shithub: libvpx

Info  •  Files  •  Log  •  Branches

vp9-rtc: Fixes to resizer for real-time

Reduce the time before sampling begins (after key)
and reduce averaging window, to make resize act
faster.

Reset RC paramaters for temporal layers on resize.

Add per-frame-bandwidth thresholds to force
downsize for extreme case, for HD input.

Change-Id: I8e08580b2216a2e6981502552025370703cd206c

--- a/test/resize_test.cc

+++ b/test/resize_test.cc

@@ -672,11 +672,11 @@

     ASSERT_EQ(info->h, GetFrameHeight(idx));

     if (info->w != last_w || info->h != last_h) {

       resize_count++;

-      if (resize_count == 1) {

+      if (resize_count <= 2) {

         // Verify that resize down occurs.

         ASSERT_LT(info->w, last_w);

         ASSERT_LT(info->h, last_h);

-      } else if (resize_count == 2) {

+      } else if (resize_count > 2) {

         // Verify that resize up occurs.

         ASSERT_GT(info->w, last_w);

         ASSERT_GT(info->h, last_h);

@@ -687,8 +687,8 @@

   }

 #if CONFIG_VP9_DECODER

-  // Verify that we get 2 resize events in this test.

-  ASSERT_EQ(resize_count, 2) << "Resizing should occur twice.";

+  // Verify that we get 4 resize events in this test.

+  ASSERT_EQ(resize_count, 4) << "Resizing should occur twice.";

   EXPECT_EQ(static_cast<unsigned int>(0), GetMismatchFrames());

 #else

   printf("Warning: VP9 decoder unavailable, unable to check resize count!\n");

--- a/vp9/encoder/vp9_ratectrl.c

+++ b/vp9/encoder/vp9_ratectrl.c

@@ -2391,6 +2391,11 @@

             cpi->resize_scale_num * lc->scaling_factor_num;

         lc->scaling_factor_den_resize =

             cpi->resize_scale_den * lc->scaling_factor_den;

+        // Reset rate control for all temporal layers.

+        lc->rc.buffer_level = lc->rc.optimal_buffer_level;

+        lc->rc.bits_off_target = lc->rc.optimal_buffer_level;

+        lc->rc.rate_correction_factors[INTER_FRAME] =

+            rc->rate_correction_factors[INTER_FRAME];

       }

       // Set the size for this current temporal layer.

       lc = &svc->layer_context[svc->spatial_layer_id *

@@ -2663,6 +2668,7 @@

   int min_width = (320 * 4) / 3;

   int min_height = (180 * 4) / 3;

   int down_size_on = 1;

+  int force_downsize_rate = 0;

   cpi->resize_scale_num = 1;

   cpi->resize_scale_den = 1;

   // Don't resize on key frame; reset the counters on key frame.

@@ -2683,10 +2689,31 @@

   }

 #endif

+  // Force downsize based on per-frame-bandwidth, for extreme case,

+  // for HD input.

+  if (cpi->resize_state == ORIG && cm->width * cm->height >= 1280 * 720) {

+    if (rc->avg_frame_bandwidth < (int)(300000 / 30)) {

+      resize_action = DOWN_ONEHALF;

+      cpi->resize_state = ONE_HALF;

+      force_downsize_rate = 1;

+    } else if (rc->avg_frame_bandwidth < (int)(400000 / 30)) {

+      resize_action = ONEHALFONLY_RESIZE ? DOWN_ONEHALF : DOWN_THREEFOUR;

+      cpi->resize_state = ONEHALFONLY_RESIZE ? ONE_HALF : THREE_QUARTER;

+      force_downsize_rate = 1;

+    }

+  } else if (cpi->resize_state == THREE_QUARTER &&

+             cm->width * cm->height >= 960 * 540) {

+    if (rc->avg_frame_bandwidth < (int)(300000 / 30)) {

+      resize_action = DOWN_ONEHALF;

+      cpi->resize_state = ONE_HALF;

+      force_downsize_rate = 1;

+    }

+  }

+

   // Resize based on average buffer underflow and QP over some window.

   // Ignore samples close to key frame, since QP is usually high after key.

-  if (cpi->rc.frames_since_key > 2 * cpi->framerate) {

-    const int window = (int)(4 * cpi->framerate);

+  if (!force_downsize_rate && cpi->rc.frames_since_key > cpi->framerate) {

+    const int window = VPXMIN(30, (int)(2 * cpi->framerate));

     cpi->resize_avg_qp += cm->base_qindex;

     if (cpi->rc.buffer_level < (int)(30 * rc->optimal_buffer_level / 100))

       ++cpi->resize_buffer_underflow;