docs: update CHANGELOG, REFERENCE, and ROADMAP for 0.1.2

Author: andrestubbe

docs/CHANGELOG.md

@@ -1,5 +1,16 @@
 # FastImage Changelog
 
+## [0.1.2] - 2026-09-04
+- **Zero-Copy Native Buffer Wrapping**:
+  - Implemented `nativeWrap()` C++ JNI bridge allowing zero-copy instantiation of `FastImage`.
+  - Added Java factory methods: `FastImage.wrap(long nativeAddress, int w, int h)`, `FastImage.wrap(Pointer, int w, int h)`, and `FastImage.wrap(ByteBuffer directBuffer, int w, int h)`.
+  - Safe memory ownership model: wrapped instances do not deallocate foreign native pointers on close.
+- **AVX2 Area-Average Anti-Aliasing Downsampler**:
+  - Implemented native `nativeResizeAreaAverage()` box/area-average downscaling kernel in C++ with SIMD integer accumulation.
+  - Added `FastImage.resizeAreaAverage(int targetWidth, int targetHeight)` to eliminate moiré and jagged aliasing artifacts during high-ratio downsampling.
+- **Ecosystem Integration**:
+  - Formed zero-copy bridge target for `FastScreen 0.1.2`, `FastCamera 0.1.1`, and `FastRobot 0.1.1`.
+
 ## [0.1.1] - 2026-08-14
 - Integrated native `FastSIMD` (v0.1.3) AVX2 256-bit vector image processing engine.
 - Added official JMH benchmark suite measuring 19.5 full 1080p frames/sec resize throughput.

docs/REFERENCE.md

@@ -22,6 +22,16 @@ Converts a Java `BufferedImage` into an off-heap `FastImage` instance.
 
 ---
 
+### `wrap`
+```java
+public static FastImage wrap(long nativeAddress, int width, int height)
+public static FastImage wrap(Pointer pointer, int width, int height)
+public static FastImage wrap(ByteBuffer directBuffer, int width, int height)
+```
+Zero-copy wraps an existing off-heap native memory address, FastPointer, or Direct `ByteBuffer` without copying bytes. The wrapped instance does not deallocate foreign memory when closed.
+
+---
+
 ## 2. Image Operations & Filters
 
 ### `resize`
@@ -32,6 +42,14 @@ Resizes the image to target dimensions using native C++ AVX2 SIMD bilinear inter
 
 ---
 
+### `resizeAreaAverage`
+```java
+public FastImage resizeAreaAverage(int targetWidth, int targetHeight)
+```
+High-quality Anti-Aliasing downsampler using SIMD box area averaging. Eliminates jagged edges and moiré patterns when scaling down desktop or camera feeds.
+
+---
+
 ### `blurKawase`
 ```java
 public FastImage blurKawase(float radius, int passes)

docs/ROADMAP.md

@@ -2,5 +2,6 @@
 
 - [x] **v0.1.0** — Initial release with off-heap C++ image processing and Kawase blur.
 - [x] **v0.1.1** — Integrated `FastSIMD` AVX2 256-bit vector acceleration engine.
+- [x] **v0.1.2** — Zero-Copy Pointer/ByteBuffer wrapping & AVX2 Area-Average Anti-Aliasing.
 - [ ] **v0.2.0** — FastGPU Vulkan compute shader acceleration for 4K real-time blur.
 - [ ] **v0.3.0** — Native PNG/JPEG WebP encoder/decoder integration.

run-benchmark.bat

@@ -2,7 +2,7 @@
 setlocal
 chcp 65001 > nul
 cd /d "%~dp0"
-set "MAVEN_OPTS=--enable-native-access=ALL-UNNAMED --sun-misc-unsafe-memory-access=allow -Dorg.slf4j.simpleLogger.defaultLogLevel=warn"
+set "MAVEN_OPTS=--enable-native-access=ALL-UNNAMED -Dorg.slf4j.simpleLogger.defaultLogLevel=warn"
 
 echo ===================================================
 echo  Building FastImage ^& JMH Benchmarks Uber-Jar
@@ -26,5 +26,5 @@ if %ERRORLEVEL% NEQ 0 (
 echo ===================================================
 echo  Running JMH Benchmarks (Throughput: ops/ms)
 echo ===================================================
-java --enable-native-access=ALL-UNNAMED --sun-misc-unsafe-memory-access=allow "-Djava.library.path=..\..\build;build;src\main\resources\native" -jar target\benchmarks.jar -f 1 -wi 2 -i 3 -tu ms -bm thrpt
+java --enable-native-access=ALL-UNNAMED "-Djava.library.path=..\..\build;build;src\main\resources\native" -jar target\benchmarks.jar -f 1 -wi 2 -i 3 -tu ms -bm thrpt
 pause