Colored Fog Issues

[NeuralStunner]

Attached is an example map with a couple quasi-liquid 3D sectors (with steps in the bac so you can get out), and two darkened sectors. The "blood" side is given a bright red fade color, the "water" side uses default fading and is at 128 brightness. The rest of the map is fully bright.

Noted problems:

1. When under the "blood", the sector fog should be bright red (0xFF0000). It's only half the correct color(0x7F0000).
2. When under a 3D surface, you can see the red zone. (Limitation of software 3D floors I guess.)
3. When inside the red area, it looks mostly correct, though the darker tones from the east sector will vanish if a 3D floor is in sight.

GZDoom has some different issues, which are posted in its forum.

[Graf Zahl]

1. is not a bug. Did you set the opacity of the fog? If not it's set to 50% which effectively makes the color 0x7f0000)

[randi]

2. I don't know that I want to try to fix this since (1) I didn't write the code and (2) there's a 3dfloors3 branch to finish that it probably wouldn't apply to anyway.
3. I can't see anything but red inside the red area. Am I supposed to be able to see something else?

[NeuralStunner]

View 1:

---

Looking at the darkened sector to the east. I figured this is due to traditional "light 0 is not fully dark" lighting.

View 2:

---

A few degrees to the left, with the 3D sector in sight. The dark areas have disappeared. (They actually vanish a bit before the FOV hits the 3D floor.)

Might be another issue for 3dfloors3, not sure.

The "underwater" lighting was the one that really surprised me, honestly. But I guess it's emulating GZDoom's method? (I never did try comparing a Transfer_Heights.)

[randi]

I can't see a difference between View 1 and View 2, even when I overlap them in Photoshop and use difference blending...

[Project Shadowcat]

I see a couple of platforms in view 1.

[Gez]

It's very slight. With contrast increased thanks to the "color curves" tool in Gimp:

---

[Graf Zahl]

Well, I'd say that this isn't a bug. Apparently some colors that differ more strongly from the fog don't get blended away completely. I guess if you used the same algorithm to draw fog boundaries with a true color renderer you'd see a very hazy image of the rest of the level so I'd write off the fact that most is indeed red, not to intent but to palette limitations.

In that case I think we can safely file this report under 'won't change'.