[Daedolon]

I'm having this weird issue where a black polygon appears on the left hand side of the screen when viewing my map in Mapster32 in Polymost mode:


In-game it produces a hall of mirrors effect (pressing esc to bring up the menu makes the affected section black):


It doesn't matter at what ang you're looking at, it will still always occur in the left side of the screen.

I have absolutely no clue why this happens, it seem to only happen in this map and only in Polymost. Corruptcheck now gives the all clear and I don't see any problematic architecture so I'm kind of clueless. I tried deleting some of the sectors at random and the problem seems to persist through that as well. Only if I delete the majority of the sectors, the problem seemingly disappears.

The map file for the curious:
https://dl.dropbox.c...ke/omat/sub.map

[Mark]

I downloaded and I'll check it out right now. That problem seems familiar. It may have been brought up in the past.

[Daedolon]

I'm pretty sure I've seen it playing or looking through other maps in Mapster but I can't be sure. Note that I'm an ATI user (HD5750) and I'm using the latest Win32 binaries.

[Mark]

No glitches in Mapster or ingame Polymost using nvidia and Vista, and close to newest eduke version

[Micky C]

I've had a similar problem of a flickering black triangle that's somehow (I think) related to sector boundaries in the left side of the screen in polymost. In fact I'm having it right now in my CBP 9 section. It's most annoying. I thought it was only me.

[Daedolon]

I should have specified how it actually appears on the screen, here's a video of it in action (earlier version of the same map):



As you can see it's not constantly visible, and it seems like if you're in a more narrow space (like a long corridor), the effect doesn't stretch out so far to the middle of the screen, but is in fact much closer to the edge of the screen.

[Mark]

I did a lot of moving around, but I'll look again.

Added: Still no problems for me.

[Micky C]

Yep, that's what's happening to me on some maps. I'm getting the exact same problem.

[Mike Norvak]

The Big Cheese, on 14 February 2013 - 12:56 AM, said:

Yep, that's what's happening to me on some maps. I'm getting the exact same problem.

Are you serious? that big cheese doesn't give you credibility...

[Mblackwell]

I get that from time to time.

[The Commander]

Let me guess, the people getting this issue are all running ATI cards?

[Plagman]

No, that's the puzzling part. A lot of people get it, no matter what card/driver they use. It doesn't seem to be tied to a specific config option or an add-on pack either. And I suspect people that get it also get the Polymost TROR glitches. I've never been able to actually observe this on any machine that I have.

[Daedolon]

I noticed one definitive thing about this issue yesterday; It ONLY occurs if your horiz changes. Looking straight ahead will not show this particular issue at all.

[Helixhorned]

Introduced by me in r2414, more analysis pending.

[Plagman]

So the difference would be having r_usenewaspect set or not? I had people try with a scratch config to rule out different config cvars and some people were still getting them..

[Daedolon]

I went through all the aspect ratios and they all had it (in r3480).

[Plagman]

But can you confirm that r2414 caused it on your end by using a snapshot before and after? Also try the TROR glitch map thing.

[Daedolon]

Both this issue and the one described in this thread happen both in-game and Mapster32 in r2402 and r2416. So I don't think it's something you did in r2414.

[Helixhorned]

You're right, the isssue is older than r2414. It's just that I did a quick bisection, and for one particular configuration, it disappeared with the preceding revision. Going back a little more, it can still be reproduced with r1500! Generally, this bug easily shows up with horiz values close, but not equal to, 100 (looking straight ahead), but is very sensitive to the exact coordinates.

[Daedolon]

Huh? Weird.

I only see it when I'm looking up or down (in latest EDuke32 versions). Whenever I look dead ahead (disable mouse aiming), I don't get this issue.

[Micky C]

He said when you're looking straight ahead you don't get it, but when you're looking at angles close to straight ahead, that's when you get it.