Updating The Knighton Manor (by Sotha)

[Tels]

I noticed a few small errors in Sotha's great mission, most significant being the custom materials needing an update for v1.04, and Sotha gave me permission to fix these errors and update the FM.

 

But if You could update Knighton's, that would be great!

 

I hereby grant thee the permission to fix

 

* Texture issues

* Barrell support

* Any z-fighting you can find

 

 

You *can* also add SEED grass, but only if the performance hit is negligible and the additions won't change gameplay or add any kind problems.

 

This thread is for:

 

1. documenting my changes
   
2. asking for people if they have seen other bugs/errors
   
3. ask for some betatesters that make sure that the update isn't broken in any way
   

 

Here is a list of changes so far:

 

• delete unnec. material files, so that f.i. the wood stack in the wooden shack renders properly under all ambient methods
  
• barrel on table in barracks gets support so it doesn't roll
  
• z-fighting beam in passage on yard fixed
  
• added SEED entities that reduce the plants/mushrooms to a single entity (more planned for the small pond)
  

 

One proposed change is to move (or rotate) the well in the garden, as between the well and the hedge it looks like the player fits, but if the guard with the lantern comes around, you find out the hard way that you actually don't fit into the space But I leave the judgement to Sotha.

 

Another thing Sotha said:

 

Technical limitation. At present, we do not have a speaker which the guards can hear and react to.

 

The alarm is a normal sound, which AI do not really hear. What the mappers CAN do at present is to give some idle guards some new targets and that is why some of the guards seem to react to the alarm and some not.

 

Once we get the possibility to have guards to hear speakers, I may update the mission.

 

While I could fix this technically, I am afraid to break the mission by changing too much, so I probably leave this for now. Never change a working thing

 

Anyway, thats it for now, I post screenshots later.

[Bikerdude]

While your at it Tels, the mission was always too clean, needs dirtying up..

[nbohr1more]

LOL Biker, unless he's solved his auto-grimer idea I think you or Fidcal would be better candidates for griming things up

[Sotha]

While I could fix this technically, I am afraid to break the mission by changing too much, so I probably leave this for now. Never change a working thing

 

Thanks Tels!

 

But let's not get carried away. Let us just stick to the things I gave the permission to change. The mission matches quite well my original vision and I do not want to see it deviate too far. I want the knighton manor play exactly it was originally planned.

 

If we get the alarm speakers, they can then be used later in another mission.

 

So let's just stick to obvious errors.

[Tels]

Here are the things that I changed, and will change, along with reasons why/why not:

 

Bugfixes

 

1. Added barrel support (the texture is now darker, this was a first test)
   
2. z-fighting beam
   
3. material was black under some ambient rendering methods
   

 

If anybody has other bugs, please post them in this thread.

 

Optimizations

 

These changes have been done under two assumptions:

 

1. Speed: adding SEED grass will change performance for the worse, so we need improve it a bit
   
2. Changes should be easy, and not break the map or impact anything
   

 

Based on this I have done the following:

 

 

1. Added 3 SEED entities that combine mushrooms, the plants and the ivy at the pond into combined entities (screenshots in next post)
   
2. Aligned a few visportals with their arch-brushes (first shot shows the arch is unnnec. cut into two halfes, second shot shows how it is better. Only done this in case there is no door, of course).
   
3. And converted a few selected brushes into func_statics (floor of two rooms before/after)
   
4. Combined a few brushes that looked like they were accidentily cut up, deleted 2 stray left-over brushes/one func_static (outside of the player-reachable area).
   

 

It is unlikely I will do any further optimizations, because the map is already fast (and small and optimized) and it is simply not worth investing more than one hour.

 

Visuals

 

1. Added some grass (SEED entities).
   

 

It is unlikely that any other visual changes will be done, not even grime decals, because A: I don't want to fix what is not broken, nor ruin the map and B: Performance will get worse and C: Sotha said nothing about this. .

 

Gameplay, AI, Triggers

 

No changes will be done here at all. No path nodes moved etc. Never touch a working system

[Tels]

Here are the screenshots that did not fit into the last post:

 

Optimization done with SEED (4 entities, instead of a 18 * 3 mushrooms plus 13 * 4 plants = 76)

 

A possible optimization would be that scene, the stairs cause excessive triangulation. But I am not sure if it is really nec., the map seems to run fine so far

 

 

Added grass around trees:

 

 

Anyway, except a few more tweaks to the grass entities, the map is done. I am sure it could be "prettied up" and "dirtied up" a lot more, but that is not my task here

 

So who wants to betatest the new version?

[nbohr1more]

Great stuff!

 

I'll play the beta but I cannot guarantee much playtime these days with our new baby around

 

Plus I've got Moddb articles to write...

 

Lowenz and Grayman seem to be beta-test maniacs who can spot any anomaly no matter how minor.

[Tels]

Haha, I noticed that Sotha (as a good mapper should do) reused the same window+window frame model quite a lot in the map. And so I began musing about models vs. func_statics. Normally it doesn't make much sense to convert func_statics to real models (ASE/LWO), but:

 

* if a model is used a lot, then you can save triangles because models are left alone, while func_statics are "cut up" during dmap (e.g. two brushes of a func_static that touch each other cause triangulation of the surfaces). That not only can make dmapping a very lenghty process, but also cause excess triangles (and thus longer rendering if instead just a few pixels had been drawn more, but only 2 instead of 10 tris).

* also, func_statics are cloned in a map file and also in memory during run-time. Models are loaded only once.

 

So I thought about converting the window frame into a model (purely to learn "what does it do and how does it bring") and the very first window I looked was an outside window, and after flying into the room to check out the inside, something didn't seem right:

 

 

[Sotha]

The optimizations look good, although I don't think they'll save so many tris they'll be very significant. If you func_staticified something, be extra sure they are monsterclipped properly.

 

The staircase might be optimized into func_statics, but that would require special rebuilding to maintain the sealing and monster_clipping for the stairs. I don't see the optimization worth it. Plus the monsterclipping and rebuilding may be very error prone. I just don't think the tris saving is worth the trouble. Do note that the upper stair bottoms seen in the image are truly part of the stairs above. That would need really careful work if it was to be optimized. I don't even know it is possible...

 

So I thought about converting the window frame into a model (purely to learn "what does it do and how does it bring") and the very first window I looked was an outside window, and after flying into the room to check out the inside, something didn't seem right:

 

Oh no! I was hoping no one will ever spot that.

 

Okay, yes I admit it:

I place windows so that they look natural, rather than realistic. The window would have looked odd inside the room if it matched with the exterior window.

 

Now there will be an endless complaining about how my missions suck because the windows are not realistically placed, but placed as per the scene seen.

 

I'll take it as a man. (Crawl under the bed and cry.)

 

Also the windows have surfaces on both sides: it is after this mission I learned to CAULK the unseen surfaces BEFORE multiplying the object everywhere...

[fllood]

I place windows so that they look natural, rather than realistic.

Sotha, actually I prefer people build their FM look natural if realism spoil atmosphere, gameplay or need to much build efforts being worthwhile or better spend elsewhere.

[Springheel]

Technical limitation. At present, we do not have a speaker which the guards can hear and react to.

 

I don't understand why not. Isn't it possible to attach a sound that propagates to AI through a speaker? I thought that was done for the statue destruction in St. Lucia? Or is the problem that we just don't have a pre-made alarm sound that propagates to AI?

[Melan]

Okay, yes I admit it:

I place windows so that they look natural, rather than realistic. The window would have looked odd inside the room if it matched with the exterior window.

I have done this before, but always after a lot of agonising and with a feeling of guilt. People probably never noticed it in the end.

[Tels]

The optimizations look good, although I don't think they'll save so many tris they'll be very significant. If you func_staticified something, be extra sure they are monsterclipped properly.

 

The staircase might be optimized into func_statics, but that would require special rebuilding to maintain the sealing and monster_clipping for the stairs. I don't see the optimization worth it. Plus the monsterclipping and rebuilding may be very error prone. I just don't think the tris saving is worth the trouble. Do note that the upper stair bottoms seen in the image are truly part of the stairs above. That would need really careful work if it was to be optimized. I don't even know it is possible...

 

Yes, I saw this only later, you are right, this would be tricky, and certainly NOT worth the effort.

 

Oh no! I was hoping no one will ever spot that.

 

Okay, yes I admit it:

I place windows so that they look natural, rather than realistic. The window would have looked odd inside the room if it matched with the exterior window.

 

Now there will be an endless complaining about how my missions suck because the windows are not realistically placed, but placed as per the scene seen.

 

I'll take it as a man. (Crawl under the bed and cry.)

 

LOL chill, I just found it amusing

 

Also the windows have surfaces on both sides: it is after this mission I learned to CAULK the unseen surfaces BEFORE multiplying the object everywhere...

 

For my amusement I plan on experimenting with a "model" window, at this time I'd also caulk them, too. But not before the weekend, probably.

 

Anyway, as written above, it doesn't really matter as the interiour spaces are fast enough, anyway. The only thing I could think of this would be effecting is dmapping speed - I think that D3 does the same work over and over again for each window (triangulate it, compute surfaces etc).

 

 

Btw, one more peculiar thing: broadheads break if you hit certain parts of the red wooden doors upstairs, but not on other parts. Is this caused by the invisible hinges? I cannot see any difference in game, neither with r_showtris or without. And yet, shooting a broadhead into the upper parts of the door makes it break, shoot the middle and it sticks in...

[Sotha]

Btw, one more peculiar thing: broadheads break if you hit certain parts of the red wooden doors upstairs, but not on other parts. Is this caused by the invisible hinges? I cannot see any difference in game, neither with r_showtris or without. And yet, shooting a broadhead into the upper parts of the door makes it break, shoot the middle and it sticks in...

 

They are prefab doors, so it might be because of the invisible hinges..

[lowenz]

Great stuff!

 

I'll play the beta but I cannot guarantee much playtime these days with our new baby around

 

Plus I've got Moddb articles to write...

 

Lowenz and Grayman seem to be beta-test maniacs who can spot any anomaly no matter how minor.

Ahahahahah

[lost_soul]

I don't understand why not. Isn't it possible to attach a sound that propagates to AI through a speaker? I thought that was done for the statue destruction in St. Lucia? Or is the problem that we just don't have a pre-made alarm sound that propagates to AI?

 

I was thinking you could just have all AIs go to combat mode if ... happens. In terms of realism, the guards would know about where I'm talking about and they would run there to find out what is going on. This is where the player would currently be too, so the guards would thus attack the player unless he moved quickly to a safe location. I know Sotha isn't planning on changin this, but it is just an idea. Even if there is no stim for AI, can't you just set them all (within a certain range) to attack mode?

Edited March 2, 2011 by lost_soul

[Tels]

Yes, I think changing their alert level must be possible, but atm I do not plan on changing this (fear of breaking the map).

 

Back to betatesting, is there anybody who has time and interest in this?

[lowenz]

Yes, I think changing their alert level must be possible, but atm I do not plan on changing this (fear of breaking the map).

 

Back to betatesting, is there anybody who has time and interest in this?

I'm glad to help, good sir

TKM is one of my favourite FMs!

Edited March 2, 2011 by lowenz

[7upMan]

Back to betatesting, is there anybody who has time and interest in this?

 

Once I get my graphics card up and running I'll be able to lend you a hand. How soon do you need me?

[Tels]

@lowenz and 7upman: I *think* next week or so, but depending on whether I find7fix another bug in SEED, I might even delay the map until v1.05 (because it doesn't make sense to work around bugs because the map can be released later without problems).

 

I post more on Monday.

[lowenz]

No problem man, I'm always here

[nbohr1more]

My offer of "some testing" still stands.

 

Can't wait to see what new SEED improvements you cook-up though...