Comparison file generation for mclt16x16_bayer

src/main/java/ImageDtt.java

@@ -22,6 +22,7 @@
  **
  */
 import java.util.concurrent.atomic.AtomicInteger;
+
 import Jama.Matrix;
 import ij.ImageStack;
 
@@ -193,6 +194,7 @@ public class ImageDtt {
 
 		for (int ithread = 0; ithread < threads.length; ithread++) {
 			threads[ithread] = new Thread() {
+				@Override
 				public void run() {
 					DttRad2 dtt = new DttRad2(dct_size);
 					dtt.set_window(window_type);
@@ -426,6 +428,7 @@ public class ImageDtt {
 
 		for (int ithread = 0; ithread < threads.length; ithread++) {
 			threads[ithread] = new Thread() {
+				@Override
 				public void run() {
 					int tileY,tileX;
 					for (int nTile = ai.getAndIncrement(); nTile < nTiles; nTile = ai.getAndIncrement()) {
@@ -505,6 +508,7 @@ public class ImageDtt {
 
 		for (int ithread = 0; ithread < threads.length; ithread++) {
 			threads[ithread] = new Thread() {
+				@Override
 				public void run() {
 					int tileY,tileX;
 					for (int nTile = ai.getAndIncrement(); nTile < nTiles; nTile = ai.getAndIncrement()) {
@@ -619,6 +623,7 @@ public class ImageDtt {
 
 		for (int ithread = 0; ithread < threads.length; ithread++) {
 			threads[ithread] = new Thread() {
+				@Override
 				public void run() {
 					int tileY,tileX;
 					for (int nTile = ai.getAndIncrement(); nTile < nTiles; nTile = ai.getAndIncrement()) {
@@ -688,6 +693,7 @@ public class ImageDtt {
 			ai.set(0);
 			for (int ithread = 0; ithread < threads.length; ithread++) {
 				threads[ithread] = new Thread() {
+					@Override
 					public void run() {
 						DttRad2 dtt = new DttRad2(dct_size);
 						dtt.set_window(window_type);
@@ -750,6 +756,7 @@ public class ImageDtt {
 		}
 		for (int ithread = 0; ithread < threads.length; ithread++) {
 			threads[ithread] = new Thread() {
+				@Override
 				public void run() {
 					DttRad2 dtt = new DttRad2(transform_size);
 					dtt.set_window(window_type);
@@ -860,6 +867,7 @@ public class ImageDtt {
 		}
 		for (int ithread = 0; ithread < threads.length; ithread++) {
 			threads[ithread] = new Thread() {
+				@Override
 				public void run() {
 					DttRad2 dtt = new DttRad2(transform_size);
 					dtt.set_window(window_type);
@@ -971,9 +979,344 @@ public class ImageDtt {
 		return clt_data;
 	}
 
-/*
- * 	
- */
+
+	public void generateFPGACompareData(
+			final double [][]         image_data, // for selected subcamera
+			final double [][]         colorCentersXY, // pixel centers per color (2 - green)
+			final int                 transform_size,
+			final int                 width,
+			DttRad2                   dtt
+			){
+		int height = image_data[0].length/width;
+		double [][][] fpga_clt_data_in = new double [3][4][];
+		double [][][] fpga_clt_data_out = new double [3][4][];
+		double [][][] fpga_clt_data_rot = new double [3][4][];
+//		double [][] fpga_fract_shiftsXY = new double[3][];
+		double [][] fpga_centersXY = new double [3][2];
+//		int    [][] color_int_shifts = new int [3][2];
+
+		double [][][][] fold_coeff = new double[3][][][];
+		int [] ctile_left = new int [3];
+		int [] ctile_top =  new int [3];
+		double [][] residual_shift = new double[3][2];
+		int    []   ishx = new int[3];
+		int    []   ishy = new int[3];
+
+		double [][] fpga_full_tile = new double [3][FPGA_TILE_SIZE * FPGA_TILE_SIZE];
+		double [][] fpga_tile =      new double [3][4*transform_size*transform_size];
+		for (int chn = 0; chn<3; chn++) for (int j = 0; j < 2; j++) {
+			fpga_centersXY[chn][j] = colorCentersXY[chn][j];
+			// Round to FPGA precision
+			fpga_centersXY[chn][j] = Math.round(128*fpga_centersXY[chn][j])/128.0;
+		}
+		for (int chn = 0; chn<3; chn++) {
+			double px = fpga_centersXY[chn][0] - transform_size;
+			double py = fpga_centersXY[chn][1] - transform_size;
+			// Was wrong rounding, fractional part gets to +0.5
+			ctile_left[chn] = (int) -Math.round(-px);
+			ctile_top[chn] =  (int) -Math.round(-py);
+			residual_shift[chn][0] = -(px - ctile_left[chn]);
+			residual_shift[chn][1] = -(py - ctile_top[chn]);
+		}
+		int lt = (FPGA_TILE_SIZE - 2 * transform_size)/2;
+		for (int chn = 0; chn < 3; chn++){
+			for (int i = 0; i < FPGA_TILE_SIZE; i++){
+				System.arraycopy(
+						image_data[chn],
+						((ctile_top[GREEN_CHN] - lt) + i) * width + (ctile_left[GREEN_CHN] - lt),
+						fpga_full_tile[chn], FPGA_TILE_SIZE * i,
+						FPGA_TILE_SIZE);
+			}
+		}
+		for (int chn = 0; chn < 3; chn++){
+
+	 		if ((ctile_left[chn] >= 0) && (ctile_left[chn] < (width - transform_size * 2)) &&
+					(ctile_top[chn] >= 0) && (ctile_top[chn] < (height - transform_size * 2))) {
+				for (int i = 0; i < transform_size * 2; i++){
+					System.arraycopy(image_data[chn], (ctile_top[chn] + i) * width + ctile_left[chn], fpga_tile[chn], transform_size * 2 * i, transform_size* 2);
+				}
+			} else { // copy by 1
+				for (int i = 0; i < transform_size* 2; i++){
+					int pi = ctile_top[chn] + i;
+					if      (pi < 0)       pi &= 1;
+					else if (pi >= height) pi = height - 2 + (pi & 1);
+					for (int j = 0; j < transform_size* 2; j++){
+						int pj = ctile_left[chn] + j;
+						if      (pj < 0)      pj &= 1;
+						else if (pj >= width) pj = width - 2 + (pj & 1);
+						fpga_tile[chn][transform_size * 2 * i + j] = image_data[chn][pi * width + pj];
+					}
+				}
+			}
+		}
+		// Fold and transform
+		for (int chn = 0; chn < 3; chn++){
+			fold_coeff[chn] = dtt.get_shifted_fold_2d ( // get_shifted_fold_2d(
+					transform_size,
+					residual_shift[chn][0],
+					residual_shift[chn][1],
+					0); // debug level
+		}
+		for (int chn = 0; chn < 3; chn++){
+			for (int dct_mode = 0; dct_mode <4; dct_mode++) {
+//				fpga_clt_data_in[chn][dct_mode] = dtt.fold_tile_debug (fpga_tile[chn], transform_size, dct_mode, fold_coeff[chn]); // DCCT, DSCT, DCST, DSST
+				fpga_clt_data_in[chn][dct_mode] = dtt.fold_tile (fpga_tile[chn], transform_size, dct_mode, fold_coeff[chn]); // DCCT, DSCT, DCST, DSST
+			}
+		}
+
+		for (int chn = 0; chn < 3; chn++) if (chn != GREEN_CHN) {
+			for (int dct_mode = 0; dct_mode <4; dct_mode++) {
+				for (int i = 0; i < 64; i++){
+					fpga_clt_data_in[chn][dct_mode][i] *= 2.0; //adding twice each number in FPGA for R and B
+				}
+			}
+		}
+
+
+		for (int chn = 0; chn < 3; chn++) {
+			for (int dct_mode = 0; dct_mode <4; dct_mode++) {
+				fpga_clt_data_out[chn][dct_mode] = fpga_clt_data_in[chn][dct_mode].clone();
+			}
+		}
+
+		double scale1 = (1 << (FPGA_DTT_IN - 9)); //  -1;
+		scale1 *= 1.0 *((1 << FPGA_WND_BITS) -1) / (1 << FPGA_WND_BITS);
+		scale1 *= 1.0 *((1 << FPGA_WND_BITS) -1) / (1 << FPGA_WND_BITS);
+//		scale1 *= 2.0;
+		System.out.println("scale1="+scale1);
+		for (int chn = 0; chn < 3; chn++) {
+			for (int dct_mode = 0; dct_mode <4; dct_mode++) {
+				System.out.println("Color="+chn+" fpga_clt_data_out[chn][dct_mode] = dtt.dttt_iv(..., scale1="+scale1);
+				fpga_clt_data_out[chn][dct_mode] = dtt.dttt_iv   (fpga_clt_data_out[chn][dct_mode], dct_mode, transform_size, scale1, ((1 << 25) -1)); // debug level
+//				fpga_clt_data_out[chn][dct_mode] = dtt.dttt_iv   (fpga_clt_data_out[chn][dct_mode], dct_mode, transform_size);
+			}
+		}
+
+		for (int chn = 0; chn < 3; chn++) {
+			for (int dct_mode = 0; dct_mode <4; dct_mode++) {
+				fpga_clt_data_rot[chn][dct_mode] = fpga_clt_data_out[chn][dct_mode].clone();
+			}
+		}
+
+// Rotate for fractional shift:
+
+		for (int chn = 0; chn < 3; chn++) {
+			fract_shift(    // fractional shift in transform domain. Currently uses sin/cos - change to tables with 2? rotations
+					fpga_clt_data_rot[chn], // double  [][]  clt_tile,
+					transform_size,
+					residual_shift[chn][0],            // double        shiftX,
+					residual_shift[chn][1],            // double        shiftY,
+					true); // debug
+		}
+
+//		int byr_shift = ((ctile_top[GREEN_CHN] & 1) <<1) | (ctile_left[GREEN_CHN] & 1);
+//GREEN_CHN
+
+		// Printout
+		System.out.println("// Debugging FPGA implementation");
+		for (int chn = 0; chn<3; chn++) {
+			System.out.println("// residual_shift["+chn+"][0]="+residual_shift[chn][0]+", residual_shift["+chn+"][1]="+residual_shift[chn][1]);
+			ishx[chn] = (int) Math.round((1 << (FPGA_SHIFT_BITS)) * residual_shift[chn][0]);
+			ishy[chn] = (int) Math.round((1 << (FPGA_SHIFT_BITS)) * residual_shift[chn][1]);
+			if (ishx[chn] >= (1 << (FPGA_SHIFT_BITS-1))) ishx[chn] = (1 << (FPGA_SHIFT_BITS-1)) - 1;
+			if (ishy[chn] >= (1 << (FPGA_SHIFT_BITS-1))) ishy[chn] = (1 << (FPGA_SHIFT_BITS-1)) - 1;
+			if (ishx[chn] < -(1 << (FPGA_SHIFT_BITS-1))) ishx[chn] = -(1 << (FPGA_SHIFT_BITS-1));
+			if (ishy[chn] < -(1 << (FPGA_SHIFT_BITS-1))) ishy[chn] = -(1 << (FPGA_SHIFT_BITS-1));
+			residual_shift[chn][0] = ishx[chn] * (1.0/(1 << (FPGA_SHIFT_BITS)));
+			residual_shift[chn][1] = ishy[chn] * (1.0/(1 << (FPGA_SHIFT_BITS)));
+			System.out.println(String.format("%4x // color %d shift_x, %d bits", ishx[chn] & ((1 << (FPGA_SHIFT_BITS)) - 1),chn,FPGA_SHIFT_BITS));
+			System.out.println(String.format("%4x // color %d shift_y, %d bits", ishy[chn] & ((1 << (FPGA_SHIFT_BITS)) - 1),chn,FPGA_SHIFT_BITS));
+			System.out.println(String.format("%4x // color %d ctile_left", ctile_left[chn],chn));
+			System.out.println(String.format("%4x // color %d ctile_top",  ctile_top[chn], chn));
+		}
+
+
+
+
+
+
+		System.out.println("\n// Full Bayer fpga tile data");
+		int id = (1 << (FPGA_PIXEL_BITS - 9)); // 8
+		for (int i = 0; i < FPGA_TILE_SIZE*FPGA_TILE_SIZE; i++) {
+			double d = 0.0;
+			for (int fpga_chn = 0; fpga_chn < 3; fpga_chn++){
+				d +=  fpga_full_tile[fpga_chn][i];
+			}
+			System.out.print(String.format("%4x ",(int) Math.round(id * d)));
+			if (((i+1) %FPGA_TILE_SIZE) == 0) {
+				System.out.println();
+			}
+		}
+		System.out.println();
+
+		for (int chn = 0; chn<3; chn++) {
+			double [] fpga_pix_lim = {0.0,0.0};
+			for (int i = 0; i < 256; i++) if (fpga_tile[chn][i] != 0){
+				if (fpga_tile[chn][i] > fpga_pix_lim[0]) fpga_pix_lim[0] = fpga_tile[chn][i];
+				if (fpga_tile[chn][i] < fpga_pix_lim[1]) fpga_pix_lim[1] = fpga_tile[chn][i];
+			}
+			System.out.println(String.format("\n// Color # %d: Pixels input range: %f ... %f", chn, fpga_pix_lim[1], fpga_pix_lim[0]));
+			System.out.println(String.format("//%x // shift_x, %d bits",ishx[chn] & ((1 << (FPGA_SHIFT_BITS)) - 1),FPGA_SHIFT_BITS));
+			System.out.println(String.format("//%x // shift_y, %d bits",ishy[chn] & ((1 << (FPGA_SHIFT_BITS)) - 1),FPGA_SHIFT_BITS));
+			for (int row = 0; row <16; row++){
+				for (int col = 0; col <16; col++){
+					System.out.print(String.format("%4x ",(int) Math.round(id * fpga_tile[chn][row*16 + col])));
+				}
+				System.out.println();
+			}
+			System.out.println();
+		}
+		System.out.println();
+		for (int chn = 0; chn<3; chn++) {
+			System.out.println("// Color="+chn+", signs table (per mode, per index - bitstring of variants, 0 - positive, 1 - negative)");
+			for (int dct_mode = 0; dct_mode <4; dct_mode++) {
+				for (int i = 0; i < 64; i++){
+					int d = 0;
+					for (int b = 0; b < 4; b++){
+						if (fold_coeff[chn][dct_mode][i][b] < 0){
+							d |= (1 << b);
+						}
+					}
+					System.out.print(String.format("%x ",d));
+					if ((i % 16) == 15){
+						System.out.println();
+					}
+				}
+			}
+			System.out.println();
+		}
+		System.out.println();
+		for (int chn = 0; chn<3; chn++) {
+			System.out.println("// Color = "+chn+", absolute values, mode0 (CC), others are the same");
+			//		for (int dct_mode = 0; dct_mode <4; dct_mode++) {
+			int dct_mode = 0;
+			for (int i = 0; i < 64; i++){
+				for (int b = 0; b < 4; b++){
+					int d = (int) Math.round(((1 << FPGA_WND_BITS) -1)* Math.abs(fold_coeff[chn][dct_mode][i][b]));
+					System.out.print(String.format("%5x ",d & ((1 << (FPGA_WND_BITS)) - 1)));
+				}
+				if ((i % 4) == 3){
+					System.out.println();
+				}
+			}
+			System.out.println();
+			//		}
+		}
+		System.out.println();
+
+
+
+		for (int chn = 0; chn<3; chn++) {
+			double [] fpga_dtt_lim = {0.0,0.0};
+			for (int dct_mode = 0; dct_mode <4; dct_mode++) {
+				for (int i = 0; i < 64; i++){
+					if (fpga_clt_data_in[chn][dct_mode][i] > fpga_dtt_lim[0]) fpga_dtt_lim[0] = fpga_clt_data_in[chn][dct_mode][i];
+					if (fpga_clt_data_in[chn][dct_mode][i] < fpga_dtt_lim[1]) fpga_dtt_lim[1] = fpga_clt_data_in[chn][dct_mode][i];
+				}
+			}
+			System.out.println(String.format("// Color= %d, DTT input range: %f ... %f", chn, fpga_dtt_lim[1], fpga_dtt_lim[0]));
+			double scale = (1 << (FPGA_DTT_IN - 9)); //  -1;
+			scale *= 1.0 *((1 << FPGA_WND_BITS) -1) / (1 << FPGA_WND_BITS);
+			scale *= 1.0 *((1 << FPGA_WND_BITS) -1) / (1 << FPGA_WND_BITS);
+			scale *= 2; // Increased twice in FPGA adding twice each number in FPGA
+			System.out.println("Color="+chn+" fpga_clt_data_out[chn][dct_mode] = dtt.dttt_iv(..., scale="+scale);
+
+//			if (chn != GREEN_CHN) scale *= 2; // adding twice each number in FPGA for R and B - done before
+			for (int dct_mode = 0; dct_mode <4; dct_mode++) {
+				for (int i = 0; i < 64; i++){
+					int idd = (int) Math.round(scale * fpga_clt_data_in[chn][dct_mode][i]);
+					System.out.print(String.format("%7x ", idd & ((1 << 25) -1)));
+					if ((i % 8) == 7) System.out.println();
+				}
+				System.out.println();
+			}
+			System.out.println();
+		}
+		System.out.println();
+
+		for (int chn = 0; chn<3; chn++) {
+			double [] fpga_dtt_lim = {0.0,0.0};
+			for (int dct_mode = 0; dct_mode <4; dct_mode++) {
+				for (int i = 0; i < 64; i++){
+					if (fpga_clt_data_out[chn][dct_mode][i] > fpga_dtt_lim[0]) fpga_dtt_lim[0] = fpga_clt_data_out[chn][dct_mode][i];
+					if (fpga_clt_data_out[chn][dct_mode][i] < fpga_dtt_lim[1]) fpga_dtt_lim[1] = fpga_clt_data_out[chn][dct_mode][i];
+				}
+			}
+			System.out.println(String.format("// Color = %d: DTT output range: %f ... %f", chn, fpga_dtt_lim[1], fpga_dtt_lim[0]));
+			// scale = (1 << (FPGA_DTT_IN - 9)); //  -1;
+//			double scale = (1 << (FPGA_DTT_IN - 8)); //  increased twice
+			double scale = (1 << (FPGA_DTT_IN - 9)); //  Do not increase - lead to overflow !
+			scale *= 1.0 *((1 << FPGA_WND_BITS) -1) / (1 << FPGA_WND_BITS);
+			scale *= 1.0 *((1 << FPGA_WND_BITS) -1) / (1 << FPGA_WND_BITS);
+
+			for (int dct_mode = 0; dct_mode <4; dct_mode++) {
+				for (int i = 0; i < 64; i++){
+					int idd = (int) Math.round(scale * fpga_clt_data_out[chn][dct_mode][i]);
+					System.out.print(String.format("%7x ", idd & ((1 << 25) -1)));
+					if ((i % 8) == 7) System.out.println();
+				}
+				System.out.println();
+			}
+			System.out.println();
+			System.out.println("// Color = "+chn+" Testing symmetry of checkerboard patterns");
+			for (int dct_mode = 0; dct_mode < 2; dct_mode++) {
+				for (int i = 0; i < 64; i++){
+					if ((i % 8) == 0) System.out.print("// ");
+					int idd = (int) Math.round(scale * fpga_clt_data_out[chn][dct_mode][i]);
+					int idd1 = (int) Math.round(scale * fpga_clt_data_out[chn][3-dct_mode][63-i]);
+					System.out.print(String.format("%7x ", (idd-idd1) & ((1 << 25) -1)));
+					if ((i % 8) == 7) System.out.println();
+				}
+				System.out.println();
+			}
+			System.out.println();
+			System.out.println("// Color = "+chn+" Testing antisymmetry of checkerboard patterns");
+			for (int dct_mode = 0; dct_mode < 2; dct_mode++) {
+				for (int i = 0; i < 64; i++){
+					if ((i % 8) == 0) System.out.print("// ");
+					int idd = (int) Math.round(scale * fpga_clt_data_out[chn][dct_mode][i]);
+					int idd1 = (int) Math.round(scale * fpga_clt_data_out[chn][3-dct_mode][63-i]);
+					System.out.print(String.format("%7x ", (idd+idd1) & ((1 << 25) -1)));
+					if ((i % 8) == 7) System.out.println();
+				}
+				System.out.println();
+			}
+			System.out.println();
+		}
+		System.out.println();
+
+		for (int chn = 0; chn<3; chn++) {
+//			double scale = (1 << (FPGA_DTT_IN - 9)); //  -1;
+//			double scale = (1 << (FPGA_DTT_IN - 8)); //
+			double scale = (1 << (FPGA_DTT_IN - 9)); //  Do not increase - lead to overflow !
+
+			// compensate for DTT scale
+			scale *= 1.0 *((1 << FPGA_WND_BITS) -1) / (1 << FPGA_WND_BITS);
+			scale *= 1.0 *((1 << FPGA_WND_BITS) -1) / (1 << FPGA_WND_BITS);
+			// compensate for rotator scale:
+			scale *= 1.0 *((1 << FPGA_WND_BITS) -1) / (1 << FPGA_WND_BITS);
+			scale *= 1.0 *((1 << FPGA_WND_BITS) -1) / (1 << FPGA_WND_BITS);
+			double [] fpga_dtt_lim = {0.0,0.0};
+			for (int dct_mode = 0; dct_mode <4; dct_mode++) {
+				for (int j = 0; j < 64; j++){
+					if (fpga_clt_data_rot[chn][dct_mode][j] > fpga_dtt_lim[0]) fpga_dtt_lim[0] = fpga_clt_data_rot[chn][dct_mode][j];
+					if (fpga_clt_data_rot[chn][dct_mode][j] < fpga_dtt_lim[1]) fpga_dtt_lim[1] = fpga_clt_data_rot[chn][dct_mode][j];
+				}
+			}
+
+			System.out.println(String.format("// Color = %d: DTT rotated, shift_x=%f. shift_y = %f", chn, residual_shift[chn][0],residual_shift[chn][1]));
+			System.out.println(String.format("// DTT rotated  range: %f ... %f", fpga_dtt_lim[1], fpga_dtt_lim[0]));
+//			scale = (1 << (FPGA_DTT_IN - 9)); //  -1;
+			for (int dct_mode = 0; dct_mode <4; dct_mode++) {
+				for (int j = 0; j < 64; j++){
+					int idd = (int) Math.round(scale * fpga_clt_data_rot[chn][dct_mode][j]);
+					System.out.print(String.format("%7x ", idd & ((1 << 25) -1)));
+					if ((j % 8) == 7) System.out.println();
+				}
+				System.out.println();
+			}
+		}
+	}
 
 
 	public double [][][][][][] clt_aberrations_quad_corr(
@@ -1194,6 +1537,7 @@ public class ImageDtt {
 		final Matrix [] corr_rots = geometryCorrection.getCorrVector().getRotMatrices(); // get array of per-sensor rotation matrices
 		for (int ithread = 0; ithread < threads.length; ithread++) {
 			threads[ithread] = new Thread() {
+				@Override
 				public void run() {
 					DttRad2 dtt = new DttRad2(transform_size);
 					dtt.set_window(window_type);
@@ -1301,7 +1645,29 @@ public class ImageDtt {
 								}
 							}
 						} // if (macro_mode) ... else
-						
+						if (FPGA_COMPARE_DATA && (globalDebugLevel > 0) && (tileX == debug_tileX) && (tileY == debug_tileY)) {
+							final int fpga_cam = 0;
+							double [][] manual_offsets={
+//									{ 1.3, -2.7},
+//									{-1.3,  2.7},
+//									{ 0.0,  0.0}};
+							{ 2.3, -2.7},
+							{-0.3,  2.7},
+							{ 0.0,  0.0}};
+
+							double [][] colorCentersXY = {
+									{centersXY[fpga_cam][0] + manual_offsets[0][0], centersXY[fpga_cam][1] + manual_offsets[0][1]}, // add manual offsets here
+									{centersXY[fpga_cam][0] + manual_offsets[1][0], centersXY[fpga_cam][1] + manual_offsets[1][1]},
+									{centersXY[fpga_cam][0] + manual_offsets[2][0], centersXY[fpga_cam][1] + manual_offsets[2][1]}
+							};
+							generateFPGACompareData(
+									image_data[fpga_cam], // final double [][]                  image_data, // for selected subcamera
+									colorCentersXY,       // final double [][]         colorCentersXY, // pixel centers per color (2 - green)
+									transform_size,       // final int                 transform_size,
+									width,                 // final int                 width
+									dtt
+									);
+						}
 
 						for (int chn = 0; chn <numcol; chn++) {
 							boolean debug_for_fpga = FPGA_COMPARE_DATA && (globalDebugLevel > 0) && (tileX == debug_tileX) && (tileY == debug_tileY) && (chn == 2);
@@ -1476,6 +1842,8 @@ public class ImageDtt {
 
 							}
 						}
+
+
 						// calculate overexposed fraction
 						if (saturation_imp != null) {
 							disparity_map[OVEREXPOSED][nTile] = (1.0 * overexp_all[0]) / overexp_all[1];
@@ -2585,6 +2953,7 @@ public class ImageDtt {
 		}
 		for (int ithread = 0; ithread < threads.length; ithread++) {
 			threads[ithread] = new Thread() {
+				@Override
 				public void run() {
 					DttRad2 dtt = new DttRad2(dct_size);
 					dtt.set_window(window_type);
@@ -2684,6 +3053,7 @@ public class ImageDtt {
 			ai.set(0);
 			for (int ithread = 0; ithread < threads.length; ithread++) {
 				threads[ithread] = new Thread() {
+					@Override
 					public void run() {
 						DttRad2 dtt = new DttRad2(dct_size);
 						dtt.set_window(window_type);
@@ -2799,6 +3169,7 @@ public class ImageDtt {
 			ai.set(0);
 			for (int ithread = 0; ithread < threads.length; ithread++) {
 				threads[ithread] = new Thread() {
+					@Override
 					public void run() {
 						int tileY,tileX;
 						int n2 = transform_size * 2;
@@ -2931,6 +3302,7 @@ public class ImageDtt {
 		final AtomicInteger ai = new AtomicInteger(0);
 		for (int ithread = 0; ithread < threads.length; ithread++) {
 			threads[ithread] = new Thread() {
+				@Override
 				public void run() {
 					int tileY,tileX;
 					for (int nTile = ai.getAndIncrement(); nTile < nTiles; nTile = ai.getAndIncrement()) {
@@ -3006,6 +3378,7 @@ public class ImageDtt {
 		final AtomicInteger ai = new AtomicInteger(0);
 		for (int ithread = 0; ithread < threads.length; ithread++) {
 			threads[ithread] = new Thread() {
+				@Override
 				public void run() {
 					int tileY,tileX;
 					for (int nTile = ai.getAndIncrement(); nTile < nTiles; nTile = ai.getAndIncrement()) {
@@ -3125,6 +3498,7 @@ public class ImageDtt {
 
 		for (int ithread = 0; ithread < threads.length; ithread++) {
 			threads[ithread] = new Thread() {
+				@Override
 				public void run() {
 					int tileY,tileX;
 					for (int nTile = ai.getAndIncrement(); nTile < nTiles; nTile = ai.getAndIncrement()) {
@@ -3159,6 +3533,7 @@ public class ImageDtt {
 
 		for (int ithread = 0; ithread < threads.length; ithread++) {
 			threads[ithread] = new Thread() {
+				@Override
 				public void run() {
 					DttRad2 dtt = new DttRad2(dct_size);
 					int tileY,tileX;
@@ -3194,6 +3569,7 @@ public class ImageDtt {
 
 		for (int ithread = 0; ithread < threads.length; ithread++) {
 			threads[ithread] = new Thread() {
+				@Override
 				public void run() {
 					int tileY,tileX;
 					double scale = 0.25;
@@ -3315,6 +3691,7 @@ public class ImageDtt {
 
 		for (int ithread = 0; ithread < threads.length; ithread++) {
 			threads[ithread] = new Thread() {
+				@Override
 				public void run() {
 					int tileY,tileX;
 					for (int nTile = ai.getAndIncrement(); nTile < nTiles; nTile = ai.getAndIncrement()) {
@@ -3372,6 +3749,7 @@ public class ImageDtt {
 
 		for (int ithread = 0; ithread < threads.length; ithread++) {
 			threads[ithread] = new Thread() {
+				@Override
 				public void run() {
 					int tileY,tileX;
 					for (int nTile = ai.getAndIncrement(); nTile < nTiles; nTile = ai.getAndIncrement()) {
@@ -3472,6 +3850,7 @@ public class ImageDtt {
 
 		for (int ithread = 0; ithread < threads.length; ithread++) {
 			threads[ithread] = new Thread() {
+				@Override
 				public void run() {
 					int tileY,tileX;
 					for (int nTile = ai.getAndIncrement(); nTile < nTiles; nTile = ai.getAndIncrement()) {
@@ -3661,6 +4040,7 @@ public class ImageDtt {
 		final AtomicInteger ai = new AtomicInteger(0);
 		for (int ithread = 0; ithread < threads.length; ithread++) {
 			threads[ithread] = new Thread() {
+				@Override
 				public void run() {
 					int tileY,tileX,chn;
 					for (int nTile = ai.getAndIncrement(); nTile < nTiles; nTile = ai.getAndIncrement()) {
@@ -4091,11 +4471,7 @@ public class ImageDtt {
 				}
 			}
 			System.out.println(String.format("// DTT input range: %f ... %f", fpga_dtt_lim[1], fpga_dtt_lim[0]));
-//			double scale = (1 << (FPGA_DTT_IN - 10)) -1;
-///			double scale = (1 << (FPGA_DTT_IN - 8)) -1;
 			double scale = (1 << (FPGA_DTT_IN - 9)); //  -1;
-///			scale /= 1.0 *((1 << FPGA_WND_BITS) -1) / (1 << FPGA_WND_BITS);
-///			scale /= 1.0 *((1 << FPGA_WND_BITS) -1) / (1 << FPGA_WND_BITS);
 			scale *= 1.0 *((1 << FPGA_WND_BITS) -1) / (1 << FPGA_WND_BITS);
 			scale *= 1.0 *((1 << FPGA_WND_BITS) -1) / (1 << FPGA_WND_BITS);
 			for (int dct_mode = 0; dct_mode <4; dct_mode++) {
@@ -4109,6 +4485,7 @@ public class ImageDtt {
 			System.out.println();
 
 			for (int dct_mode = 0; dct_mode <4; dct_mode++) {
+				System.out.println("Color= 2? clt_tile[dct_mode] = dtt.dttt_iv(..., scale="	+scale);
 				clt_tile[dct_mode] = dtt.dttt_iv   (clt_tile[dct_mode], dct_mode, transform_size, scale, ((1 << 25) -1)); // debug level
 			}
 			fpga_dtt_lim[0] = 0.0;
@@ -4390,6 +4767,7 @@ public class ImageDtt {
 
 		for (int ithread = 0; ithread < threads.length; ithread++) {
 			threads[ithread] = new Thread() {
+				@Override
 				public void run() {
 					DttRad2 dtt = new DttRad2(dct_size);
 					dtt.set_window(window_type);
@@ -4457,6 +4835,7 @@ public class ImageDtt {
 
 		for (int ithread = 0; ithread < threads.length; ithread++) {
 			threads[ithread] = new Thread() {
+				@Override
 				public void run() {
 					int tileY,tileX;
 					double [] dct1 = new double [dct_size*dct_size];