Due to unoptimizations in the game, blocking sightlines with ceilings does not actually work to solve lag. To block sightlines, you will want to outright delete sectors, making a "void" in the map of sorts.
That is partially what I meant actually. To makes voids work as an optimization in a non-jarring way (in an outdoor level), you have to slope the ceilings around it as smooth as possible. This is much easier to do with equation slopes and vertices.
Why slope the ceilings, if you have them high above the surface you end up with high pillars in the middle of the level, which may not be what the theme called for. Sloping them can reduce the height of the walls dramatically and ensure flying characters do not look at that part of the level from above.
I was not very impressed with this set of maps. I played it and wanted to be as honest as I could but I found myself getting frequently lost, there was a lot of not too great level conveyance, and then there was a moment like this which feels like it was made exclusively for people who play in first person only. Running through this section just caused me to run straight into enemies I had absolutely no way of seeing. There's a lot of cramped spaces in Lawn World. And there's a lot of really weird decisions made that I just don't like at all.
Found this while I was trying to point something else out
However I think the worst thing to me is level rule consistency and conveyance and I feel . And I don't feel that any other moment makes that more clear (imo) than this
The fact that I can't easily jump back to an earlier part of the level here had me confused the first time I played this level. When I was going through the level I straightup got lost because when I tried to make this jump from the reverse was met with an invisible wall which made me think I wasn't supposed to go that way. And given that there's this wall any time there's an invisible wall had me believe that this was a dead end initially. This kind of stuff is present all over the level pack. This stage in particular has a problem of having too much terrain that looks playable/explorable surrounding it leading to a very confused death.
I can appreciate these levels from the confines of it being a location, but I do not appreciate them at all as Sonic stages. The conveyance isn't very great and I feel like you could do a lot more to help guide the player towards the right path.
Overall I did not have a great time playing through this level pack and feel like a lot of work can be done to help improve the experience.
New version, much smaller this time, but I am getting back to working with the binary format so I can resume releases. I will also start using Software as my renderer again to reduce the chance of major issues sneaking by.
Fixed - backwards logic for the Brute Crawlas (in terms of checking z distance) was fixed. The broken logic made them more annoying than they should be
Fixed - Poly Object rendering in software, for the WIP zone Greenhouse Scamper
NEW - Lawn World act 1 now has a new route to take, just remember to not lose the whirlwind shield at the beginning to access it.
FIXED - SOC error breaking a Skyward Garden emblem
- A couple of texture alignments were also fixed (tm).
- Easier to hit starpost near the start of LWZ 1
In addition, a couple of changes in 0.25 I forgot to add. More playable space in BCZ 2 with rings and goodies to find, slight changes to make a few starposts (not the LWZ ones) easier to hit, and some extra...
Sorry for the release noise, but this fixes the display of emblem hints for Lawn World Acts 1 and 2 and an attempted fix for the graphics of an FoF where a LWZ 1 emblem is.
NEW: Greenhouse Scamper Zone is complete (for now, like other zones it is subject to future changes and enhancements). Tested in both Software and OpenGL (though OpenGL is still recommended due to how it handles transparency and lighting).
Record Attack emblem requirements were updated in response.
Unfortunately, the games data had to be reset again (to avoid what would otherwise be an issue with GSZ's emblems in existing games). The level thumb is still relevant as...
Actually apply some of the newer textures in a couple of spots that were missed previously (areas made before the textures were made in Lawn World.
Experimented with replacing a number of greenhouse glass FOF's with transparent and solid greenhouse glass midtextures (in LWZ2 and GSZ). The result is little to no render glitching where a lot of it went on before (in software mode) while looking the same. I am aiming to emphasize software only effects more so...
More utilization of alternative solid FoF types (for the purpose of bringing performance improvements to Skyward Garden Act 1, Broadside Cliffs Act 2, and Greenhouse Scamper).
Optimized the lua scripts in several areas.
Slight optimization to the map of BCZ2 itself. All optimizations listed, based on what I noticed, has improved performance in performance-sensitive areas (in Software rendering, OpenGL was always fast).
Updated preview for SGZ2 (with 3 emblems already in position to collect). There are a couple of visual issues with software rendering at the moment, but I apparently found one of the more notable ones related to sprites is actually fixed in next. The completed version may not be ready until the next release anyway.
So I played this with a small group of friends in Multiplayer, and the lag that we encountered was too much for us to go through. But when I played this solo, the lag was greatly reduced. Granted, there was still lag, but not nearly as much as when my friends joined. So I did a little digging... I think I know why this lags for a lot of people - it's not the levels themselves! (...I think?) But, I found a solution for optimization! It's a LUA script!
There's a LUA script called LUA_SPL2 that's meant to temporarily lock the player's movement when they use a horizontal or diagonal spring. It checks every player instance to basically see if they exist. This check happens on the "PreThinkFrame". The problem is that it also does this check for the "MobjThinker".
What I think is going on is that the game is having every object scanning all players every game tic, which depending on where the players are located can cause huge slowdown! The more Things being rendered to constantly check the players' existence, the more lag is created.
The addHook("MobjThinker"... should be removed or changed. Doing this eradicated most - if not all lag. And the spring-player-movement-locking feature still works, even for players that join a Multiplayer game mid-session.
I hope that this helps!
I do have to ask, why was the "MobjThinker" added in the script? I checked the earliest version of this map - v0.1 - and I noticed that the script had not included "MobjThinker".
So I played this with a small group of friends in Multiplayer, and the lag that we encountered was too much for us to go through. But when I played this solo, the lag was greatly reduced. Granted, there was still lag, but not nearly as much as when my friends joined. So I did a little digging... I think I know why this lags for a lot of people - it's not the levels themselves! (...I think?) But, I found a solution for optimization! It's a LUA script!
There's a LUA script called LUA_SPL2 that's meant to temporarily lock the player's movement when they use a horizontal or diagonal spring. It checks every player instance to basically see if they exist. This check happens on the "PreThinkFrame". The problem is that it also does this check for the "MobjThinker".
What I think is going on is that the game is having every object scanning all players every game tic, which depending on where the players are located can cause huge slowdown! The more Things being rendered to constantly check the players' existence, the more lag is created.
The addHook("MobjThinker"... should be removed or changed. Doing this eradicated most - if not all lag. And the spring-player-movement-locking feature still works, even for players that join a Multiplayer game mid-session.
I hope that this helps!
I do have to ask, why was the "MobjThinker" added in the script? I checked the earliest version of this map - v0.1 - and I noticed that the script had not included "MobjThinker".
Interesting find, it certainly created a major improvement in FPS for me in the slower areas. Perhaps my mistake was that I assumed it was only looping through the Spring objects (since SRB2 Lua scripting was still a bit new to me and I was still a bit unfamiliar with how everything worked).
Though the next version will have many more optimizations to things like level geometry and FoF's on top of that. The experiment that is replacing the sides of some FoF's with midtextures is going well, meaning it not only speeds up Software, but it fixes various visual issues as well. I am using UDMF now so it has to wait until 2.2.11 is released.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.
