Examples: Add DepthTextureUtils

Author: gchoqueux

examples/files.json

@@ -219,6 +219,7 @@
 		"webgl_postprocessing_afterimage",
 		"webgl_postprocessing_backgrounds",
 		"webgl_postprocessing_transition",
+		"webgl_postprocessing_interactive_displacement",
 		"webgl_postprocessing_dof",
 		"webgl_postprocessing_dof2",
 		"webgl_postprocessing_fxaa",

examples/jsm/postprocessing/DepthSavePass.js

@@ -0,0 +1,60 @@
+import {
+	WebGLRenderTarget,
+	FloatType
+} from 'three';
+
+import { SavePass } from 'three/addons/postprocessing/SavePass.js';
+
+
+/**
+* A pass that saves the depth contents of the current read  buffer in a render target.
+*
+* ```js
+* const depthSavePass = new DepthSavePass( customRenderTarget );
+* composer.addPass( depthSavePass );
+* ```
+*
+* @augments SavePass
+* @three_import import { SavePass } from 'three/addons/postprocessing/SavePass.js';
+*/
+
+class DepthSavePass extends SavePass {
+
+	constructor( renderTarget ) {
+
+		if ( renderTarget === undefined ) {
+
+			renderTarget = new WebGLRenderTarget( 1, 1, { type: FloatType } ); // will be resized later
+			renderTarget.texture.name = 'DepthSavePass.rt';
+
+		}
+
+		super( renderTarget );
+
+	}
+
+	/**
+	 * Performs the depth save pass.
+	 *
+	 * @param {WebGLRenderer} renderer - The renderer.
+	 * @param {WebGLRenderTarget} writeBuffer - The write buffer. This buffer is intended as the rendering
+	 * destination for the pass.
+	 * @param {WebGLRenderTarget} readBuffer - The read buffer. The pass can access the result from the
+	 * previous pass from this buffer.
+	 * @param {number} deltaTime - The delta time in seconds.
+	 * @param {boolean} maskActive - Whether masking is active or not.
+	 */
+
+	render( renderer, writeBuffer, readBuffer, /* deltaTime , maskActive */ ) {
+
+		this.uniforms[ 'tDiffuse' ].value = readBuffer.depthTexture;
+
+		renderer.setRenderTarget( this.renderTarget );
+		if ( this.clear ) renderer.clear();
+		this._fsQuad.render( renderer );
+
+	}
+
+}
+
+export { DepthSavePass };

examples/jsm/utils/DepthTextureUtils.js

@@ -0,0 +1,101 @@
+import { Vector2, Vector3 } from 'three';
+
+const NDC = new Vector3();
+const cursorNDC = new Vector2();
+const z_Axis = new Vector3( 0, 0, - 1 );
+/**
+ * Converts depth to view position.
+ * @param {Vector2} cursorNDC - Normalized device coordinates of the cursor.
+ * @param {number} depth - Depth value.
+ * @param {Camera} camera - Camera object.
+ * @returns {Vector3} - View position.
+ */
+function depthToViewPosition( cursorNDC, depth, camera ) {
+
+	// depth to Z NDC
+	const zNDC = 2.0 * depth - 1.0;
+
+	NDC.set( cursorNDC.x, cursorNDC.y, zNDC );
+
+	const viewPosition = NDC.applyMatrix4( camera.projectionMatrixInverse );
+
+	return viewPosition;
+
+}
+
+/**
+ * Converts logarithmic depth to view position.
+ * @param {Vector2} cursorNDC - Normalized device coordinates of the cursor.
+ * @param {number} logDepth - Logarithmic depth value.
+ * @param {Camera} camera - Camera object.
+ * @returns {Vector3} - View position.
+ */
+function logDepthToViewPosition( cursorNDC, logDepth, camera ) {
+
+	const w = ( camera.far + 1.0 ) ** logDepth - 1;
+
+	NDC.set( cursorNDC.x, cursorNDC.y, - 1 );
+	const viewPosition = NDC.applyMatrix4( camera.projectionMatrixInverse );
+
+	const angle = viewPosition.angleTo( z_Axis );
+	viewPosition.setLength( w / Math.cos( angle ) );
+
+	return viewPosition;
+
+}
+
+const buffer = new Float32Array( 4 );
+
+/**
+ * Computes the 3D world position corresponding to a given 2D mouse position on the screen,
+ * using depth information from a render target.
+ *
+ * @param {Object} mouse - The mouse position in screen coordinates.
+ * @param {number} mouse.x - The x-coordinate of the mouse on the screen.
+ * @param {number} mouse.y - The y-coordinate of the mouse on the screen.
+ * @param {THREE.WebGLRenderer} renderer - The WebGL renderer used to access the render target and canvas size.
+ * @param {THREE.WebGLRenderTarget} renderTarget - The render target that contains depth data.
+ * @param {THREE.Camera} camera - The camera used for the scene. Supports both perspective and orthographic cameras.
+ * @param {THREE.Vector3} target - A pre-allocated Vector3 to store the resulting world position.
+ * @returns {THREE.Vector3} The computed world position corresponding to the mouse input.
+ *
+ * This function:
+ * 1. Converts the mouse position from screen space to canvas space.
+ * 2. Reads the depth value from the render target at that position.
+ * 3. Converts the position from normalized device coordinates (NDC) and depth into view space.
+ * 4. Transforms the result into world space using the camera's world matrix.
+ *
+ * It supports both regular and logarithmic depth buffers.
+ */
+
+const pickWorldPosition = ( mouse, renderer, renderTarget, camera, target ) => {
+
+	const canvasRect = renderer.domElement.getBoundingClientRect();
+
+	const left = mouse.x - canvasRect.left;
+	const bottom = canvasRect.bottom - mouse.y;
+
+	renderer.readRenderTargetPixels( renderTarget, left, bottom, 1, 1, buffer );
+
+	const depth = buffer[ 0 ];
+
+	cursorNDC.setX( ( left / canvasRect.width ) * 2 - 1 );
+	cursorNDC.setY( ( bottom / canvasRect.height ) * 2 - 1 );
+
+	if ( renderer.capabilities.logarithmicDepthBuffer && camera.isPerspectiveCamera ) {
+
+		target.copy( logDepthToViewPosition( cursorNDC, depth, camera ) );
+
+	} else {
+
+		target.copy( depthToViewPosition( cursorNDC, depth, camera ) );
+
+	}
+
+	target.applyMatrix4( camera.matrixWorld );
+
+	return target;
+
+};
+
+export { pickWorldPosition };