DOC TDB – Heuristics underwent a thorough refactor that isn’t documented yet. See the example project!
Heuristics Attribute allows fine-grained and precise control over pathfinding constraints by leveraging user-defined attributes.
When dealing with modifiers, keep in mind that lower values are considered better than higher ones by the
⊚ Search algorithms.
Checklist
Properties
| Property | Description |
|---|---|
| Reference Weight | The reference weight is used internally by heuristics to “adapt” its scale against modifiers. In other words, if the heuristic module itself was outputing a value like a modifier, this would its Weight property.
|
| Modifier List | A list of individual modifiers. Their values are accumulated per Source, and the final sum is used by the ⊚ Search algorithm. |
Individual Modifier
Properties
| Property | Description |
|---|---|
| Enabled | If disabled, this modifier will be ignored by the pathfinding process. This is basically a QoL toggle to facilitate experimentation and better understand the impact of a specific modifier when they start pilling up. |
| Source | Source define if the attribute value of the modifier is fetched from the Vtx point data, or from the Edge point data.See  Pathfinding.
|
| Weight | The weight of a modifier represent its final maximum value. |
| Selector | Attribute to read modifier value from. Reads a double.See  Attribute Selectors.
|
| Local Weight | If enabled, the weight used for the modifier is fetched from a local attribute (same Source` as base value), allowing for per-point weight. |
| Score Curve | The score curve is a key control element of the final modifier value. It allows your to fully leverage the range of the input attribute value in any way you see fit. More infos below. |
Weight
Under the hood, the value of a modifier is remapped from its min/max to 0-1. That remapped value is used to sample the (Score Curve)[#score-curve] and then that value is multiplied by the weight. Depending on the selected Search, this value is post-processed to ensure it is absolute (non-negative)
Score Curve 📌
The score curve is a key control element of the final modifier value. It allows your to fully leverage the range of the input attribute value in any way you see fit.
The curve is expected to stay normalized in the 0-1 range on the
yaxis, and is sampled in the 0-1 range on thexaxis. Just keep in mind the resulting sampled value will be made absolute (non-negative), and is used as a multiplier.
Examples
Say we set-up a point attribute modifier using MyWeightAttribute, which has raw values ranging from -50 to 50, with a weight of 100.
- Raw values will be remapped to 0-1
- The curve will be sampled, then multiplied by the weight.
A - Linear
This is the default, linear curve. It doesn’t modulate weight, so the final weight will closely match the attribute range.
| Raw attribute value | → Remapped to | → Sampled at | → Final weight |
|---|---|---|---|
| -50 | 0.0 | 0 | 0 |
| -25 | 0.25 | 0.25 | 25 |
| 0 | 0.5 | 0.5 | 50 |
| 25 | 0.75 | 0.75 | 75 |
| 50 | 1.0 | 1.0 | 100 |
B - Expo
Exponential curve is very useful to emphasize higher values and “ignore” lower ones.
This can be especially useful when the attribute is for example the edge length, the resulting weight would strongly discourage long edges
| Raw attribute value | → Remapped to | → Sampled at | → Final weight |
|---|---|---|---|
| -50 | 0.0 | 0.0 | 0 |
| -25 | 0.25 | 0.01 | 1 |
| 0 | 0.5 | 0.05 | 5 |
| 25 | 0.75 | 0.25 | 25 |
| 50 | 1.0 | 1.0 | 100 |
C - Inverse Expo
Inverted curve (exponential or not) curve is very useful to, well, inverse the weight. This is especially handy if you want to discourage smaller values and ignore higher ones.
| Raw attribute value | → Remapped to | → Sampled at | → Final weight |
|---|---|---|---|
| -50 | 0.0 | 1 | 100 |
| -25 | 0.25 | 0.25 | 25 |
| 0 | 0.5 | 0.05 | 5 |
| 25 | 0.75 | 0.01 | 1 |
| 50 | 1.0 | 0 | 0 |
D - Yoyo
This is really just here to make a comparison with other more classic approaches. This resulting weight would ignore extreme values inside the range and strongly discourage anything closer to the center.
| Raw attribute value | → Remapped to | → Sampled at | → Final weight |
|---|---|---|---|
| -50 | 0.0 | 0 | 0 |
| -25 | 0.25 | 0.5 | 50 |
| 0 | 0.5 | 1 | 100 |
| 25 | 0.75 | 0.5 | 50 |
| 50 | 1.0 | 0 | 0 |