Pitch + Angle Based Sprite Masking

[Major Cooke]

Done. New pull request issued due to merge conflict, so that's fixed as well.

Code: Select all

VisibleAngles 0, 90
VisiblePitch 0, -20

Anything else, or still waiting on more feedback?

[Major Cooke]

Updated once again for the latest series of changes.

Is there anything else?

[Graf Zahl]

No idea. As you may see I have been extremely busy the last week so I haven't taken a closer look at this code.

[Graf Zahl]

One note about the pull request: Unless you change the handler for some property you should not bump savegame versions anymore, i.e. only change it if you need to add some 'if (SaveVersion >/< xxx)' to implement it.

[Graf Zahl]

So what's the reason for this:

Code: Select all

	player_t* pPlayer = players[consoleplayer].camera->player;

	if (pPlayer && pPlayer->mo)

You are essentially disabling the feature for non-player cameras entirely. You alre also not comparing the angle toward the actual camera but to the player actor attached to the camera, which is not necessarily the same. Why do you require

players[consoleplayer].camera->player->mo

and not just

players[consoleplayer].camera

?

Last but not least - this does not consider chasecam etc. Since this is a RENDERER feature you cannot do comparisons against PLAYSIM values, you have to use the renderer's ViewPos to determine where the camera is, not the camera actor's position.
Too bad I only noticed this after committing, it should be easy to fix, though.

[Major Cooke]

Couple other bug fixes I'll need to put in too which were slightly overlooked. I'll host them in a new bug thread.

[Graf Zahl]

I'm sorry but that's still not correct. You have to be very careful here with portals. There is no guarantee that your calculations end up in the same portal group. I had a quick look and it seems that the needed info is currently not stored globally. I'll have to think about it a bit.

[Major Cooke]

Got it all fixed for ya. You actually don't need to do anything else, it appears.

Yeah I meant to say that first PR was just for seeing how I was doing, and I had run into a little snag. However, I figured out how to fix it all and I even tested it with Arookas' map.

[Edward-san]

Your buggy version of AActor::IsInsideVisibleAngles managed to trigger an llvm backend internal error when trying to compile release mode with clang, then it stopped with your latest fixes.

[Major Cooke]

So I don't need to worry about it then?

[Edward-san]

You don't, indeed.