From ca5044f09cc031e9f44d50683d44d3f29bec6d73 Mon Sep 17 00:00:00 2001
From: k4yt3x <i@k4yt3x.com>
Date: Thu, 12 Dec 2024 00:00:00 +0000
Subject: [PATCH] feat(libvideo2x): add optimization for scene detection

Signed-off-by: k4yt3x <i@k4yt3x.com>
---
 include/libvideo2x/encoder.h   | 2 +-
 src/encoder.cpp                | 5 +++--
 src/libvideo2x.cpp             | 6 ++----
 tools/video2x/src/argparse.cpp | 4 ++--
 4 files changed, 8 insertions(+), 9 deletions(-)

diff --git a/include/libvideo2x/encoder.h b/include/libvideo2x/encoder.h
index d85a4a0..7a83f10 100644
--- a/include/libvideo2x/encoder.h
+++ b/include/libvideo2x/encoder.h
@@ -19,7 +19,6 @@ struct EncoderConfig {
     bool copy_streams = true;
 
     // Basic video options
-    int frm_rate_mul = 0;
     AVPixelFormat pix_fmt = AV_PIX_FMT_NONE;
 
     // Rate control and compression
@@ -59,6 +58,7 @@ class Encoder {
         EncoderConfig &enc_cfg,
         int width,
         int height,
+        int frm_rate_mul,
         int in_vstream_idx
     );
 
diff --git a/src/encoder.cpp b/src/encoder.cpp
index 74b4520..a83a055 100644
--- a/src/encoder.cpp
+++ b/src/encoder.cpp
@@ -35,6 +35,7 @@ int Encoder::init(
     EncoderConfig &enc_cfg,
     int width,
     int height,
+    int frm_rate_mul,
     int in_vstream_idx
 ) {
     int ret;
@@ -123,9 +124,9 @@ int Encoder::init(
         spdlog::debug("Auto-selected pixel format: {}", av_get_pix_fmt_name(enc_ctx_->pix_fmt));
     }
 
-    if (enc_cfg.frm_rate_mul > 0) {
+    if (frm_rate_mul > 0) {
         AVRational in_frame_rate = get_video_frame_rate(ifmt_ctx, in_vstream_idx);
-        enc_ctx_->framerate = {in_frame_rate.num * enc_cfg.frm_rate_mul, in_frame_rate.den};
+        enc_ctx_->framerate = {in_frame_rate.num * frm_rate_mul, in_frame_rate.den};
         enc_ctx_->time_base = av_inv_q(enc_ctx_->framerate);
     } else {
         // Set the output video's time base
diff --git a/src/libvideo2x.cpp b/src/libvideo2x.cpp
index 67dbec0..07f1acf 100644
--- a/src/libvideo2x.cpp
+++ b/src/libvideo2x.cpp
@@ -93,9 +93,6 @@ int VideoProcessor::process(
         return handle_error(-1, "Failed to determine the output dimensions");
     }
 
-    // Update encoder frame rate multiplier
-    enc_cfg_.frm_rate_mul = proc_cfg_.frm_rate_mul;
-
     // Initialize the encoder
     Encoder encoder;
     ret = encoder.init(
@@ -106,6 +103,7 @@ int VideoProcessor::process(
         enc_cfg_,
         output_width,
         output_height,
+        proc_cfg_.frm_rate_mul,
         in_vstream_idx
     );
     if (ret < 0) {
@@ -389,7 +387,7 @@ int VideoProcessor::process_interpolation(
 
     // Check if a scene change is detected
     bool skip_frame = false;
-    if (prev_frame.get() != nullptr) {
+    if (proc_cfg_.scn_det_thresh < 100.0 && prev_frame.get() != nullptr) {
         float frame_diff = get_frame_diff(prev_frame.get(), frame);
         if (frame_diff > proc_cfg_.scn_det_thresh) {
             spdlog::debug(
diff --git a/tools/video2x/src/argparse.cpp b/tools/video2x/src/argparse.cpp
index 1426a30..ff9b532 100644
--- a/tools/video2x/src/argparse.cpp
+++ b/tools/video2x/src/argparse.cpp
@@ -137,9 +137,9 @@ int parse_args(
             ("frame-rate-mul,m", po::value<int>(&proc_cfg.frm_rate_mul)
                 ->notifier([](int v) { validate_min(v, "frame-rate-mul", 2); }),
                 "Frame rate multiplier")
-            ("scene-thresh,t", po::value<float>(&proc_cfg.scn_det_thresh)->default_value(20.0f)
+            ("scene-thresh,t", po::value<float>(&proc_cfg.scn_det_thresh)->default_value(100.0f)
                 ->notifier([](float v) { validate_range<float>(v, "scene-thresh", 0.0, 100.0); }),
-                "Scene detection threshold")
+                "Scene detection threshold (20 means 20% diff between frames is a scene change)")
         ;
 
         po::options_description libplacebo_opts("libplacebo options");