[FFmpeg-devel] [PATCH] WIP: swscale/range_convert: fix truncation bias in range conversion (PR #20836)

[FFmpeg-devel] [PATCH] WIP: swscale/range_convert: fix truncation bias in range conversion (PR #20836)

[Ramiro Polla via ffmpeg-devel]

PR #20836 opened by Ramiro Polla (ramiro)
URL: https://code.ffmpeg.org/FFmpeg/FFmpeg/pulls/20836
Patch URL: https://code.ffmpeg.org/FFmpeg/FFmpeg/pulls/20836.patch

384fe39623e932e68fe35af7d5b51fcd0a6c28fb introduced a regression in the
range conversion offset calculation, resulting in a slight green tint.

The offset being calculated was not taking into consideration a bias
needed for correctly rounding the result from the multiplication stage,
leading to a truncated value.

Fixes issue #11646.

Note this is still missing updating all fate results.


>From dec7df004722525fc0ba3c8b98c7248a90a7e617 Mon Sep 17 00:00:00 2001
From: Ramiro Polla <ramiro.polla@gmail.com>
Date: Tue, 4 Nov 2025 16:18:47 +0100
Subject: [PATCH] [WIP] swscale/range_convert: fix truncation bias in range
 conversion

384fe39623e932e68fe35af7d5b51fcd0a6c28fb introduced a regression in the
range conversion offset calculation, resulting in a slight green tint.

The offset being calculated was not taking into consideration a bias
needed for correctly rounding the result from the multiplication stage,
leading to a truncated value.

Fixes issue #11646.

Note this is still missing updating all fate results.
---
 libswscale/swscale.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/libswscale/swscale.c b/libswscale/swscale.c
index c3d64a67b8..f4c7eccac4 100644
--- a/libswscale/swscale.c
+++ b/libswscale/swscale.c
@@ -581,7 +581,8 @@ static void solve_range_convert(uint16_t src_min, uint16_t src_max,
     int total_shift = mult_shift + src_shift;
     *coeff = AV_CEIL_RSHIFT(((uint64_t) dst_range << total_shift) / src_range, src_shift);
     *offset = ((int64_t) dst_max << total_shift) -
-              ((int64_t) src_max << src_shift) * *coeff;
+              ((int64_t) src_max << src_shift) * *coeff +
+              (1U << (mult_shift - 1));
 }
 
 static void init_range_convert_constants(SwsInternal *c)
-- 
2.49.1

_______________________________________________
ffmpeg-devel mailing list -- ffmpeg-devel@ffmpeg.org
To unsubscribe send an email to ffmpeg-devel-leave@ffmpeg.org