Augmenting 2d maps for game engines

[johnvanvliet]

I use "shape form shading " for photos
an example from a different forum i posted a bit back

textured and mesh only


this is the original image
http://photojournal.jpl.nasa.gov/catalog/PIA19899

http://photojournal.jpl.nasa.gov/jpeg/PIA19899.jpg

but this dose not work well for most hand drawn things

some can be done ,after a bit
https://www.cartographersguild.com/s...l=1#post262058


a dungeon type map ?

start with hand painting a displacement map

there are some tools that ?? say ??? they can convert a image to a normal map ???
-- they DO NOT really work well except for a few things
like crazybump and "NormalmapGenerator"
http://www.crazybump.com/ -- NON FREE
http://crazybump.com/buy.html

https://github.com/Theverat/NormalmapGenerator

[ArtbyV]

So the problem is trying to get 3d elements on the map without using multiple meshes. To keep frame rate and such down, they just want to use a single plane as the mesh. So no separate meshes for trees, mountains, hills, or elevation.

I tried Crazy Bump. I have the paid version and I use it for a lot of stuff and it works great. Just not that great for hand drawn maps. I'll have to look into Wilbur. Never heard of it before, but I've seen other people use it on here and they are getting good results.


Sent from my iPhone using Tapatalk

[Redrobes]

So the problem is trying to get 3d elements on the map without using multiple meshes. To keep frame rate and such down, they just want to use a single plane as the mesh.

If thats the reason then its crazy. If you want the frame rate high then your much better off having one tree as a mesh and reusing it rather than to send down one enormous tessellation with trees baked into it. The game images that your showing do not seem like the sort of thing that would cause an issue with the frame rate unless your using hardware that is very slow in which case thats even more of a case of not using a large mess for tessellation.

Regardless of what the mesh actually has on it have you tried some speed tests between rendering trees and one large mesh ? People generally use one large mesh for terrain when the terrain has a lot of detail in it. When the trees are all identical then thats when it makes no sense to make one mesh with virticees that are duplicated over it.

[Katto]

OK, I see.
This is a quick heightmap made from your posted map. Of course you need to add more gradients and variations to the shapes. You can then add an additional normal map for finer details. Creating organic shapes like trees is impossible with a heightmap, but you can add stylized ones using cones for example. Walls and houses should be possible. It all depends on the overall scale. Use a 16-bit greyscale map to avoid stairs effects.
I don't think the one mesh thing makes sense, but it would be interesting to see the results.
Hope it helps

[ArtbyV]

If thats the reason then its crazy.

Well, to be fair, it isn’t the only reason. The final game is supposed to look more like a boardgame. As you can see in the pics, the game pieces that represent units are 2d tokens. They are walking the line between boardgame and something slick with mock 3d elements.

Regardless of what the mesh actually has on it have you tried some speed tests between rendering trees and one large mesh?

I have not done speed tests. I did suggest doing the ground as a single mesh and doing separate meshes for trees, but there was concern that adding those as 3d elements might make the game pieces look bad and there is no desire to replace those with character models. The other issue is how can a flat game token sit in an elevated wood. It’s pre-alpha so this is all style experimentation.

OK, I see. This is a quick heightmap made from your posted map. Of course you need to add more gradients and variations to the shapes. You can then add an additional normal map for finer details.

I appreciate you taking the time. Thanks man.

So do you just do edit the heightmap directly in Photoshop? I never tried editing heightmaps by hand. When I do, I get massive changes to the surface. Maybe I’m too heavy handed, but I feel like I have no control over it.

Here’s a test I did using tessellation in Unity.

bandicam_2016-08-10_23-12-52-395.jpg

Trees look decent, but the mountains are terrible. Still trying to figure out how mountains and hills are going to look. Mountains are impassable, so no problem with token placement, but trees and hills are elevated. Trying to decide if the pieces will break the 2d plane or not. Here is what I was thinking for the hills, but I'm not sure I like this.

hillstest.jpg

[johnvanvliet]

as in post #14

you can separate objects and make a pseudo heightmap
the lighter the black and white tone the higher the pixel

I am ASSUMING that the brown bit is a creak bed and not a road
-- OR is it a road



this can be rendered in any tool ( i use blender )




i used the Gmic plugin in gimp and the "local similarity mask "

[ArtbyV]

you can separate objects and make a pseudo heightmap
the lighter the black and white tone the higher the pixel

That looks great! Much more what I am after. I appreciate you taking the time.

I should just be doing this in Maya with a height map. I don't have the Unity files on my computer to test, and it would make a lot more sense to test this locally.

I am ASSUMING that the brown bit is a creak bed and not a road
-- OR is it a road

Kinda looks like a creek bed, but it is a road. I thinking of switching to bricks or some sort of cobblestone for clarity.

i used the Gmic plugin in gimp and the "local similarity mask "

That plugin looks pretty slick. I'm using Photoshop CS6, but I might just download Gimp and give it a try. Thanks dude!

[johnvanvliet]

Gmic has come a LONG way in the last few years

it is basically my "swiss army knife" tool

[ArtbyV]

So after tons of tests, it turns out that old school flat map is what worked out the best for the game. Though I did learn some techniques with both height maps and tessellation that could be used for another project at a later date. Here's the final map I did for the demo/proof of concept level.

LastBastionMap.jpg

It's not perfect by any means, but I think it works well enough.

Here's it is in game working with the movement, combat, and conversation systems.



It does get rather pixelated in the battle screen, so that will be something that needs to be addressed if the game ever makes it to full production.