51

Re: Tones Levels (the thread formerly known as "level feedback?")

tones, i just played the new version of runner and wanted to let you know about two new shortcuts i found. using the first down ramp it's possible to bounce back and land inside the tube, and using the same technique you can bounce over to the brown paths and bounce back down to the narrow green strip. you can still take Dave's shortcut over to the teleport tube as well.

I don't think you should, or maybe even can, change these. my two techniques are much more difficult to pull off than the shortcuts in the previous version (i didn't actually finish, just proved they were possible).

i had replays of both but i named the first one with a name already used, but it wasn't actually overwritten (maybe a bug?). i still have the second if you want to see it

It's a taste i have acquired.

52

Re: Tones Levels (the thread formerly known as "level feedback?")

FPS is good on my system (XP/Radeon9800Pro).

Level is very interesting, love trying it...

It is very good training for camera control...

Keep up the good work!

Currently Playing:
Celeste and Electronic Super Joy

53 (edited by themacmeister 2005-11-15 11:13:59)

Re: Tones Levels (the thread formerly known as "level feedback?")

"Runner" level --

OK, made the 3 swiches on the 'cross' with only about 10 seconds to go...

About how much time should I have at this point to finish ? (It looks like I was about 30 or 40 seconds away from the finish (if there *IS* a finish!)...

Currently Playing:
Celeste and Electronic Super Joy

54

Re: Tones Levels (the thread formerly known as "level feedback?")

That is actually the map called "check"
And yeah, it's a bit of a hustle to make the goal in time.
I'm planning onspeeding up the approach pattern of the "cross" platform before you get on it,
less time sitting there waiting for it to land.
There's also a short cut near the end which helps a lot.

55 (edited by Dave 2005-11-15 18:45:24)

Re: Tones Levels (the thread formerly known as "level feedback?")

I played through the first half (or so, I think!) of your new level. It is very fun and quite creative! I got stuck in a rut trying to freestyle the opening section, and I almost pulled off a really great trick - but not quite. I need to finish to level first, before I start that stuff(I can't help it though, since it's so fun!)

I also noticed, you have lots of "straight hills"... Did you make those by hand? They are really great. I was wondering if it would be useful to anyone if I included an option like that in the curve program(maybe it should be a separate program). You specify a length, width, a number of lumps, total rise in the middle, and any desired "slope"/difference in beginning height to ending height... It would be a piece of cake to make it, but I'd only do it if people were interested.

Also, in terms of adding detail without decreasing performance I have had a flash of what I hope is a brilliant idea. Perhaps we need rlk to confirm this, but it makes sense to me that it would work this way : Lumps which are structural are taken into account with the physics engine, and are processed for collision each frame. Lumps which are detail are not processed for physics collisions. Any surface that is textured will be processed by the graphics portion of the engine(EXCEPT invisible textures). Therefore, if those statements are true:

We could use one big invisible/structural lump for the physics part of the platform, and use textured/detail lumps to provide the variety of texturing. For example : any platform with an "edge", like the wooden edges around many of the ones in your new level, could be converted to "detail" brushes with a few simple mouse clicks, and a single invisible one which is still structural can be placed to occupy the entire space of the platform. If I'm correct, only ONE lump will be taken into account for the physics part of the engine, even though 5 (or many more) lumps would be drawn by the graphics part of the engine. This technique, if it's correctly implemented(and I'm not wrong about how the game engine is working), could drastically improve performance in very large maps, since honestly, most of the big levels in Neverball are not going to tax the graphics cards as much as the CPU(except cards which can't handle reflection, but that should be turned off, if that's the case.) Since many of the sides of lumps(all of the "inside" faces) have the invisible texture, anyway, when you think about it, there are not a whole lot of actual triangles being drawn, but the physics code is calculating each individual structural lump and determining whether the ball is touching it or going to have a collision with it.

Would you mind trying this idea out, tones? it should be relatively easy to do in your new level. Just select all of the brushes/lumps that are making up a single platform, change them to detail, and create a new single invisible brush to place in the same spot. If you were noticing performance problems on your machine, hopefully you'll notice if they get better(or worse! but I really don't think that would happen!) after trying this out. (I'd save it as a different filename, and hold on to the one you've got right now, just to be safe.)

Sorry for the long post guys, but at least I don't think it was boring tongue

(edit: I couldn't possibly post that much without messing something up)

56 (edited by paul 2005-11-16 15:01:53)

Re: Tones Levels (the thread formerly known as "level feedback?")

Ok, I finished "check" in 4:27.24. the secret is to keep moving, try not to stop unless you absolutely have to.

I noticed that the grain of the wood textures doesn't line up on the large moving x platform, but i didn't see anything else.

It's a taste i have acquired.

57 (edited by tonesfrommars 2005-11-16 18:36:41)

Re: Tones Levels (the thread formerly known as "level feedback?")

I fixed that misaligned wood grain on the "cross", thanks Paul.
I think you were the first to download this map, there's a more recent version up with little fixes like that one.

I really like your idea Dave, and I'm going to try it out for sure when I have some time.
The only hitch I may run into, (we'll see when I try it) is that the lump count in this level is right around 1000.
What I understand from reading the forums is that there's a static limit on # of lumps per map (in mapc, that is)
I think at one point someone hacked mapc to do away with that limit so they could experiment with
converting Quake levels into Neverball levels.
I wonder if detail brushes are considered actual lumps in mapc?
Anyway, when I first completed the architecture of that level, I compiled, and got a "lump overflow" error.
If your idea pans out, it might be time to pursue limitless mapc again.
To build a really large level like this, I find it's necessary to be able to go all out with fancy texture placements to keep things visually interesting.

Another idea along these lines is to create a custom texture for certain "building blocks" within a given level.
For instance, the wooden borders could be applied to different textures as 16 unit strips along the outer edge of a 128 unit square. This would reduce # of lumps, and also keep things a little less complicated during mapping.

As far as generating slopes, I don't know that such a program would save much time for me personally.
I usually drag a big block out along the desired X-Y, then cut the Z slope out from a side view. It's pretty quick, and I don't have to think in numbers.

I think it would be really useful to make a program that could generate straight slopes along a 45 degree angle.
I would imagine this could be derived from curvec without too much grief..?  or what about "straight" curves at 45 deg?
I end up dragging vertices around when trying to create angled slopes, which sometimes crashes radiant.
The beginning platform with its four arms in "check" is an example of what I mean by angled slope.

(touche, Dave. this post is pretty damn long as well. Why don't we see about adding forum categories to the nevertable. Longest post, most edits, etc tongue)

58

Re: Tones Levels (the thread formerly known as "level feedback?")

the static limits of mapc are easily rewritten. I had to do it in at least three different categories when I was making a level some time ago that had approximately 200 func_trains, 300-400 switches, and more than 1000 path corners. Only Tuxmym has ever seen it, and I'm not sure if it will ever be made public(certainly not before the 1.4.1 dev release...), as it still has some issues... But the point is that I had to overcome those (somewhat artbitrary) static limits, and it's not hard to do at all. Once rlk gives us the 1.4.1 dev release, all of those static limits will be done away with - at least that's what I think he said! tongue

(if you don't know how to use Dev-c++, I'd be happy to modify the limits and send you a working copy of mapc that will allow you to have lots more lumps. Just let me know)

59

Re: Tones Levels (the thread formerly known as "level feedback?")

I'd definitely make it to the top when it comes to most edits...

The 'check' level is awesome. (Here's where I hit the limits of my English vocabulary.) Seriously, I've never seen anything like it, it's very pretty and original, the texturing of the start platform is brilliant. There are, of course, some minor t-intersections in various places... I can provide screenshots if needed, but I'd like to check if I have the latest map first. And I'd like some more coins in it, no idea on where and how, though.

And, just for the record tongue, lumps are limited to 1024.

#define MAXL    1024

60

Re: Tones Levels (the thread formerly known as "level feedback?")

Cool..

Dave, would you mind doing a quick tweek for me on mapc, I'm nervous I'd screw it up.
Also, I'm in OS X, using a .app version of mapc that Luc sent me.
Can I compile mapc in the same way I compiled curvec ?
Then I will use it manually as an executable.

I suspect the answer is yes, in which case, if I could get the modified code, I should be set.

Thanks for the feedaback Parasti, I'm glad you like the level.
Maybe you could tell me which sections you're seeing the t-intersections in, I can look for them.
No need for making screenshots, thanks.
(btw, you probably have the latest version, I only updated once, shortly after posting the first.)

Also, I do plan to add some coins, work out a max coins challenge. Good idea.

61

Re: Tones Levels (the thread formerly known as "level feedback?")

OK, so... there's one on the turf-blue'd moving platform (x : 2816, y : -1304, top view, if that helps), and then there are two at the 'end' of the back-red textured slope (x : -1792, y : - 1112). I also noticed a missing texture on the grey platform that gets connected with the cross (lump located at x : 24, y : 872). Hope the numbers help. (I'm sure I would be far off if I tried to describe them.)

62

Re: Tones Levels (the thread formerly known as "level feedback?")

tonesfrommars wrote:

Dave, would you mind doing a quick tweek for me on mapc, I'm nervous I'd screw it up.

I'd be happy to do it. It is very, very simple to make the changes, though. You wouldn't mess anything up doing it yourself. You just find the line that parasti quoted above, and change the value to twice that much.

tonesfrommars wrote:

Can I compile mapc in the same way I compiled curvec ?

Unfortunately, it's not quite as simple as mapc. Right now, you have to link in the SDL libraries, so that SDL_image can read the width and height of each texture as the map compiles... So if you have ever compiled an SDL program before, then it shouldn't be a problem. If not, then I'm not sure where to even begin on Mac OS. I can send you a new executable for windows, but not Mac.

rlk plans on dropping the dependency on SDL_image (for mapc) by eventually just accessing the jpeg.dll file directly, but of course, by the time we get that version, we won't be dealing with these arbitrary limits!

So, do you still want me to do it? If so, I'll send it(the mapc.c file) over email to you.

63

Re: Tones Levels (the thread formerly known as "level feedback?")

Thanks Dave,
sounds like I can manage the modification to the code.
I will contact Luc and ask him how to go about compiling for OSX.
I don't think it'll be too hard.

I wonder if themacmeister might have some insight into this, as he seems to be well  versed in Linux as well as Mac-based OS....?

64

Re: Tones Levels (the thread formerly known as "level feedback?")

Okay, Luc or themacmeister will probably be able to help you.

I'd still be very interested to learn whether or not your frame rate improves as a result of trying my invisible platform/detail textures trick... Maybe I'll open it up in radiant myself and try it(I get good frame rates anyway, but maybe they'll go up?) I'll let you know if it makes any difference.

65

Re: Tones Levels (the thread formerly known as "level feedback?")

Sorry Tones,

I haven't used a Macintosh for over 2 years, and even then, I was just getting into AppleX11, and compiling stuff.

I assume the .app version you are using was made with X-Code, which leaves me totally out of the equation, as I have *never* used it.

You can get the devel libraries from libsdl.org and you should be able to add -l/path/to/libraries/SDL to a command line compile (or is it a capital L). Anyways, with the help of man pages and Luc, you should be set...

I am still unaware whether CLI programs compiled with xterm, work under Apple Terminal ?!? That's how clueless I am with OSX smile

Currently Playing:
Celeste and Electronic Super Joy

66

Re: Tones Levels (the thread formerly known as "level feedback?")

Hmmm... I just did the changes to about 9 platforms. Here's what mapc tells me after I compiled it.

mapc wrote:

c:\nbdev\neverball>mapc check.map .\data
check.sol (995/5813/$10)
  mtrl  vert  edge  side  texc  geom  lump  path  node  body
    13  2631  5755  1548  5091  5461  1003    32   217     5
  coin  goal  view  jump  swch  bill  ball  char  indx
     1     1     1     0     7     0     1     0 29741

c:\nbdev\neverball>mapc check.map .\data
check.sol (766/5813/$10)
  mtrl  vert  edge  side  texc  geom  lump  path  node  body
    13  2631  5799  1418  5091  5461   932    32   203     5
  coin  goal  view  jump  swch  bill  ball  char  indx
     1     1     1     0     7     0     1     0 29549

The first time is the default level that tones supplied. It has, as you can see, about 1003 lumps. I figured, if I go in and change a few platforms, I'll add about half a dozen lumps and it should still compile. BUT LO! Notice that the lump count has dropped by approximately 70 lumps! Apparently, details lumps do NOT get counted as regular lumps! This is good, I think.

Also, I was pleasantly surprised that the FPS on my machine went up. Noticeably. (from ~85ish to ~100ish) It was smooth in either case, but on weaker hardware, it would probably make a big difference.

tones, I will send you the .map file I made if you wish, but of course, you could easily do it yourself in radiant. It only took me about 5 minutes to do 9 different platforms, possibly it was quicker.

Using my technique, you'll have less lumps in your map, keeping all of the glorious detail, and it will perform quicker. Sounds like an advertisment, doesn't it?

67

Re: Tones Levels (the thread formerly known as "level feedback?")

Thanks, themacmeister, for your response.
I am even more clueless than you in this realm. btw..
How did you learn that which you do know in this realm?
I'm kind of at a loss for how to further educate myself.

OK, Dave. This is great news.
I suppose this means it will be less pressing for me to compile a new mapc right away.
Thanks for offering your map, I will probably start the process myself anyhow, just to be thorough,
and keep track of whats going on.
There are "Filter: details" and "Filter: structural" check boxes in radiant, so we can toggle the detail brushes when building the structure up or vice versa. Oh boy Oh boy Oh boy
I'm thrilled about this development.

Dave, your initiative and creative thinking are really expanding the horizons for those of us trying to create neverball maps.
Thanks, I'm looking forward to a bunch of great new levels from all of us!

68 (edited by tonesfrommars 2006-01-25 10:04:56)

Re: Tones Levels (the thread formerly known as "level feedback?")

Dave, do you have a sense for whether or not "detail" brushes need to be free of T-intersections?
What about T-intersections occuring between visible and invisible brushes?
I broke this rule a bit while designing a level, and didn't see any artifacts during play,
however, I wonder if it's possible that it might affect performance..?  Probably not.
I have a feeling RLK would frown on it, what do you think?

69

Re: Tones Levels (the thread formerly known as "level feedback?")

i have a feeling that the width of a spiral may affect the the smoothness (it might be impossible to make wide curves uniformly smoothe along the inner radius), but i would definitely try it with fewer brushes just to see.

as for performance, with my mobile intel 915GM/GMS,910GML chipset, FPS is barely affected (meaning i can only tell if i've already been playing for a while)

btw, the only visual bugs i see are some kind of t-intersections at the downramp onto the spiral (didn't check the .map to make sure).

It's a taste i have acquired.

70 (edited by Dave 2005-11-18 05:10:49)

Re: Tones Levels (the thread formerly known as "level feedback?")

tones,
I am 100% positive that T-intersections in the detail brushes are a no-no. you will see them.
I am also 100% positive that T-intersections in the structural invisible lumps don't matter one single bit. you won't see them, and they won't affect the physics negatively.
And of course - T-intersections between invisible(structural) and visible(detail) brushes also don't matter.

I'm looking forward to trying your new level. I've been busy trying to solve my problems in Wings3D modifying some of rlk's .obj's. I finally succeeded, though. Yay! I'm updating my current levelset right now, so I'll play your level soon.

71

Re: Tones Levels (the thread formerly known as "level feedback?")

tonesfrommars wrote:

That is actually the map called "check"
And yeah, it's a bit of a hustle to make the goal in time.
I'm planning onspeeding up the approach pattern of the "cross" platform before you get on it,
less time sitting there waiting for it to land.
There's also a short cut near the end which helps a lot.

It also might be fun to include "ricochet" pads. In the upper reaches of the "check" level, there are many straight long paths, which then angle off usually at 90 degrees from the direction you were traveling. It would be fun it there were an angled lump at each direction change which would allow courageous players to see just how fast they could take those corners and stay in control of the ball. It would also allow one to finish the level a little bit quicker. Overall, though, it's a great level.

72

Re: Tones Levels (the thread formerly known as "level feedback?")

Great idea (ricochet pads), I will definitely implement that when I get back to fleshing out "check".
Thanks for the clarification also re: t-intersections.

73

Re: Tones Levels (the thread formerly known as "level feedback?")

Very good work tonesfrommars !
I like check.map very much. It is good-looking and clever.

74 (edited by Dave 2005-11-26 15:53:59)

Re: Tones Levels (the thread formerly known as "level feedback?")

Dave wrote:

Hmmm... I just did the changes to about 9 platforms. Here's what mapc tells me after I compiled it.

mapc wrote:

c:\nbdev\neverball>mapc check.map .\data
check.sol (995/5813/$10)
  mtrl  vert  edge  side  texc  geom  lump  path  node  body
    13  2631  5755  1548  5091  5461  1003    32   217     5
  coin  goal  view  jump  swch  bill  ball  char  indx
     1     1     1     0     7     0     1     0 29741

c:\nbdev\neverball>mapc check.map .\data
check.sol (766/5813/$10)
  mtrl  vert  edge  side  texc  geom  lump  path  node  body
    13  2631  5799  1418  5091  5461   932    32   203     5
  coin  goal  view  jump  swch  bill  ball  char  indx
     1     1     1     0     7     0     1     0 29549

... BUT LO! Notice that the lump count has dropped by approximately 70 lumps! Apparently, details lumps do NOT get counted as regular lumps! ...

I was wrong about 1 thing. Detail lumps do count as "regular" lumps. I cannot explain why the total number of lumps is lower in the one I compiled second... But if you look at the numbers which are in ()'s after check.sol, the first number is the total structural lumps. You can see that it dropped by over 200, which is why the performance of the level increased...

Maybe I accidentally deleted some lumps as I was modifying the level - this would account for the reduction in total number of lumps(1003 to 932...)

Using this technique, I was able to change my holes1.sol from ~600 structural lumps to ~440 structural lumps. I can probably reduce it a little more, too.

Just a little update, that's all.

(EDIT: now my holes.sol is down to ~370 structural lumps. That's about 40% less! Cool...)

75

Re: Tones Levels (the thread formerly known as "level feedback?")

Yippee,

I just finished 'Runner' with 32 seconds remaining.

As I am only an amateur player, I took no shortcuts, and I paused a few times - I believe you can safely take 30sec off the total time for this level. It then makes it more a matter of memory, as well as speed.

What do you think?

Currently Playing:
Celeste and Electronic Super Joy