ok guys heres the deal. the new map array has two levels of height rather than one. so we can have more gently sloping hills
now the New GFX engine will take sprites 2x larger than current (128x64, rather than 64x32) and will take 32Bit PNG's RGBA.
so here is my question. do the below hills look better or worse than the current. namely are smooth interiors and straight lined edges better or worse than having it all straight (the current graphics have all the internal bends as straight lines as well as external bends)
another important factor is that the track layout system will be being changed to a easier and more flexible system. so the bottom of the hills being rounded should not interfere with the graphics of the tracks (if using the current layout system they would slightly)
sorry guys i am unsure which max height of tile is correct (hence the two images below :S). the tallest height in the image should be the same as in current OTTD tiles. the shallow slopes in the images should be 1/2 that height.
Alltaken wrote:sorry guys i am unsure which max height of tile is correct (hence the two images below :S). the tallest height in the image should be the same as in current OTTD tiles. the shallow slopes in the images should be 1/2 that height.
Alltaken wrote:oh i was under the impression there was a "correct" height. and i wasn't sure what that height was
but i guess i can come up with a visually ideal height if the height can be re-done.
Alltaken
I don't know if there are the correct one, but I do know if it looks ok when I see it. Maybe we should talk this over before rendering too many "it could be this one" before we actually are sure of what we want
Bjarni wrote:Maybe we should talk this over before rendering too many "it could be this one" before we actually are sure of what we want
well the difference between changing ALL the heights of the tiles. is about..... let me think......30 secs work. select them all and then press S and scale it down a bit then press render again
but yeah i agree we should work out a height, i can render a few options and make it a poll.
In the current graphics engine, a tile is 64 pixels from the left to the right corner. Raising one corner means its height is increased by 8 pixels. (in orthogonally projected space). In the new map where there is space for one level in between, this means 4 pixels y-displacement per height level.
Assume a tile is 1x1 world units in size. Considering that the angle between the screen and the playing field is 60°, 8 pixel displacement on the screen, for a 64 pixel diagonal tile means sqrt(6)/16 ~ 0.153, world units in altitude.
With the new map, this means: A raise of 1 level is sqrt(6)/32 times the length of one side of a tile and a raise of 2 levels is sqrt(6)/16 times the tile width.
(Assuming my calculations are correct...)
Regarding experimenting with different heights:
We want to keep the current unit height as it makes height displacements pixel exact (1 height level = n projected pixels on the screen). Also, for the current zoom levels, (1:2 and 1:4) it is pretty much impossible to use any other height values. 4 pixel displacement for the smallest height difference is exactly 1 pixel in 1:4 zoom. We don't want to use fractional pixel units.
hey oskar, (the guy i wanted to see, coz i know you have the answers)
so you are saying if the tile is 1x1 units. that the height is therefore 0.153 units (the ratio of height to width being 0.153:1)
because if this is the case then the top image that i posted (of the two) is approximatly 40% to high. and that seems weird to me (visually)
i didn't particularly follow your math. but i assume its correct (i could understand it if i wern't so tired LOL)
the length of the sides would need to be passed through the isometric scale in order to find a real world size. then the height can just be counted off.
cos30=a/h (h equals ((32^2)+(16^2))^0.5
h= 35.7770876399966351425467786997
:. a=30.983866769659335081434123198259 pixels
8 pixels in displacement for verticle you say.
so the ratio is 30.983866769659335081434123198259:8
= 1:0.2581988897471611256786176933189
hmm tell me if my math is wrong.
this means that the top image is very close to the correct angles (and my visual test of this leads me to believe it too.
Yeah, my math seems wrong, but so does yours. (The use of a euclidian distance in a projected space feels wrong.) Anyway, here is a new attempt:
- In the world, we can say that a tile has a diagonal of 64 pixels. The orthogonal projection makes this 32 units in the vertical direction on the screen. This is consistent with an angle of 60° (cos(60°) = 1/2).
- 8 pixels on the screen, with an angle of 60° to the ground plane, corresponds to 8/sin(60°) = 16/sqrt(3) pixels of height in the world.
- The 64 pixels diagonal of a tile gives it a 64/sqrt(2) pixel side. The height/side relation is therefore 16/sqrt(3) / (64/sqrt(2)) = 1 / (2*sqrt(6)) ~ 0.204...
are there going to be any sheer verticle edges in the game? One of the things I did like in RCT/Loco was the ability to terraform the environment to a better extent than in TTD
smooth looks very very nice and but i'm wondering whether or not it'll be more difficult to play on.
edit: if the new graphics engine is truely 3D then you'll be able to spin the map round and see the "backs" of the hills right? if this is the case, it'll be much easier to play on.
...Synthetic Intelligent Organism Normally for Infiltration and Dangerous Exploration...
...sionide.net...
select them all and then press S and scale it down a bit then press render again 
