How do you "pass-through" all remaining tiles not matched by the rules?

[nima (Nima Talebi)]

Good evening,

I have a few questions that I’m hoping someone could shed some light on.

Question 1

I have a rules.txt file with two entires:

% cat rules.txt
rules/p2f.tmx
rules/f2t.tmx

We start with a layer called “prototype” and the first rule will map that input layer to its output layer called “foundation”. This second layer is an intermediate tilemap, however, and that’s where the next rule kicks in and using that as its input layer, generates the final product, which is called “template”.

This works well using the autotile feature and the rules, however it is lacking in one functionality that I can’t figure out.

p2f: every tile is mapped either to the “finished product” which we expect to see in “template”, or it maps it to a “special” tile.

f2t: all “special” tiles are now mapped to their “finished product”, however, all other already-finished tiles now need to be mapped over to the template layer as-is, like an identity function to map all remaining (not matched by any rule) tiles as-is to the final layer.

So to recap, we have two rules, and there layers:

Layers:
0. prototype (we edit this with a single special tile, let’s call that tile A)

1. foundation (this is the intermediate layer to allow for me to get to the final desired tilemap, it contains either the finished product tiles, or a new special tile, let’s call that tile B)
2. template (this is the final product tilemap that the game engine will look to for rendering and collision data; it should contain zero special tiles)

Here is the first layer, prototype; this is the hand-drawn part:

Next, here is the first generated layer, foundation, which serves as both the first layer’s output, and the second layer’s input:

Finally, here is the third and final layer:

Now that if we flatter the last two layers, we would get the desired tilemap, however, as can be seen by the darkened layer 2 beneath layer 3 tiles, the second layer’s non-special tiles never made it to layer 3.

What rule would allow for this? I’ve tried all sorts of things, all to the tune of trying to mimic this behavior:

1. for all “other” tiles unmatched by any other rule in layer 1, copy
2. in layer 2, paste directly.

---

Question 2

Are rules aware of the border of the tilemap in any capacity? For example can we say in a rule, any tile next to the border, or in one of the four corners of the tilemap?

[eishiya]

Question 1: I’m not sure I understand why you’re doing things the way you are, so please forgive me if this doesn’t quite answer your question.
In general, if you want to copy a tile to another layer, due to the lack of “generic” output tiles, you will need a rule for every possible tile. So, for each slope, you’d need input_foundation → output_template rules that just copy the tile. It would probably be simpler to instead output those tiles directly as part of the rule that populates the foundation layer, i.e. have your rules that deal with the prototype layer also output the slopes to the template layer, and then your foundation → template rules can modify them if need be.

Question 2: No. Depending on the contents of your map, you may be able to identify borders via the various options for matching outside the map, but in the general case, it’s not possible to know whether a tile is at the border. There’s a feature request for it on GitHub: Automapping: Explicitly detect map edges · Issue #3858 · mapeditor/tiled · GitHub For now, if you need to know where the border is, you could use a guide layer to let you know. For example, if you use MatchOutsideMap with no other settings and have the guide layer filled with some tile, then any cell that has a tile is inside the map, and any empty cell is outside the map. You can populate the guide layer with Automapping as well (a simple input_guide Ignore → output_guide SomeTile), since Automapping can’t populate anything out of bounds. Of course, this only works for finite maps.

[nima (Nima Talebi)]

Quick update 1/2 to add some context:

Rule 1 Input/Output

Input:

Output:

[nima (Nima Talebi)]

And updattet 2/2:

Rule 2 Input/Output

Input:

Output:

The aster tile is my feeble attempt at resolving Q1; that is, an Ignore ⇨ Ignore, though I’ve tried I think almost all other sensible permutations of the automat tiles.

[nima (Nima Talebi)]

Thank you @eishiya, I’ve read so many of your comments to date, first time I’m posting here and you’ve responded so quickly; thank you!

Question 2 — right, that makes sense.

Question 1 — I did try to have my rules do this:

Layer1 ⇨ { Layer 2 AND Layer 3} via Rule 1, followed by
Layer2 ⇨ { Layer 3 } via Rule 2

That is what you’re suggesting. If that should work, then that would resolve my issue. Last I tried this, I kept getting a blank template layer, so thought maybe you shouldn’t have multiple rules share output layers.

I’m going to give that a try, since it sounds like it should be possible, which means I probably did something wrong.

[eishiya]

{input_prototype → output_foundation, output_template} should work just fine, I have a bunch of rules that modify multiple layers. Make sure you don’t give them different output indices or something. Or perhaps you had some other error in the rules or outputs. You would need the slope tiles in both output layers, perhaps you only put them in one?

[nima (Nima Talebi)]

Will test this out tonight and report back, thank you again.

[nima (Nima Talebi)]

I’m trying the one-in, two-out approach, what I notice when defining my layers this way:

output2_template // (L3)
output1_foundation // (L2)
input_prototype // (L1)

Is that any application of the input rule component seems to get some probabilistice chance of rendering to output1, or output2, but never both. I don’t have exclusive rules for L1⇨L2 and L1⇨L3, rather I have 3 layers all superimposed, and expected L1 to render one set of tiles to L2, and another set to L3.

Is that not going to work?

[nima (Nima Talebi)]

Here is one component, for example, from the rule:

L1:

L2(+L1):

L3(+L1):

Sorry for multi-posting, I’m a new user here so restricted in 3 max images per post.

[eishiya]

You did the thing I told you not to do: you added different indices to the outputs. Because you want both to output at the same time, they need to have the SAME index. Just output_, remove the 1 and 2.

Tiled picks a single index to output (and as of a recent version, outputs with no index are always output, even if other indices are present). If you have 1 and 2, you’ll get either 1 or 2. If you want both, they need to have the same index. Output indices exist specifically to randomise output. You don’t need to use indices at all for your rules.

[nima (Nima Talebi)]

I apologize, I did exactly the opposite, I think I read what I expected you to say, instead of what you actually said.

Thank you again!