Eleazar's Terrain Improvements

[zookeeper]

We'd just mass-convert Ggf to Gg^Flo (or whatever string it gets) in all maps. Doesn't sound like a problem, what with wmllint and all.

OK, so to break it down:
* i'd make the new terrain in trunk
* ask somebody to convert the maps with wmllint
* then remove the obsolete terrain

Yes. We might want to leave the obsolete terrain in for a release or two for backwards compatibility, but if that's problematic for one reason or another, then it's not necessary this early in the dev cycle, I think.

There's just one problem with that: we currently can't have more than two layers of terrain: can't have Re^Gr^Li for example, meaning that if grass was an overlay terrain, you couldn't have another overlay, like rocks or flowers on it.

If someone (that'd be mordante) can lift that limitation then that'd be awesome of course.

I tend to think the addition of another overlay layer, while occasionally useful, would add more confusion than benefit. There are less code-intensive ways to get what we need.

A better thing with the grass might be to make the grass itself an "overlay", slightly vertically offset on the tile, so that a grass tile below will stick up and slightly overlap a grass tile above.

Offsetting all grass doesn't change the way grass interacts. The simplest way to make the grass "stick up and slightly overlap a grass tile above" is by remaking the transitions to stick up more on the north sides, which would work well with other tiles too. I haven't touched transitions yet, except to recolor them appropriately, but I plan to make the North edges stick up.

Actually, I think there's some confusion here: I think that, despite Jetrel just talking about grass overlapping other grass, he also meant that the grass would slightly overlap any other terrain or units above it, just like forests do. That is, if you had grass and then a castle north of it, the grass blades would overlap the castle wall. That we could do, and it might look really nice too, as long as the effect is subtle enough (that is, the reeds aren't very tall).

If necessary, we could also use both "flat" grass and "sticking out" grass at the same time (part of the grass would be in the "flat layer", and then there'd be an extra "sticking out layer", with units and neighbouring terrain getting drawn in between those layers).

[Eleazar]

Actually, I think there's some confusion here: I think that, despite Jetrel just talking about grass overlapping other grass, he also meant that the grass would slightly overlap any other terrain or units above it, just like forests do. That is, if you had grass and then a castle north of it, the grass blades would overlap the castle wall. That we could do, and it might look really nice too, as long as the effect is subtle enough (that is, the reeds aren't very tall).

If necessary, we could also use both "flat" grass and "sticking out" grass at the same time (part of the grass would be in the "flat layer", and then there'd be an extra "sticking out layer", with units and neighbouring terrain getting drawn in between those layers).

I'd thought about that, but i don't see how it could work in practice without creating something nightmarishly complicated.

Note the screenshot. Terrains that we might want to overlap on the front edge with grass come down into the southern hex in very different amounts. It's OK with trees, because trees are tall enough to cover them all up. But grass that stuck up a few pixels beyond the edge of mountains and hills would be about as high as the battlements of some castles. And it wouldn't be high enough to reach most unit's feet.

[zookeeper]

Actually, I think there's some confusion here: I think that, despite Jetrel just talking about grass overlapping other grass, he also meant that the grass would slightly overlap any other terrain or units above it, just like forests do. That is, if you had grass and then a castle north of it, the grass blades would overlap the castle wall. That we could do, and it might look really nice too, as long as the effect is subtle enough (that is, the reeds aren't very tall).

If necessary, we could also use both "flat" grass and "sticking out" grass at the same time (part of the grass would be in the "flat layer", and then there'd be an extra "sticking out layer", with units and neighbouring terrain getting drawn in between those layers).

I'd thought about that, but i don't see how it could work in practice without creating something nightmarishly complicated.

Note the screenshot. Terrains that we might want to overlap on the front edge with grass come down into the southern hex in very different amounts. It's OK with trees, because trees are tall enough to cover them all up. But grass that stuck up a few pixels beyond the edge of mountains and hills would be about as high as the battlements of some castles. And it wouldn't be high enough to reach most unit's feet.

Yeah... You're right, it'd probably be an order of magniture trickier to do than I initially thought, so I guess I can't really suggest worrying about it now.

[thespaceinvader]

The trouble is that in not doing it, we create the appearance that all of our castles are hovering slightly above the grass - this wasn't a problem with the older, shorter grass, but when the grass seems to be between a few inches and a few feet long, it look a bit strange. I think some overlap in the transitions is probably necessary - but doing it right for each terrain would be very difficult. Is there room to have transition tiles around each castle?

[Eleazar]

The trouble is that in not doing it, we create the appearance that all of our castles are hovering slightly above the grass - this wasn't a problem with the older, shorter grass, but when the grass seems to be between a few inches and a few feet long, it look a bit strange. I think some overlap in the transitions is probably necessary - but doing it right for each terrain would be very difficult.

I'm surprised at the importance people are ascribing to grass-overlap -- especially with more obvious faults in plentiful supply.
Maybe the wall (on the right in my last screenshot) looks like it is floating above the grass, but i wouldn't say that about anything else i've looked at.

If the hard lower edges bother people, castles and such could be given some irregular semi-transparency on their lower edges. That might also be beneficial in blending with non-grass base tiles too. But it's not something i'm gonna mess with-- the benefit/effort ratio is not good enough.

Is there room to have transition tiles around each castle?

This question is too vague to answer.

[thespaceinvader]

I'm being dense and asking the wrong thing is why. We blend some of our villages into the surrounding terrain by having small borders of semi-transparent dirt colours, I guess to avoid exactly the impression I mentioned. Could we not do this with castles?

Apologies if I'm asking stupid questions, terrain art is not a strong area of mine.

[Eleazar]

1) Made a tiny mushrooms embellishment terrain. It probably makes more sense in caves, but i don't want to look at cave terrain right now.

2)

I'm being dense and asking the wrong thing is why. We blend some of our villages into the surrounding terrain by having small borders of semi-transparent dirt colours, I guess to avoid exactly the impression I mentioned. Could we not do this with castles?

Nope. The south end of all castles overhangs onto the adjacent terrain, which might be ice, snow, or water, or chasm. Dirt sticking out doesn't make sense in many of those situations.


3) What's the deal with villages? I can't change the base terrain on any of them with the editor. How come i can put whatever base i want under forests but not villages?


4) This wall doesn't transition nicely with anything that sticks up. There's your example of a terrain that floats.

Picture 8.png (192.63 KiB) Viewed 5182 times

5) How do i make a new group in the editor?

[Jetrel]

Note the screenshot. Terrains that we might want to overlap on the front edge with grass come down into the southern hex in very different amounts. It's OK with trees, because trees are tall enough to cover them all up. But grass that stuck up a few pixels beyond the edge of mountains and hills would be about as high as the battlements of some castles. And it wouldn't be high enough to reach most unit's feet.

Yeah... You're right, it'd probably be an order of magniture trickier to do than I initially thought, so I guess I can't really suggest worrying about it now.

No, this is a false alarm. You're assuming all grass would be a big hex-filling solid lump of grass, when we can in fact use essentially the same technique as forests, and have some small variants. Grass would be mixed enough in distribution that this would look fine.

This would be more complicated in terms of needing those small variants, but it'd be much simpler because by having grass based on an underlying terrain like dirt, we'd no longer need to make transitions for grass.

Variance in distribution is one of the biggest improvements we need. Right now, this new grass is a dramatic improvement, but it's still suffering from fairly regular distribution on a wide-scale.


Note: I really don't want to detract from any of these improvements. We don't need to do these suggestions of mine right away, and indeed virtually everything you're doing is very valuable and doesn't conflict with them at all.

[Eleazar]

Yeah... You're right, it'd probably be an order of magniture trickier to do than I initially thought, so I guess I can't really suggest worrying about it now.

No, this is a false alarm. You're assuming all grass would be a big hex-filling solid lump of grass, when we can in fact use essentially the same technique as forests, and have some small variants. Grass would be mixed enough in distribution that this would look fine.

No i'm not assuming that.
If we weren't trying to make it stick up slightly over the south edge of castles and so forth, then yes it would be possible using the forest method. But i see no great advantages to that. Feel free to make a prototype to prove that this can work if you want.

Variance in distribution is one of the biggest improvements we need. Right now, this new grass is a dramatic improvement, but it's still suffering from fairly regular distribution on a wide-scale.

I don't see how this improves variance distribution. 8 tiles made with equal skill that are all grass are going to have the same variance as 8 tiles with holes in them showing a dirt tile underneath. Sure they are technically more varied since different dirt tiles will be hooked up with different grass tiles, but the eye is going to latch onto the borders between grass/dirt, mostly ignoring the relatively smaller differences. One bit of dirt showing through won't be significantly different from another bit of dirt showing through the same hole, unless the dirt tiles are significantly redone.

If variance distribution was really a concern, i could double or triple the number of unique grass tiles with less effort. And it's likely i'll make more after i stare at them for a while.

[Espreon]

5) How do i make a new group in the editor?

Take the following code, adapt it to your desires, and add it to data/core/editor/terrain-groups.cfg:

Code: Select all

[editor_group]
    id=special
    name= _ "special"
    icon="group_x"
    core=yes
[/editor_group]

I really like what you are doing; keep up the excellent work.

[Eleazar]

Thanks.
I committed the new "embellishments" group, and added desert plants to it.

[Eleazar]

I didn't want to get bogged down on this, but the snow had obvious repeating pattern problems. I quickly cleaned it up, and made another tile.

*click to zoom*

[Eleazar]

I made a quick terrain "leaf litter" to replace the olive, "forest floor" since none of the grass colors looked right to be under pine trees. It's not the highest quality, but it establishes color --important for this harmonization-- and is at least as good as what it replaces.

It could be argued that different bases should go under the various forests by default, but what we have now is much closer to it should be.

[Boucman]

And i'm back....

Eleazar, you're new terrains are absolutely great, and the new organization makes so much sense we should have thought of it earlier...

I'm not sure what technical questions you need answering at this point, but for animations, there is the new "builder" system... I convert the top level macros one by one as I need them, but if you plan to animate something, drop me a line and I'll adapt it for you

[Iris]

3) What's the deal with villages? I can't change the base terrain on any of them with the editor. How come i can put whatever base i want under forests but not villages?

I believe this was rejected by the MP developers because of the vastly different stats combinations that could be created this way, in regards to movement costs and defense.