commit 4e26d6674dd9cc5e8d6b3f7b0b5553c76603b374
author Luc Trudeau <luc@trud.ca> Mon Sep 11 13:08:40 2017 -0400
committer Luc Trudeau <luc@trud.ca> Thu Sep 28 11:09:50 2017 +0000
tree 6e8b86e2d4dd294a74811085d53c095799c63791
parent 5a010aac24c75a9fabc5f212a9ac71877d67cfc0

[CFL] Move Subsampling in Store

Pixels are subsampled when they are stored in the CfL prediction
buffer. This allows to avoid having two buffers. No impact on the
bitstream.

Results on subset1 (Compared to parent)

  PSNR | PSNR Cb | PSNR Cr | PSNR HVS |   SSIM | MS SSIM | CIEDE 2000
0.0000 |  0.0000 |  0.0000 |   0.0000 | 0.0000 |  0.0000 |     0.0000

https://arewecompressedyet.com/?job=cfl-avg-on-fly%402017-09-23T21%3A38%3A04.440Z&job=cfl-Sub-On-Store%402017-09-24T15%3A01%3A41.161Z

Change-Id: If051e34d6d7078d77609542305a2afebef5cb177

av1/encoder/rdopt.c

diff --git a/av1/encoder/rdopt.c b/av1/encoder/rdopt.c
index b97797f..b1ea0e5 100644
--- a/av1/encoder/rdopt.c
+++ b/av1/encoder/rdopt.c
@@ -5845,7 +5845,7 @@
 #endif  // CONFIG_EXT_INTRA
 
 #if CONFIG_CFL
-static int64_t cfl_alpha_dist(const int *y_pix_q3, const uint8_t *src,
+static int64_t cfl_alpha_dist(const int16_t *pred_buf_q3, const uint8_t *src,
                               int src_stride, int width, int height,
                               int dc_pred, int alpha_q3,
                               int64_t *dist_neg_out) {
@@ -5870,7 +5870,7 @@
   for (int j = 0; j < height; j++) {
     for (int i = 0; i < width; i++) {
       const int uv = src[i];
-      const int scaled_luma = get_scaled_luma_q0(alpha_q3, y_pix_q3[i]);
+      const int scaled_luma = get_scaled_luma_q0(alpha_q3, pred_buf_q3[i]);
 
       // TODO(ltrudeau) add support for HBD.
       diff = uv - clamp(scaled_luma + dc_pred, 0, 255);
@@ -5880,7 +5880,7 @@
       diff = uv - clamp(-scaled_luma + dc_pred, 0, 255);
       dist_neg += diff * diff;
     }
-    y_pix_q3 += MAX_SB_SIZE;
+    pred_buf_q3 += MAX_SB_SIZE;
     src += src_stride;
   }
 
@@ -5906,23 +5906,23 @@
   const int height = cfl->uv_height;
   const int dc_pred_u = cfl->dc_pred[CFL_PRED_U];
   const int dc_pred_v = cfl->dc_pred[CFL_PRED_V];
-  const int *y_pix_q3 = cfl->y_down_pix_q3;
+  const int16_t *pred_buf_q3 = cfl->pred_buf_q3;
 
   int64_t sse[CFL_PRED_PLANES][CFL_MAGS_SIZE];
-  sse[CFL_PRED_U][0] = cfl_alpha_dist(y_pix_q3, src_u, src_stride_u, width,
+  sse[CFL_PRED_U][0] = cfl_alpha_dist(pred_buf_q3, src_u, src_stride_u, width,
                                       height, dc_pred_u, 0, NULL);
-  sse[CFL_PRED_V][0] = cfl_alpha_dist(y_pix_q3, src_v, src_stride_v, width,
+  sse[CFL_PRED_V][0] = cfl_alpha_dist(pred_buf_q3, src_v, src_stride_v, width,
                                       height, dc_pred_v, 0, NULL);
 
   for (int c = 0; c < CFL_ALPHABET_SIZE; c++) {
     const int m = c * 2 + 1;
     const int abs_alpha_q3 = c + 1;
     sse[CFL_PRED_U][m] =
-        cfl_alpha_dist(y_pix_q3, src_u, src_stride_u, width, height, dc_pred_u,
-                       abs_alpha_q3, &sse[CFL_PRED_U][m + 1]);
+        cfl_alpha_dist(pred_buf_q3, src_u, src_stride_u, width, height,
+                       dc_pred_u, abs_alpha_q3, &sse[CFL_PRED_U][m + 1]);
     sse[CFL_PRED_V][m] =
-        cfl_alpha_dist(y_pix_q3, src_v, src_stride_v, width, height, dc_pred_v,
-                       abs_alpha_q3, &sse[CFL_PRED_V][m + 1]);
+        cfl_alpha_dist(pred_buf_q3, src_v, src_stride_v, width, height,
+                       dc_pred_v, abs_alpha_q3, &sse[CFL_PRED_V][m + 1]);
   }
 
   int64_t dist;