From what I understand the engine uses only up facing polys to create the path, and absolutely discards down facing ones, that can be used for "special rendering" effects like the grass one, or for cam blocking using obscuring terrain. as I said, just be careful of the XY coordinates, and you'll get no issues. I am sure your method is as good as this, I just hate notepad and ascii edits and prefer shortcuts
Tileset Dynamic Grass Question
Débuté par
Pstemarie
, août 18 2012 03:30
#26
Posté 19 août 2012 - 03:48
OTR: I dunno how does it look in renderaabb, and I cant test cuz I already trashed the model ahah!
From what I understand the engine uses only up facing polys to create the path, and absolutely discards down facing ones, that can be used for "special rendering" effects like the grass one, or for cam blocking using obscuring terrain. as I said, just be careful of the XY coordinates, and you'll get no issues. I am sure your method is as good as this, I just hate notepad and ascii edits and prefer shortcuts
From what I understand the engine uses only up facing polys to create the path, and absolutely discards down facing ones, that can be used for "special rendering" effects like the grass one, or for cam blocking using obscuring terrain. as I said, just be careful of the XY coordinates, and you'll get no issues. I am sure your method is as good as this, I just hate notepad and ascii edits and prefer shortcuts
#27
Posté 19 août 2012 - 04:51
OTR - do all of the grass meshes you produce using your method reder using renderaabb
#28
Posté 19 août 2012 - 05:16
They do. The real heart of my question wasn't so much about whether they render in renderaabb or the color it chooses to display the faces...but what the actual positions of the triangles in the walkmesh were, according to it. If you add something and it messes with what the triangulation should be on the walkmesh, I'd consider that as something which was likely going to be problematic down the road.henesua wrote...
OTR - do all of the grass meshes you produce using your method reder using renderaabb
That's not to say it certainly would, though. 1.69 is a lot more forgiving about that kind of stuff. Not necessarily the toolset, though.
#29
Posté 19 août 2012 - 05:50
With regard to the downward facing triangles thing, you should be careful with placing them as if it's possible to get the camera underneath them (for instance if they're atop a raise/lower or cliff) it's likely to get trapped. Placing them slightly underneath the main walk mesh rather than on top is a good idea.
#30
Posté 19 août 2012 - 05:59
Six:downward grass doesnt block camera, i think only cam blocking is done by obscuring type, and if you create them, you already should have them carefully modelled for that blocking effect you are after
#31
Posté 19 août 2012 - 06:27
Meh. This flipping verts/faces and so forth all sounds so complex - much more than I care to dive into. Anytime I have a tile with two woks it crashes the toolset. IMO, all this work for a little splash effect isn't worth it for me - its far easier to just render the grass and forget about the splash vfx.
#32
Posté 19 août 2012 - 07:35
Cool beans anyway, we discovered something new today!
#33
Posté 19 août 2012 - 07:41
Pstemarie, we should change your handle to Pshaw!
#34
Posté 19 août 2012 - 08:02
The real coolness would be cloning the grass walkmesh type to get multiple styles of grass to throw in the tilesets
#35
Posté 19 août 2012 - 09:08
henesua wrote...
Pstemarie, we should change your handle to Pshaw!
I had to look that word up
#36
Posté 19 août 2012 - 09:26
having different types of grass is something i already thought about. Whenever i look at some village, rural or whatever landscape image there is at least 2 variations that stand out. I want a dominant green and some higher yellowish grass in one tileset. And to cover so many spots with placeables is not the ideal way for me. Not sure if that will work one day or how thats doable. But your discovery is great anyway and guys, thanks for your ambition:)
#37
Posté 19 août 2012 - 11:21
Wait... its possible to have two types of grass in NWN?
That would be awesome. Thats... thats almost enough to motivate me to cancel work on Arnheim and make some grassland tilesets.
That would be awesome. Thats... thats almost enough to motivate me to cancel work on Arnheim and make some grassland tilesets.
#38
Posté 20 août 2012 - 03:59
That would seem to be quite a challenge! Is this something someone's already done?
I'm kind of embarassed that I can't remember the filename but I could have sworn that grass was a six-vert rectangular model with danglymesh, and the model is created and destroyed in a relatively narrow area around the player, but only around them, and on the client side.
You can change the grass texture in the set because of the change in 1.67 but because it's a model, the sneakiest you could get (AFAIK) is attach a TXI to the grass texture you name in the .SET file. But to do what, I wonder?
Since there's no way (again, AFAIK) to start on a random frame using a TXI with proceduretype cycle (you can do it with emitters), you couldn't use self-randomizing emitters. However, if you were willing to give up the "creature wind" (i.e. danglymesh) of the grass, you could embed self-randomizing emitters into the geometry of each tile.
Even that I don't think would look very good, though it certainly could be used in really narrow cases to maybe make automatic variations on moss or something like that. You could maybe do a mix of regular tileset grass and embedded self-randomizing emitters. That might look better.
OnEdit: Or would the static nature of tiles get in the way of the self-randomizing emitter firing?
I'm kind of embarassed that I can't remember the filename but I could have sworn that grass was a six-vert rectangular model with danglymesh, and the model is created and destroyed in a relatively narrow area around the player, but only around them, and on the client side.
You can change the grass texture in the set because of the change in 1.67 but because it's a model, the sneakiest you could get (AFAIK) is attach a TXI to the grass texture you name in the .SET file. But to do what, I wonder?
Since there's no way (again, AFAIK) to start on a random frame using a TXI with proceduretype cycle (you can do it with emitters), you couldn't use self-randomizing emitters. However, if you were willing to give up the "creature wind" (i.e. danglymesh) of the grass, you could embed self-randomizing emitters into the geometry of each tile.
Even that I don't think would look very good, though it certainly could be used in really narrow cases to maybe make automatic variations on moss or something like that. You could maybe do a mix of regular tileset grass and embedded self-randomizing emitters. That might look better.
OnEdit: Or would the static nature of tiles get in the way of the self-randomizing emitter firing?
Modifié par OldTimeRadio, 20 août 2012 - 04:04 .
#39
Posté 20 août 2012 - 09:31
If the grass is a model wouldn't it be just as easy to add it to the water tiles where one wants the grass to appear?
#40
Posté 20 août 2012 - 06:25
#41
Posté 20 août 2012 - 06:33
IMO, it depends on how many tiles you're doing it to and how densely you embed them. Without the grass manager (which only shows the grass out to where you set it on the "Grass Far Render Distance" in your nwn.ini) auto-culling the grass, the impact from all those dangly+transparent models hanging around could stack up.Pstemarie wrote...
If the grass is a model wouldn't it be just as easy to add it to the water tiles where one wants the grass to appear?
Static geometry with a 24-bit texture is godly. Dangly with a 24-bit texture is still pretty cheap. But when you start multiplying danglies with transparencies, I think it might cause more of a headache than it's worth. All 3 are running in the scenegraph on the GPU but danglies+transparencies is putting a lot (relative to my first two examples) of transforms down per mesh, per frame.
But like I said, it really all depends on what kind of density you place them with and also probably what the player's "wind" settings are and whether they're using anti-aliasing and a bunch of other things.
Very nice looking screenshot!
#42
Posté 22 août 2012 - 07:32
Well OTR, I attempted to download your sample, got hit with several popups, and finally the page was completely blocked by my avg anti-virus program as a known and bad phishing site. That may explain why you are not getting downloads.
#43
Posté 22 août 2012 - 07:51
i got the file 2 days ago i think. My AVAST didnt reported anything back the day and i had no popup. I´m using firefox 14.





Retour en haut







