Improve quality of coalescing-as-needed; add options to limit feature…

… count (#25)

* Add an option to limit geometry vertex count

* Update docs and changelog

* Track desired feature count and geometry size in strategies

* Remove dead code for a long-forgotten inaccessible option

* Fix the option name in changelog

* Refine coalesce-smallest to only coalesce onto other small features

* Clean polygons before coalescing-as-needed

* Don't accumulate tiny polygon holes as negative dust

* Add the option not to limit the feature count at maxzoom

* Add options to limit feature count more abruptly in each tile

* Revert "Add the option not to limit the feature count at maxzoom"

This reverts commit ace173a.

* Revert "Fix the option name in changelog"

This reverts commit 38dbce8.

* Revert "Add an option to limit geometry vertex count"

This reverts commit d379fdf.

* Remove test for reverted option

* Remove more leftovers from geometry size limiting

* Update changelog

CHANGELOG.md

@@ -1,3 +1,9 @@
+## 2.13.0
+
+* Add --limit-tile-feature-count and --limit-tile-feature-count-at-maximum-zoom
+* Coalesce small features only onto other small features with `--coalesce-smallest-as-needed`, never to large features
+* Clean coalesced-as-needed features before simplifying them, to improve simplification quality
+
 ## 2.12.0
 
 * Add `--drop-denser` option to drop points in dense clusters in preference

README.md

@@ -528,6 +528,8 @@ the same layer, enclose them in an `all` expression so they will all be evaluate
 
  * `-M` _bytes_ or `--maximum-tile-bytes=`_bytes_: Use the specified number of _bytes_ as the maximum compressed tile size instead of 500K.
  * `-O` _features_ or `--maximum-tile-features=`_features_: Use the specified number of _features_ as the maximum in a tile instead of 200,000.
+ * `--limit-tile-feature-count=`_features_: Abruptly limit each tile to the specified number of _features_, after ordering them if specified.
+ * `--limit-tile-feature-count-at-maximum-zoom=`_features_: Abruptly limit each tile at the maximum zoom level to the specified number of _features_, after ordering them if specified.
  * `-pf` or `--no-feature-limit`: Don't limit tiles to 200,000 features
  * `-pk` or `--no-tile-size-limit`: Don't limit tiles to 500K bytes
  * `-pC` or `--no-tile-compression`: Don't compress the PBF vector tile data. If you are getting "Unimplemented type 3" error messages from a renderer, it is probably because it expects uncompressed tiles using this option rather than the normal gzip-compressed tiles.

geometry.cpp

@@ -632,12 +632,12 @@ drawvec reduce_tiny_poly(drawvec &geom, int z, int detail, bool *reduced, double
 				// inner rings must just have their area de-accumulated rather
 				// than being drawn since we don't really know where they are.
 
-				// i.e., this ring (inner or outer) is small enough that we are including it
+				// i.e., this outer ring is small enough that we are including it
 				// in a tiny polygon rather than letting it represent itself,
 				// OR it is an inner ring and we haven't output an outer ring for it to be
 				// cut out of, so we are just subtracting its area from the tiny polygon
 				// rather than trying to deal with it geometrically
-				if (std::fabs(area) <= pixel * pixel || (area < 0 && !included_last_outer)) {
+				if ((area > 0 && area <= pixel * pixel) || (area < 0 && !included_last_outer)) {
 					// printf("area is only %f vs %lld so using square\n", area, pixel * pixel);
 
 					*accum_area += area;
@@ -659,10 +659,13 @@ drawvec reduce_tiny_poly(drawvec &geom, int z, int detail, bool *reduced, double
 					}
 				}
 				// i.e., this ring is large enough that it gets to represent itself
+				// or it is a tiny hole out of a real polygon, which we are still treating
+				// as a real geometry because otherwise we can accumulate enough tiny holes
+				// that we will drop the next several outer rings getting back up to 0.
 				else {
 					// printf("area is %f so keeping instead of %lld\n", area, pixel * pixel);
 
-					for (size_t k = i; k <= j && k < geom.size(); k++) {
+					for (size_t k = i; k < j && k < geom.size(); k++) {
 						out.push_back(geom[k]);
 					}
 

main.cpp

@@ -83,6 +83,8 @@ int cluster_distance = 0;
 int tiny_polygon_size = 2;
 long justx = -1, justy = -1;
 std::string attribute_for_id = "";
+size_t limit_tile_feature_count = 0;
+size_t limit_tile_feature_count_at_maxzoom = 0;
 unsigned int drop_denser = 0;
 
 std::vector<order_field> order_by;
@@ -2851,6 +2853,8 @@ int main(int argc, char **argv) {
 		{"Setting or disabling tile size limits", 0, 0, 0},
 		{"maximum-tile-bytes", required_argument, 0, 'M'},
 		{"maximum-tile-features", required_argument, 0, 'O'},
+		{"limit-tile-feature-count", required_argument, 0, '~'},
+		{"limit-tile-feature-count-at-maximum-zoom", required_argument, 0, '~'},
 		{"no-feature-limit", no_argument, &prevent[P_FEATURE_LIMIT], 1},
 		{"no-tile-size-limit", no_argument, &prevent[P_KILOBYTE_LIMIT], 1},
 		{"no-tile-compression", no_argument, &prevent[P_TILE_COMPRESSION], 1},
@@ -2981,6 +2985,10 @@ int main(int argc, char **argv) {
 					exit(EXIT_ARGS);
 				}
 				break;
+			} else if (strcmp(opt, "limit-tile-feature-count") == 0) {
+				limit_tile_feature_count = atoll_require(optarg, "Limit tile feature count");
+			} else if (strcmp(opt, "limit-tile-feature-count-at-maximum-zoom") == 0) {
+				limit_tile_feature_count_at_maxzoom = atoll_require(optarg, "Limit tile feature count at maxzoom");
 			} else if (strcmp(opt, "drop-denser") == 0) {
 				drop_denser = atoi_require(optarg, "Drop denser rate");
 				if (drop_denser > 100) {

main.hpp

@@ -53,6 +53,8 @@ extern size_t max_tile_features;
 extern int cluster_distance;
 extern std::string attribute_for_id;
 extern int tiny_polygon_size;
+extern size_t limit_tile_feature_count;
+extern size_t limit_tile_feature_count_at_maxzoom;
 
 struct order_field {
 	std::string name;

man/tippecanoe.1

@@ -688,6 +688,10 @@ the line or polygon within one tile unit of its proper location. You can probabl
 .IP \(bu 2
 \fB\fC\-O\fR \fIfeatures\fP or \fB\fC\-\-maximum\-tile\-features=\fR\fIfeatures\fP: Use the specified number of \fIfeatures\fP as the maximum in a tile instead of 200,000.
 .IP \(bu 2
+\fB\fC\-\-limit\-tile\-feature\-count=\fR\fIfeatures\fP: Abruptly limit each tile to the specified number of \fIfeatures\fP, after ordering them if specified.
+.IP \(bu 2
+\fB\fC\-\-limit\-tile\-feature\-count\-at\-maximum\-zoom=\fR\fIfeatures\fP: Abruptly limit each tile at the maximum zoom level to the specified number of \fIfeatures\fP, after ordering them if specified.
+.IP \(bu 2
 \fB\fC\-pf\fR or \fB\fC\-\-no\-feature\-limit\fR: Don't limit tiles to 200,000 features
 .IP \(bu 2
 \fB\fC\-pk\fR or \fB\fC\-\-no\-tile\-size\-limit\fR: Don't limit tiles to 500K bytes

mbtiles.cpp

@@ -372,6 +372,12 @@ std::string stringify_strategies(std::vector<strategy> const &strategies) {
 			any = true;
 		}
 
+		if (strategies[i].feature_count > 0) {
+			state.json_write_string("feature_count_desired");
+			state.json_write_number(strategies[i].feature_count);
+			any = true;
+		}
+
 		state.json_end_hash();
 	}
 	state.json_end_array();

options.hpp

@@ -11,7 +11,6 @@
 #define A_PREFER_RADIX_SORT ((int) 'R')
 #define A_CALCULATE_FEATURE_DENSITY ((int) 'g')
 #define A_INCREASE_GAMMA_AS_NEEDED ((int) 'G')
-#define A_MERGE_POLYGONS_AS_NEEDED ((int) 'm')
 #define A_DROP_DENSEST_AS_NEEDED ((int) 's')
 #define A_DROP_FRACTION_AS_NEEDED ((int) 'd')
 #define A_DROP_SMALLEST_AS_NEEDED ((int) 'n')

tests/loop/out/-z0_-O200_--cluster-densest-as-needed.json

@@ -8,7 +8,7 @@
 "maxzoom": "0",
 "minzoom": "0",
 "name": "tests/loop/out/-z0_-O200_--cluster-densest-as-needed.json.check.mbtiles",
-"strategies": "[ { \"coalesced_as_needed\": 999 } ]",
+"strategies": "[ { \"coalesced_as_needed\": 999, \"feature_count_desired\": 1000 } ]",
 "type": "overlay",
 "version": "2"
 }, "features": [

tests/loop/out/-z0_-O200_--drop-densest-as-needed.json

@@ -8,7 +8,7 @@
 "maxzoom": "0",
 "minzoom": "0",
 "name": "tests/loop/out/-z0_-O200_--drop-densest-as-needed.json.check.mbtiles",
-"strategies": "[ { \"dropped_as_needed\": 999 } ]",
+"strategies": "[ { \"dropped_as_needed\": 999, \"feature_count_desired\": 1000 } ]",
 "type": "overlay",
 "version": "2"
 }, "features": [

tests/loop/out/-z0_-O200_--drop-fraction-as-needed.json

@@ -8,7 +8,7 @@
 "maxzoom": "0",
 "minzoom": "0",
 "name": "tests/loop/out/-z0_-O200_--drop-fraction-as-needed.json.check.mbtiles",
-"strategies": "[ { \"dropped_as_needed\": 811 } ]",
+"strategies": "[ { \"dropped_as_needed\": 811, \"feature_count_desired\": 1000 } ]",
 "type": "overlay",
 "version": "2"
 }, "features": [

tests/muni/out/-Z11_-z13_-O100_--cluster-densest-as-needed.json

@@ -8,7 +8,7 @@
 "maxzoom": "13",
 "minzoom": "11",
 "name": "tests/muni/out/-Z11_-z13_-O100_--cluster-densest-as-needed.json.check.mbtiles",
-"strategies": "[ {  }, {  }, {  }, {  }, {  }, {  }, {  }, {  }, {  }, {  }, {  }, { \"dropped_by_rate\": 4080, \"coalesced_as_needed\": 662 }, { \"dropped_by_rate\": 2973, \"coalesced_as_needed\": 1747 }, { \"coalesced_as_needed\": 4342 } ]",
+"strategies": "[ {  }, {  }, {  }, {  }, {  }, {  }, {  }, {  }, {  }, {  }, {  }, { \"dropped_by_rate\": 4080, \"coalesced_as_needed\": 662, \"feature_count_desired\": 687 }, { \"dropped_by_rate\": 2973, \"coalesced_as_needed\": 1747, \"feature_count_desired\": 755 }, { \"coalesced_as_needed\": 4342, \"feature_count_desired\": 856 } ]",
 "type": "overlay",
 "version": "2"
 }, "features": [