I'm working on a layout for a forest map, I may need advice

Tutorials and discussions about Mapping - Introduce your own ones!
JackGriffin
Godlike
Posts: 3774
Joined: Fri Jan 14, 2011 1:53 pm
Personal rank: -Retired-

Re: I'm working on a layout for a forest map, I may need adv

Post by JackGriffin »

For sure don't try to use tree models. If you want to make this map you are almost surely stuck with your original idea of alpha channel tree walls. There is absolutely no way the engine will handle even a super simple tree model duplicated hundreds of times. It will be a lagfest at the minimum and unplayable if you use any sort of custom lighting besides zonelight.

I think you are overthinking the thing. If the player understands that the forest is "off-limits" then they will just accept the tree line and it won't be visually distracting. Think about all the Nintendo games that use this very dynamic. It didn't ruin those maps/games for you did it? It won't here either.

Edit- forgot to add this. Nord's idea on the tree thing is usable but I ran into all kinds of problems once you add lighting. If you are zonelighting and you won't be using a sun to cast shadows then it's maybe possible but if you use any sort of lighting to create shadows then trick visuals like this really stand out. The human's eye is so damn good at discerning shadows that you just about can't trick that without real effort and anything that approaches uncanny valley trying to recreate it gets noticed super fast.
So long, and thanks for all the fish
User avatar
papercoffee
Godlike
Posts: 10443
Joined: Wed Jul 15, 2009 11:36 am
Personal rank: coffee addicted !!!
Location: Cologne, the city with the big cathedral.
Contact:

Re: I'm working on a layout for a forest map, I may need adv

Post by papercoffee »

What about sprites?
User avatar
Le Impact Frag
Average
Posts: 37
Joined: Fri Jan 29, 2016 9:37 pm
Personal rank: Aftermatch janitor
Location: Kaunas, Lithuania

Re: I'm working on a layout for a forest map, I may need adv

Post by Le Impact Frag »

papercoffee wrote:What about sprites?
You mean forward-facing sprites? That would probably be a nice idea. using it with the previously mentioned wall-of-trees. I'm thinking it should look like something from earlier racing games. here's an example.
However, I'm concerned with the collision with the sprites. That can really complicate this idea...
Spoiler
I'm sorry if the picture doesn't appear. I'm still pretty n00bish with the post editor here :help:
Attachments
Untitled.png
The funniest thing about this particular signature is that by the time you realize it doesn't say anything it's too late to stop reading it.
nogardilaref
Masterful
Posts: 577
Joined: Tue Jun 20, 2017 1:00 pm
Personal rank: ⚋⚊⚌☰⚞⌖⚟☰⚌⚊⚋

Re: I'm working on a layout for a forest map, I may need adv

Post by nogardilaref »

JackGriffin wrote:For sure don't try to use tree models. If you want to make this map you are almost surely stuck with your original idea of alpha channel tree walls. There is absolutely no way the engine will handle even a super simple tree model duplicated hundreds of times. It will be a lagfest at the minimum and unplayable if you use any sort of custom lighting besides zonelight.
That depends on quite a few factors actually.
The renderers are able to render about 20K polys in total before starting to lag in an average PC if I remember my own benchmarking right from years ago, although if the renderers actually used the GPU for most of the workload, this would be around 20x more at the very least with a dedicated graphic card.

Which means that, if we count about 100 polys per tree (which makes up a rather detailed tree already if done correctly), that would be a limit of around 200 trees in view.
Of course, it would be less in practice since it would almost lag and there are all the other models to take into account (players, pickups, weapons, etc).

However, from here, to be able to increase or at least maintain this number, the LODBias can be tweaked so that not all polys are rendered at once, and there's also the VisibilityRadius and VisibilityHeight which might make the farthest trees to simply disappear at one point.

More advanced things could be done, but perhaps go way beyond the scope of OP's objective with this map, such as actually building a custom tree with actual LOD levels: depending on the distance, a different model would be rendered entirely, and at one point, it could be simply be a sheet with a masked texture or a sprite.
This is what is done even in modern games nowadays, although most generate the lower poly versions of the models in an automatic fashion (Simplygon comes to mind).
JackGriffin wrote: I think you are overthinking the thing. If the player understands that the forest is "off-limits" then they will just accept the tree line and it won't be visually distracting. Think about all the Nintendo games that use this very dynamic. It didn't ruin those maps/games for you did it? It won't here either.
I agree that the OP is perhaps overthinking it, although I do sympathize with his perfectionism in a way. :mrgreen:
Using Nintendo as an example is unfair though, the level of detail there and here are completely different. There it's accepted because the rest of the game is not very different visually wise, but in UT you have a lot more detail overall, so when a Nintendo-like tree line stands out, it's actually very distracting in an ugly way (to me at least).
JackGriffin wrote: Edit- forgot to add this. Nord's idea on the tree thing is usable but I ran into all kinds of problems once you add lighting. If you are zonelighting and you won't be using a sun to cast shadows then it's maybe possible but if you use any sort of lighting to create shadows then trick visuals like this really stand out. The human's eye is so damn good at discerning shadows that you just about can't trick that without real effort and anything that approaches uncanny valley trying to recreate it gets noticed super fast.
Shadows are tricky, but not that tricky.
If the density is high enough, a darkened floor or something to cast a big shadow (like a mover) should be more than enough, if it isn't, perhaps a nice trick is to open the masked texture in a program, and turn the non-masked parts in a dark grey solid color, and turn the masking color to gray (RGB(127,127,127), I think) and then use it to be "fake" shadows cast by the trees using a modulated style. :)
JackGriffin
Godlike
Posts: 3774
Joined: Fri Jan 14, 2011 1:53 pm
Personal rank: -Retired-

Re: I'm working on a layout for a forest map, I may need adv

Post by JackGriffin »

I'm just curious... Are you one of these guys that just enjoys contrarianism because nearly every post you make is you disagreeing with the person above you (in great detail no less)? We get it, you are intelligent so you can quit trying so hard to convince everyone of it. Pipe down a bit and let people have opinions and viewpoints and stop trying to systematically dismantle them.

I'm adding you to my ignore list. It's just too much to read when I can simmer all your replies to "I disagree and I'll explain why using big words so I sound authoritative". It's not our fault you got your ass kicked in school, stop trying to make up for it.

/s
(or not)
So long, and thanks for all the fish
nogardilaref
Masterful
Posts: 577
Joined: Tue Jun 20, 2017 1:00 pm
Personal rank: ⚋⚊⚌☰⚞⌖⚟☰⚌⚊⚋

Re: I'm working on a layout for a forest map, I may need adv

Post by nogardilaref »

If you already added me to your ignore list, you're probably not going to see this, but I am going to retort anyway, even if just as a public record on my side of this.

1 - I do grant that making long ass detailed posts is a terrible trait of mine, which I am actively trying to fix for the longest time, believe me or not.
Any suggestions are welcome. :)

2 - A forum and a community are all about engagement and contributing, and not a chat. This also means that no post whatsoever is exempt from feedback or criticism, regardless if it's just an opinion or the statement of a fact.
Just like you're entitled to your opinion, I am entitled to mine, and if I do not agree with your opinion I am free to reply to it accordingly, as you're free to reply to mine.

3 - The way I see it, you're simply mad at me for actually presenting what may amount to potentially feasible solutions forward that you dictated as not being possible, and since I am just some random dude in the web who appeared out of nowhere, and presented nothing forward of my own to "prove" myself thus far unlike everyone else around, you're dismissing anything I say as just contradicting for the sake of contradicting, and attacking me on top of it.
Maybe if I did show something, your attitude would be different, but that would defeat the whole purpose of having an unbiased, healthy and engaging discussion around any subject at all in this forum.

I honestly expected you to continue the discussion forward in a productive way, even if just for the sake of the OP, and that you would tell me that X wouldn't work because of Y, or propose something better, or even agree with me, who knows, but you decided to halt it completely instead, impeding any useful ideas and knowledge we could trade, just because you felt "threatened" somehow for some reason, when there wasn't any good reason to begin with.
And the really sad part is that I know you're way better than this.

But well... have it your way then. :thuup:
Red_Fist
Godlike
Posts: 2163
Joined: Sun Oct 05, 2008 3:31 am

Re: I'm working on a layout for a forest map, I may need adv

Post by Red_Fist »

Remember, you must see the forest through the trees, :mrgreen:

But was going to say it's not only trees, you should use some rock deco,(or BSP) to also fill in spaces, and some water stream and pools around.

What you HAVE to do is start working it in the editor, you have the basic paths to walk on but when you actually do it, THEN you will see what looks good or works.

I say the outer rim should be BSP rock walls high enough to not see over (not touching the outer wall), then the skybox will help a lot on the walls to look like sky,. Throw a few trees around to break-up the outer wall, or some plants sticking out of the walls.
Binary Space Partitioning
User avatar
EvilGrins
Godlike
Posts: 9668
Joined: Thu Jun 30, 2011 8:12 pm
Personal rank: God of Fudge
Location: Palo Alto, CA
Contact:

Re: I'm working on a layout for a forest map, I may need adv

Post by EvilGrins »

JackGriffin wrote:Are you one of these guys that just enjoys contrarianism
You gotta admit, under some circumstances (possibly not these) that can be fun.
nogardilaref wrote:If you already added me to your ignore list
Wait... we have access to ignore lists?!?

*ahem*

Moving right along...

The DarkForest maps I've been editing have highly realistic trees in them, that come out of the WSWestern_Pack1.u file.

Hope that helps.
Attachments
WSWestern_Pack1.zip
(1.86 MiB) Downloaded 63 times
http://unreal-games.livejournal.com/
Image
medor wrote:Replace Skaarj with EvilGrins :mrgreen:
Smilies · viewtopic.php?f=8&t=13758
User avatar
sektor2111
Godlike
Posts: 6403
Joined: Sun May 09, 2010 6:15 pm
Location: On the roof.

Re: I'm working on a layout for a forest map, I may need adv

Post by sektor2111 »

EG, here is about terrain/brushes not meshes, meshes are gone from sight if you get closer any don't see the center of them. The majority doesn't like them + Bot goes retarded at visibility firing trees like an ass.
This is a blender "brushing spree" (with UED build errors included of course), I keep watching topic because I want to see something finalized and what is doable related to a realistic forest ambience.
Else, speaking about meshes, there are trees 10×cute than that, no worries. I used... very few of them heading to an initial 40 MB map size... for clarification see DM-UTR-Kamath - now THOSE are trees.
JackGriffin
Godlike
Posts: 3774
Joined: Fri Jan 14, 2011 1:53 pm
Personal rank: -Retired-

Re: I'm working on a layout for a forest map, I may need adv

Post by JackGriffin »

Models, even super simple ones will not work well if they are duplicated too many times. This is especially true if you have spots with longer draw distances. I did a ton of online testing for this very problem when I was doing the groundwork for DayZ in UT. I used MH-LostInItime (by permission) and attempted to make pockets of forest within the map using multiple methods. Duplicating simple models works but once you add light actors performance goes to shit. Sprites work but look unconvincing as do X-style sheet trees. Inability to make a convincing forest was a main reason I never pursued the project beyond the pre-work.

Before people start quoting engine poly limits at me (etc) I would urge you to do your own hard online tests beforehand. I'm not particularly fond of saying "UEngine can handle X things" because I've seen that line move a very large deal depending on situation.
So long, and thanks for all the fish
User avatar
OjitroC
Godlike
Posts: 3605
Joined: Sat Sep 12, 2015 8:46 pm

Re: I'm working on a layout for a forest map, I may need adv

Post by OjitroC »

It is very difficult to think of a good forest map and this dearth really confirms the points made by JackGriffin - they are very difficult to make. Most of those with Forest in the map name are sparsely wooded areas, with the trees few and widely spaced, rather than densely treed forests.

I think DM-UTR-Kamah demonstrates a skilful way to avoid the problems of long view distances and lag. Admittedly, the trees are not particularly dense, though the smaller plants are, and used in other contexts or in greater numbers, they can be quite laggy. Most of the terrain in Kamah is small, steeped sided ravines or gullies so that views along them, and upwards and outwards, are very limited. Larger clearing-type areas are used sparingly.

The overall ambiance of Kamah is, of course, tropical but the general approach to the design and layout of the terrain and the placing of vegetation is perhaps something for the OP to consider in the development of his map.
Last edited by OjitroC on Tue Aug 08, 2017 5:01 pm, edited 1 time in total.
Red_Fist
Godlike
Posts: 2163
Joined: Sun Oct 05, 2008 3:31 am

Re: I'm working on a layout for a forest map, I may need adv

Post by Red_Fist »

Does this deco lag problem happen if it's in the view, or if it's the sum total of the map ?

I want to use a lot of barrels and pottery, the map has walls so you can never see them all. But it will be a lot of them.
And they also can be smashed or broken and disappear unlike trees or a lot of polys like a tree.
So far though it is kind of a big-ish map and only one zone.
Last edited by Red_Fist on Tue Aug 08, 2017 4:59 pm, edited 1 time in total.
Binary Space Partitioning
Eviltom
Skilled
Posts: 150
Joined: Tue Jul 12, 2016 12:08 pm

Re: I'm working on a layout for a forest map, I may need adv

Post by Eviltom »

OjitroC wrote:I Most of the terrain in Kamah is small, steeped sided ravines or gullies so that views along them, and upwards and outwards, are very limited. Larger clearing-type areas are used sparingly.

The overall ambiance of Kamah is, of course, tropical but the general approach to the design and layout of the terrain and the placing of vegetation is perhaps something for the OP to consider in the development of his map.
Best advice so far, the engine is old and limited, experiment by all means,but maps always look and play better when you work within the engines imits. Steep sided ravines/cliffs with a few mesh trees/plants, and some masked tree sheets on top of the cliffs will work well
Red_Fist wrote:Does this deco lag problem happen if it's in the view, or if it's the sum total of the map ?

I want to use a lot of barrels and pottery, the map has walls so you can never see them all. But it will be a lot of them.
And they also can be smashed or broken and disappear unlike trees or a lot of polys like a tree.
So far though it is kind of a big-ish map and only one zone.
Its my understanding that If they arnt in view they won't render, and zoning also helps with this but correct me if I'm wrong!

Edited because of double post :ironic:
Last edited by Eviltom on Tue Aug 08, 2017 5:04 pm, edited 1 time in total.
User avatar
sektor2111
Godlike
Posts: 6403
Joined: Sun May 09, 2010 6:15 pm
Location: On the roof.

Re: I'm working on a layout for a forest map, I may need adv

Post by sektor2111 »

ON-Line is the bigger surprise in UT. If things apparently works OFF-Line, you might be bad surprised around folks pointing you as a low skilled one ON-Line. Meshes wherever they are, if are not centered in view field they are gone soon - this is rendering. Meshes are for weapons, pawns, small decoration not for build mountains. Beside this, some of them (decorations) have bStasis, is a sort of resource saver, actor animating does nothing if player is not there - yeah look at monster behind a glass or through a warp-zone ON-line and see it stuck after some time (relevanttimeout) - genius type setup no wonder. Here you can take in account meshes in a skybox - not much deal, Boulders from CTF-Face are not visible ON-line as a sample about that "super duper" CTF map. By forcing actors to be visible if you go over a few hundreds, troubles are showing up shortly - servers using tools and all sort of thing, will cause ugly limits. A good deal is BRUSH not Mesh - if you can do it. UE1 is optimized for modems - 3-5 kb/s and that's why ON-Line channels are Limited.

Again how many polys supports engine ? Many - until actors number (brushes and the rest) is reaching at boundaries and game is crashing in PostBeginPlay - yeah "unlimited" (my ass) - work-space in Editor is not Unlimited and we can use max 64 zones and then everything is mixed in one big lagger. What are for very giant areas ? For nothing. - Want a confirmation ? Look for some map MH-MysticalForest or such. That's utter crap speaking about rendering at long ranges - not gonna ask about traces there how many resources will use - I did not even open that in Editor for details, I removed it shortly.
nogardilaref
Masterful
Posts: 577
Joined: Tue Jun 20, 2017 1:00 pm
Personal rank: ⚋⚊⚌☰⚞⌖⚟☰⚌⚊⚋

Re: I'm working on a layout for a forest map, I may need adv

Post by nogardilaref »

I just created a small test map with lots of trees, so you guys can benchmark its performance if you wish, and report back your fps.
The limit, at least in my current rig, is around 384 trees, with 17 polys each, rendering about 5.5K polys out of around 6.5K (when I use "stat mesh" and "stat fps"), at which it drops a bit from 60fps.

This is 4x less than what I remember from back then, so either I was mistaken in the number, or this new rig of mine is worse than my older one for these old renderers.
I used Chris Dohnal's renderers, both D3D9 and OpenGL, and XOpenGL from Smirftsch. I didn't try D3D10 yet.

The number of lights does not seem to influence this (at least reducing from 5 lights to 1 light didn't seem to make any difference).
Surprisingly, XOpenGL (Smirftsch's renderer) is 2x slower than D3D9 and OpenGL from Chris Dohnal for this specific case at least, which I will try to figure out later why is this, as I expected the opposite.

I will try to create a proper benchmark test later on so we can all make measurements and check what's the current average limit, and clear out some doubts.
The real benchmark will try to test several different environments: many actors with less polys, less actors with more polys, more lights, less lights, etc.
Attachments
DM-PolyTest.zip
(11.61 KiB) Downloaded 61 times
Post Reply