Why the limitation?

visigoth18 · June 2013

I just wanted to vent for a moment...

1500 Details is ok for a small map, but for medium, large and extra large maps...It's a Joke!

zahinder · June 2013

While I've run into 1500 details (I think my 'dark forest' in the second mission ran to 1400 and change), there is a strong argument to be made that people will find just about any limit too limiting. And it demands some creative vision and developing an ability to scope out just how to present a map to a player to make good use of space and detail.

That said, the wildly diverging cost of stuff can also bite us, like the difference between a dwarven wall (1 cost) and a Blacklake wall (9 or something?)

Oof.
(I'm glad most of what I've used has been dwarven or otherwise cheap)

visigoth18 · June 2013

I find it ludicrous, go take a look at any [Cryptic] map and tell me they only used 1500 points worth of details >.<

x0y1 · June 2013

Very very unlikely and yeah I agree 1500 details is a joke. Everytime I make a medium overland map I run into this problem. The last one was bad. Was building a castle under siege and run out on details very quickly....

And no, I don't like it to split a map in more the on part. I dislike loading screens, worse offender was in CO "Dr. Destroyer's Robot Factory" and it seems Crytpic did not learn from the past.

I hope someday the increase at last the max. based on the map. So large maps can use e.g.2500, medium maps e.g. 2000 ......

zovya · June 2013

Some info from the Devs:

A word from Devs regarding performance mapping

Performance Considerations

Hello Foundry authors! This is an informational post for those looking to get the best performance from their Foundry quests. This post will get very technical, but I recommend reading it for any avid Foundry user, as it will give you a better understanding of performance in Neverwinter.

Draw Calls

Draw calls are one of the biggest bottlenecks for performance in Neverwinter as well as many other modern games. A draw call is when your GPU has to look up a new material for a given set or strip of polygons.

Facts About Draw Calls

1. Every material per-object constitutes a separate draw call, so one object can have multiple draw calls. This also means less materials equals less draw calls.

2. Lighting can affect draw calls as well by breaking object instancing. More info on this below.

3. In general, draw calls are more impactful to performance than polycount, so a high poly object can potentially be cheaper than a low poly, multi-material object.

There is no way for you to tell exactly how many draw calls an object has in the editor, but the budget costs in the foundry were partially calculated from draw calls, so an object with a high cost probably has multiple draw calls. Sometimes you can tell if an object has multiple materials just by looking at it. A building object may have separate materials for the walls, trim, windows, roof, Ivy foliage, and any extra details that couldn’t be worked into a single texture sheet for a single material.

How Lighting Affects Draw Calls

1. In an unlit environment, all identical objects will “instance”. This means they will draw in a single pass on the GPU, all at once. This makes them much cheaper to draw, and you can have many of them with very little cost.

2. Adding a dynamic light will “break” instancing, meaning it will force some of the objects to draw on a separate pass on the GPU. This is generally more expensive and has an impact on draw calls and the amount of time the GPU has to draw stuff.

For example, let’s say you have a row of identical tables in a rectangular room, and one side of the room was lit with a dynamic light while the other side was completely unlit. The tables on the lit side would be in a separate instance group than the ones on the unlit side of the room, effectively increasing the number of draw calls in the room. Each table that the light affects are instanced with each other, but not with the unlit tables.

3. Static lights, or as we say internally “vertex lights”, break instancing as well, but in a worse way. Any object touched by a static light becomes unique, because the lighting data is stored per-vertex, and each vertex is duplicated in memory. Using the same rectangular room with tables example above, if one side of a room is lit with a static light, each table within that light’s radius is now a unique draw call. However, once an object is no longer instanced due to vertex lighting, you can place as many vertex lights on that object as you want and it will not decrease performance. This kind of light can work fine for smaller scope environments, but can be a real problem for larger complex ones.

AI Path Nodes

When you publish a quest, the bulk of the time spent in the publishing process has to do with AI pathing. Our server takes what you’ve built and organizes and compacts the data to a more efficient and readable format. The server then creates a layer of path nodes throughout your map, covering any surface that can possibly be reached by the player or an enemy critter. Essentially, the server starts from your map’s player spawn point position and creates a branching network of nodes that represent the topology of your map. These nodes are required for the AI to function properly, but they also use significant resources on the server.

Facts about Path Nodes

1. Any areas that are completely sealed off with collision from the spawn point will not be populated with path nodes, however areas like the tops of buildings and such do get path nodes. This is because the server isn’t very aware of the intricate flow of your map, so it doesn’t know if there’s a teleporter that takes you to a rooftop or if the player can use a power to knock an enemy into a previously unreachable area.

2. Dense geometry, especially on the ground, attracts more path nodes. For example, a large map with relatively smooth terrain may have far fewer path nodes than a small map with very rugged and rocky terrain. The more geometrically complex your map is, the more complex the path nodes have to be (and the longer they take to publish).

3. Most AI behavior requires path nodes to function properly with the exception of patrols, which only require path nodes when their patrol path is obstructed. This is why AI pathing does not work well in preview mode… There is simply no data for the AI to use yet.

4. Path nodes eat up resources on the server. If they get out of hand, they can eat up quite a tremendous amount of resources. For this reason, there are different categories of maps that the server handles in different ways. An adventure zone is allowed more resources than your typical instanced quest map, but zones also allow many more players in the map at the same time. They are also able to create multiple optimized instances of that zone when the player cap is reached, and some data and resources are shared between these instances. Quest maps do not have these resource-saving features enabled, nor do they have the larger resource allocation from the server that zones enjoy.

Path nodes typically do not affect things like the framerate on the client, but they can cause stalls on the server if they are over budget limits. As an example, in previous beta weekends, there were some quests that used entire zone exteriors for some maps. Since those zones were built with a different budget cap in mind, the server wasn’t expecting such complex maps to be used for instanced quests. As a result, every time those maps were opened by someone playing the quest, the server would have to quickly allocate additional resources to those maps. This was causing a solid 3-5 second freeze for all players on the server every time one of those maps loaded, which just made the entire game significantly less stable.

Optimizing Your Map for AI Path Nodes

You don’t really have a lot of control over path nodes, but there are some things you can do to help your map will work well with them.

1. Avoid excessive amounts of objects with complex geometry in walkable areas. A pile of rocks will have many more path nodes covering it than flat ground. If your entire map is one big debris field with extremely uneven rough topology, it will attract significantly more path nodes than a map with simpler topology.

2. Seal off non-playable areas with collision walls. If you don’t intend for a player to be able to reach a spot on your map, take the time to completely seal it behind invisible walls to ensure it does not get path nodes.

3. Objects that do not have collision, such as ground decals and grass, do not add to the path node count.

4. Placing a collision ceiling as low as possible on your map will reduce the number of nodes. This works best on maps where you have high places and rooftops that the player never gets to and that can be excluded from the playable area.

5. Break your map into multiple maps if you can. This not only improves how the server handles your content, but can also make the AI function better in some cases.

6. Interiors tend to use far fewer path nodes than exteriors, and are also quicker to load for the player. All of the above tips still work for interiors.

This is a lot of information to consider, but hopefully now you have a better understanding of map performance both on the client and on the server.

meedacthunist · June 2013

visigoth18 wrote: »

I find it ludicrous, go take a look at any [Cryptic] map and tell me they only used 1500 points worth of details >.<

The difference between our maps and Cryptic ones it that our maps are assembled from many placeables while Cryptic maps are very likely baked as a one thing. It do matter considering performance and best proof is that Foundry maps have usually longer load time than Cryptic premades (not counting PE because it's crowded by players and their models).

Personallty I don't like building maps from the scratch. Witch hunt has quite detailed areas and no single map is even close to 1000, let alone 1500, and it's possible only because it's all made from mixing and matching premades with placeable details. My work map for the next scenario is the largest available winter exterior and it's already no more than 300 and it's after building detailed village. I do not estimate more than 1000 for the whole map, at worst.

Though it also helps that I'm making heavy use of invisible walls.

ovaltine74 · June 2013

1500 details is plenty if you are actually designing. If you are just throwing stuff randomly all over the map, you certainly could run out. Stay away from clusters of objects and hand place every detail. Learn the importance of negative space. You don't have to put stuff everywhere.

visigoth18 · June 2013

zovya wrote: »

Some info from the Devs:

Good post "thanks"

meedacthunist wrote: »

The difference between our maps and Cryptic ones it that our maps are assembled from many placeables while Cryptic maps are very likely baked as a one thing.

I know there completely different...

that's my gripe/ wish "I wish the system didn't build maps piece by piece, but instead as a whole once published...like the [Cryptic] Maps

>.< and I seriously hate that I can't transition back and forth between different maps!

ovaltine74 wrote: »

1500 details is plenty if you are actually designing. If you are just throwing stuff randomly all over the map, you certainly could run out. Stay away from clusters of objects and hand place every detail. Learn the importance of negative space. You don't have to put stuff everywhere.

Seriously...dont you think thats a bit...

I've been working on one of the largest outdoor maps available "I'm building a Dwarven City" and I have just barley scratched the surface of what I want to do and I only have maybe 300 points left.
each item is individually laid... Try building 'Actual' buildings with legos and come back and tell me 1500 is more then enough, I never "randomly throw stuff"

"No one needs to tell me the Logic behind why it is they way it is 'I do get it' I just get frustrated not being able to go full out >.<"

Restrictions/ Limitations make me sad ;,(

lonnehart · June 2013

visigoth18 wrote: »

I've been working on one of the largest outdoor maps available "I'm building a Dwarven City" and I have just barley scratched the surface of what I want to do and I only have maybe 300 points left.
each item is individually laid... Try building 'Actual' buildings with legos and come back and tell me 1500 is more then enough, I never "randomly throw stuff"

"No one needs to tell me the Logic behind why it is they way it is 'I do get it' I just get frustrated not being able to go full out >.<"

Oh wow.... I was working on building an old mansion on a hill and I'm definitely taking tips/notes from that.

I'm "cheating" by the way. With the limits in place what I'm going to do is create a facade of the place. You can't actually enter and if you entered the structure you'd find nothing inside. This how The Elder Scrolls Oblivion did their maps. If you were to fly into an Oblivion city on the overhead map, you'd find nothing but simplified cardboard cutout buildings as you're not meant to fly in that game.

After building the facade (closed off by a static gate), I can create a new map for the interiors. And I have a 15 map limit to work with (apparently that's how many maps a quest can contain), though this mansion probably won't take up that many maps.

Some great games, such as the Resident Evil Remake for the Gamecube also cheat in this way. All those nicely detailed rooms you saw in the game? They were high detail bitmaps with some invisible polygons that controls where your character can go... hence why the game used static camera angles.

visigoth18 · June 2013

:] I wold love it if we could transition back and forth between maps "like how teleporters work" that would be awesome >D

lolsorhand · June 2013

visigoth18 wrote: »

:] I wold love it if we could transition back and forth between maps "like how teleporters work" that would be awesome >D

^ That

, even if you just had to duplicate an area. ( Which you can't? Atleast not as far as i know. ) I can't even duplicate one of my maps, let alone a single area. Oh well.

zovya · June 2013

lolsorhand wrote: »

^ That , even if you just had to duplicate an area. ( Which you can't? Atleast not as far as i know. ) I can't even duplicate one of my maps, let alone a single area. Oh well.

There is something wrong then. You should be able to duplicate a map. You have to add that map to your storyline though.

zahinder · June 2013

The nice thing about map transitions is they function like 'save points' in your quest.
A bad disconnect/outage/spouse agro stinks if you're 4 hours into someone's 5 hour one-map epic...

Map changes also affords you additional ability to focus on particular areas and details. Like maybe you want to start with an overall map of an area, then create additional maps that delete areas the players will not see and add lots of details they will only see close-in.

I'm facing this issue myself, though in my case the map changes were ultimately demanded by lighting (most of which is map-wide) -- I was going to have a single underwater map that starts in brightly-lit shallows and then deep into the dark depths... but, well, that doesn't really work well on one continuous map.
But I also had to admit that the quest flow worked perfectly fine in a relatively linear series of moves into the depths.

eldarth · June 2013

The problem with the map transitions is the never return to the previous map limitation.

In the current quest I'm working on I've had to build an entire mayor's mansion that can be entered and exited at will. All the villagers are likewise forced to "come out of" their houses when doors are "knocked on" in order to have dialogs. I've only used about a third of the exterior map, and have had to use some craggy boulder clusters and pine tree clusters to limit size/views of the unpopulated space. Thankfully my large lake is taking up most of the center of the map as well. I'm at about 1300/1500 and I still need to flesh out a few rooms in the mansion.

ovaltine74 · June 2013

All right, here is my solution. Duplicate your current map and on the duplicate, change the time of day. At some point in your quest have a player enter a building. While he/she is inside, a transition of time takes place and when he exits the building the sky has gone from mid day to dusk. You can then block off areas the player has already explored and refocus your detail in areas the player is going to go. To the player it all appears familiar and the map transition makes sense.

raphaeldisanto · June 2013

Frankly, if the story in your quest relies upon the player being able to transfer between maps at will, you may not have the right -type- of story for the Foundry.

It's a good tool, but it has limits. I have found it best to design stories that work well within those limits.

beeblebrox69 · June 2013

ovaltine74 wrote: »

All right, here is my solution. Duplicate your current map and on the duplicate, change the time of day. At some point in your quest have a player enter a building. While he/she is inside, a transition of time takes place and when he exits the building the sky has gone from mid day to dusk. You can then block off areas the player has already explored and refocus your detail in areas the player is going to go. To the player it all appears familiar and the map transition makes sense.

That's sort of what I did in my quest. I actually have 5 versions of my village: afternoon (normal), evening (very small area blocked in by invisible walls, with a few minor changes hinting that something wicked this way comes), night (village is on fire, trees have fallen down, undead attacking, dogs and cats living together...), morning/aftermath (village mostly in ruins, tightly walled-in area for villagers to say goodbye), denouement (small area to close out story).

I don't know what was more fun--building my little village, or destroying it.

torquedsoul · June 2013

The only problem with using multiple maps is there is no continuity between maps regarding object interaction and potential triggers other than the main quest line. I think multiple maps would be more appealing if I could remember object interaction from previous maps. Using quest items to track actions is still a bit limited as they can't be used to trigger events effectively.

yospeck · June 2013

torquedsoul wrote: »

The only problem with using multiple maps is there is no continuity between maps regarding object interaction and potential triggers other than the main quest line. I think multiple maps would be more appealing if I could remember object interaction from previous maps. Using quest items to track actions is still a bit limited as they can't be used to trigger events effectively.

If you pick up an item, then when you go to the next map you can still have "Required Item" setup, whether it's in your storyline or not. You don't have to use the items that you acquired on that map, they can be for a future map.

meedacthunist · June 2013

For me reason why would be nice to have ability to freely travel between maps, forward and backward, is that it would give me an opportunity to include completely optional small dungeons just for exploring. Like small ice caves completely not linked to my next map, with optional hack&slash and miniboss only for sake of having them. It would be a gimmick, but nice to have. Also, it would allow us to have two optional entrances to the next map (story-related dungeon), so main dungeon playtrough could be made in completely different ways. Currently we have this kind of bottleneck - only one entrance to the next map.

All those small things are adding to replayability.

torquedsoul · June 2013

yospeck wrote: »

If you pick up an item, then when you go to the next map you can still have "Required Item" setup, whether it's in your storyline or not. You don't have to use the items that you acquired on that map, they can be for a future map.

I understand that and that is what I currently do. But it would be nice to be able to trigger based off of some other object interaction from a previous map. For instance, talking to an NPC on the first map can be used as a trigger on the second map.

zahinder · June 2013

What I would REALLY like on that front is simply a few 'quest variables.' Given the amount of space being used already in storing and running quests, it can't be that hard to store, say, a few numbers.

Heck, just a few 1 digit registers (0-9, or alphanumeric?) plus some visibility/enable logic (NPC1 is visible if Register0 = 1, Dialog option 5 is visible if Register1 = 5...)

That would add a HUGE amount of flexibility in a mission, even if we only had a few registers.

oggbrox · June 2013

visigoth18 wrote: »

I just wanted to vent for a moment...

1500 Details is ok for a small map, but for medium, large and extra large maps...It's a Joke!

I think the "50 NPCs"-stuff is very nasty... I got some NPCs who are placed 4-5 times at the map because of the multiple-interaction-"bug". And yeah... I got a medium map and now there is no place for sound-objects anymore... will delete some trees and grass.

eldarth · June 2013

torquedsoul wrote: »

I understand that and that is what I currently do. But it would be nice to be able to trigger based off of some other object interaction from a previous map. For instance, talking to an NPC on the first map can be used as a trigger on the second map.

What's stopping you?

NPC-A drops ItemA when dialog complete.
Map Transition
If PC has ItemA make Door1 visible
If PC does NOT (i.e. Invert if option) have ItemA make Door2 visible

visigoth18 · June 2013

Yeah that's how you can do it with the tools were given, but that dosent mean its the best way "or even a good way" I kept getting comments like "I think 4 quest items at one time is too much 'it fills what little bag space I have'" which is a fair comment on me using four items to control a side objective...

oggbrox wrote: »

the "50 NPCs"-stuff is very nasty...

Yeah I tried to do a full on Undead Siege and that stopped me

x0y1 · June 2013

ovaltine74 wrote: »

1500 details is plenty if you are actually designing. If you are just throwing stuff randomly all over the map, you certainly could run out. Stay away from clusters of objects and hand place every detail. Learn the importance of negative space. You don't have to put stuff everywhere.

I place every item and don't throw stuff around.

But thanks to this bugfix

Previous projects that were working, then stopped working due to budget problems, should be working again. Lights were accidentally costing too much in the Foundry budget.

One of my maps with a former 1500/1500 budget is now down to 1207/1500. That should do it.:cool:

eldarth · June 2013

ovaltine74 wrote: »

Stay away from clusters of objects and hand place every detail.

I wouldn't go quite that far. I've found clusters to be quite useful. One of the nice things is you can go into detail mode and "move" all the individual little pieces around and they remain part of the cluster. For instance, I've taken a cluster of pine trees and cut-n-pasted sections of them into a much longer-thinner area perfect for showing a tree-line along mountainous area where the trees are only in a strip about 50' deep by maybe 500' wide. I can then go back into cluster view and slide my tree "wall" exactly where I want it. I can also copy/paste it/rotate it into other areas.

Edit: Oh yeah, you can also delete individual detail objects in a cluster and recover some budget points and still have everything else remain as part of the cluster.

celantra · June 2013

I agree clusters are a brilliant tool. They use up no more budget then placing individual items and allow you to group a set of objects. There are usually size options that have different counts and you can delete from the counts if you think they are too large.

Why the limitation?

Comments