Jump to content
Search In
  • More options...
Find results that contain...
Find results in...
kristus

Hexen mode, node building issues.

Recommended Posts

So, yeah I'm working on a Hexen WAD. But something that I am sufering with right now is that there's a lot of nodebuilding issues cropping up.

The issues I get is bleeding sectors, sorting issues with sprites and clipping issues. I've yet to figure out why any of them appear, and they are fairly plentyful.

I'm using DB2 and Zennode btw. But I've tried with ZDBSP as well, and it doesn't appear to be doing any difference.

http://www.doglike.org/temp/Screenshot_Hexen_20100912_195254.png
http://www.doglike.org/temp/Screenshot_Hexen_20100912_195420.png
http://www.doglike.org/temp/Screenshot_Hexen_20100912_195446.png

Share this post


Link to post

I have seen in the last few days that someone just upgraded zdbsp. I sadly can not remember where I saw that, however. I am not even sure if it were an "official" release or just someone hacking it for zdoom/hexen format. Report was that it improved it alot. It may have been the zdoom forums or most anywhere on the web. Sorry.

If you can locate that new one it may well work for you. You could also try the deepsea bsp which is quick and good.

Not sure if this will help or not.
http://forum.zdoom.org/viewtopic.php?f=3&t=26831&p=511741&hilit=bsp+update#p511741

Share this post


Link to post

If the map causes problems with several different node builders there might actually be a mapping error.

Another possibility is a precision error. You can test this by building extended nodes with ZDBSP and testing in ZDoom. I have seen these on rare occasions and they tend to produce rather ugly visual glitches.

Share this post


Link to post

I had an issue exactly like the first screenshot this week. Where a lower texture turned into HOM part of the way down. After much grief, I finally tried re-tracing all the linedefs in the sectors related to the offending HOM - effectively making the sectors over again, and the issue disappeared completely.

If that hadn't worked, my next step would have been to try deleting the NODES lump in the wad, and seeing if anything changes from a fresh nodes rebuild (though I think nodebuilders start from scratch every time, so this would have been along the lines of prayer).

Share this post


Link to post
Graf Zahl said:

If the map causes problems with several different node builders there might actually be a mapping error.

Yeah, I've considered that, and I don't think so. All the major issues I've really looked hard at and I can make them go away by fucking with the verticies it relates to, but they will come back when I rebuild the nodes a second time.

As for the hole in the floor issue. (shot 3) I can probably fix that by putting a few self referencing linedefs in that area. But I can't figure out any reason why the issue exist in the first place.

Plus, this is something I've not experienced before. So it's unlikely that I've suddenly started making technical mapping errors that I can't figure out how to fix. :p That I get these kinds of errors could of course be because of new software. The only thing that is truly new is that though is that I am using Hexen map format. Somewhat new is that I am using DB2 of course. But since I appear to be rather alone in experiencing these recurring issues it's unlikely that either of those are truly to blame for it.

Graf Zahl said:

Another possibility is a precision error. You can test this by building extended nodes with ZDBSP and testing in ZDoom. I have seen these on rare occasions and they tend to produce rather ugly visual glitches.


Yeah, I've considered this as well. But it's not like these maps are bigger or more complex/detailed than the average Zdoom map, so I really can't say why this is happening to me so frequently.

Also, what confuses me is that the blockmap issue is in both Vanilla and Zdoom. AFAIK, Zdoom isn't using the blockmap anymore, so the same bug shouldn't be appearing in both, unless there is a mapping error. But the sector that got the issue isn't even a very complicated one and I can not find anything wrong with it.

EDIT: I fixed the blockmap issue by moving a vertice one unit in relation to another vertice. (They were 1x2 units away from eachother before that so they are now 2x1). This did fix the hole.
I don't think it was because of the relation to the other vertice that caused the issue though. But quite likely in relation to the closely adjacent line. Which suggests that this one at least. Was due to imprecision.

But none of the really complex structures I've made has shown any bleeding issues or similar. All the issues are on generally mundane structures. Though what they tend to have in common is that they are all rounded shapes that are off grid.

EDIT2: Seems like I've fixed the worst issues now... We'll see what happens later. Haven't tested in Vanilla yet. But I am using ZDBSP instead of Zennode now and I moved a few of the culprit verticies to a higher grid line state (2x2 instead of 1x1) if that makes sense to anyone. The issue shown in the second screenshot remains however. But that is only showing up in ZDoom of the engines I've tested in so far.

Share this post


Link to post
kristus said:

EDIT2: Seems like I've fixed the worst issues now... We'll see what happens later. Haven't tested in Vanilla yet. But I am using ZDBSP instead of Zennode now and I moved a few of the culprit verticies to a higher grid line state (2x2 instead of 1x1) if that makes sense to anyone. The issue shown in the second screenshot remains however. But that is only showing up in ZDoom of the engines I've tested in so far.


If that helped it was most likely a precision issue. Remember that the original BSP structure stores vertices in full integer precision only. So even if you see it glitch only at 'mundane' structures it's most likely caused by some high-precision geometry trying to force splits at points where they can't be inserted cleanly. To handle these things properly fixed point vertex precision would be needed - but that'd require an advanced port that allows using those.

Share this post


Link to post

Seconded. Given the close proximity of the vertexes in question, subsequent splits to the surrounding geometry are very likely to induce 'holes'. This can be better described as a 'collapsed subsector' - in the sense that due to welding and/or rounding it's size is reduced to zero (Z=0 plane).

Share this post


Link to post

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
×