

One could also alter the returns of the functions based upon the active game.

I have created a first suggestion based upon the global variables approach:
master...Noseey:crispy-doom:Heretic_TC_Transparency_Fix

I have created a first suggestion based upon the global variables approach: master...Noseey:crispy-doom:Heretic_TC_Transparency_Fix

Thanks for that! However, you know, I'm a friend of function pointers and generally prefer them over global variables. So, you could keep the current I_BlendOverTinttab()/I_BlendOverAltTinttab() functions with the constant opacity values but under different names, and let an init function which takes the gamemission as a parameter (c.f. d_mode.h) decide which of both functions the actual I_BlendOverTinttab()/I_BlendOverAltTinttab() function pointers point to.

... and let an init function which takes the gamemission as a parameter (c.f. d_mode.h) decide which of both functions the actual I_BlendOverTinttab()/I_BlendOverAltTinttab() function pointers point to.

Would it be acceptable to modify V_LoadTintTable to do so? It would - depending on the compile-switch - either load the tinttable or assign those functions for the blending (which acts like a "tinttable"-selection for truecolor).

Could you prepare a PR, please. It's easier to discuss actual code changes than some written ideas.