alowe Posted March 24, 2018 23 minutes ago, kb1 said: Not sure what you mean. I make observations. For example, *you* seem triggered. Which is strange, because #4: "Nothing wrong with your method". Again, less drama is an option. I've done what I can to help answer OP's question, so I'm also done here. Yeah, I'm equally done because people are accusing me of trying to own the thread when all I am doing is answering people's days of questions. Blocking method works reliably. Other methods don't. That's what I've said all along. People can do what they want with that. I'll continue doing what works because it works and that to me is all that is important. 0 Share this post Link to post
zokum Posted March 25, 2018 (edited) On 22/03/2018 at 5:41 AM, fraggle said: Assuming you're mapping for Doom-in-Doom format, you might find that BSP does a better job than other node builders in avoiding slime trails. Lee Killough added some heuristics years ago to try to avoid them and it helps. Nothing is perfect but maybe give it a try. I think ZenNode and therefor ZokumBSP also has code to avoid slime trail. It will compute how inaccurate the floating point to fixed point conversion is and will try to avoid splitting diagonal lines over a certain threshold. I think it's this line: https://github.com/zokum-no/zokumbsp/blob/2a1ac55cdb18ffc607a31b4eee6e4b85da39707f/src/zokumbsp/zennode.cpp#L1211 The code does seem a bit fishy, I'll have to look into it at some date. There seems to be possible optimizations to do as well. It should at the very least if it has to, pick the line with the least error, not just the "first" line with any error if they all have an error. Edited March 25, 2018 by zokum 4 Share this post Link to post
kb1 Posted March 27, 2018 On 3/25/2018 at 4:46 PM, zokum said: The code does seem a bit fishy, I'll have to look into it at some date. There seems to be possible optimizations to do as well. It should at the very least if it has to, pick the line with the least error, not just the "first" line with any error if they all have an error. I can't tell - is it really calculating the amount of deviation from integer coords at the split point? (It might be - I just can't tell from looking at it locally.) If I was working on it, I'd have to animate the split process to verify it - trying to just follow the math blows my mind, you know? :) I wonder if a nodebuilder could be coerced into moving the vertices as needed, kinda like the FixSlimeTrails function, but during map build. Still try to avoid such splits, but if not possible to avoid a split, move the vertices. Do you think that would produce better results? 0 Share this post Link to post
Szymanski Posted April 6, 2018 Some sources of slime trails are possible to view in GZDoombuliders node viewer, you'll often find very thin sectors where the vertices aren't where you'd expect. Sometimes voids (fully black areas) between areas. The node builder works best with lines that are parallel, It's sometimes possible to avoid these issues by adding sectors which guide the node builder. Sometimes just moving vertices around will help through trial and error, that's best left for final edits. 1 Share this post Link to post
zokum Posted April 8, 2018 I wouldn't rely too much on the GZDB node viewer. Different node builders can produce different nodes. 0 Share this post Link to post
Soundblock Posted April 9, 2018 On 4/6/2018 at 5:38 PM, Szymanski said: The node builder works best with lines that are parallel This explains my nightmare with Plasmaplant rather well. 4 Share this post Link to post
zokum Posted April 9, 2018 I can only speak for ZenNode here. It will cache a list of colinear lines on the map, that's not the same as parallell. Colinear lines are lines that would if they had infinite length, be the exact same line. 0 Share this post Link to post
Szymanski Posted April 9, 2018 11 hours ago, Soundblock said: This explains my nightmare with Plasmaplant rather well. Very pretty, reminds me of a childhood t-shirt 12 hours ago, zokum said: I wouldn't rely too much on the GZDB node viewer. Different node builders can produce different nodes. It's all I have, there's no independent tools (that I'm aware of). 1 Share this post Link to post
Catpho Posted April 9, 2018 11 hours ago, Soundblock said: This explains my nightmare with Plasmaplant rather well."pic" That layout is beauty and nightmare at the same time. 1 Share this post Link to post
alowe Posted April 9, 2018 22 hours ago, Soundblock said: This explains my nightmare with Plasmaplant rather well. Nice use of lighting in that map too. Something I tend to forget about... 0 Share this post Link to post