x/ffmpeg
1
0
Fork 0
mirror of https://git.ffmpeg.org/ffmpeg.git synced 2026-06-11 08:13:06 +00:00
Code Issues Packages Projects Releases Wiki Activity

avfilter/vf_unsharp: fix amount scaling in the high-bit-depth path

Browse Source 
The 16-bit kernel is dispatched for every non-8-bit pixel format
(9/10/12/16-bit content, all stored in uint16_t). It's supposed to
undo the Q16 scaling that set_filter_param() applies to `amount`:

    fp->amount = amount * 65536.0;

but the shift written in the kernel is `>> (8+nbits)`, which for the
nbits=16 instantiation of the macro comes out to `>> 24` instead of
`>> 16`. Because of this, on any non-8-bit input, unsharp applies ~1/256
of the user's requested strength and is effectively a no-op. The
8-bit kernel (nbits=8) happens to be correct because 8+8 == 16.

This commit also widens the intermediate product to int64 before the
shift, to avoid a potential overflow. Take a 16-bit pixel at the
edge of a sharp white/black region, with the user-facing `amount`
set to its declared maximum of 5.0.

    *srx       = 65535
    blur       = 32768
    diff       = *srx - blur                  = 32767
    amount_q16 = 5.0 * 65536                  = 327680

Then the kernel computes:

    product    = diff * amount_q16
               = 32767 * 327680               = 10,737,090,560     (~1.07e10)

which overflows INT32_MAX. Widening to int64 keeps the
multiplication in range; the subsequent `>> 16` brings it back to
sample range and the final cast to int32 is then safe. The widening
is a semantic no-op for 8/9/10/12-bit content where the product
always fits in int32 (worst case at 12-bit: 4095 * 327680 ~ 1.34e9).

Introduced by ee792ebe08 (2019-11-08, "avfilter/vf_unsharp: add 10bit
support"). The fate-filter-unsharp-yuv420p10 reference added in the
same series was generated from the broken kernel and is regenerated
here. fate-filter-unsharp (8-bit) is unaffected.

Repro:

    python3 -c "import numpy as np; y=np.tile(np.where(np.arange(128)//8 & 1, 512, 256).astype('<u2'), (128,1)); c=np.full((64,64), 512, '<u2'); open('in.yuv','wb').write(y.tobytes()+c.tobytes()*2)"

    ffmpeg -f rawvideo -pix_fmt yuv420p10le -s 128x128 -i in.yuv \
        -lavfi "split=2[a][b];[b]unsharp=la=1[bs];[a][bs]psnr" \
        -f null - 2>&1 | grep PSNR

Before: `PSNR y:66.50 ...` -- the filter is effectively a no-op,
        so the sharpened output matches the input almost exactly.
After:  `PSNR y:28.27 ...` -- the filter actually sharpens, so
        output and input differ as expected.

Signed-off-by: Nil Fons Miret <nilf@netflix.com>
Made-with: Cursor
This commit is contained in:
Nil Fons Miret
2026-04-30 21:15:58 +00:00
committed by Kyle Swanson
co-authored by Kyle Swanson
parent 68ea660d83
commit e294b390a0
2 changed files with 23 additions and 23 deletions
Download Patch File Download Diff File Expand all files Collapse all files
libavfilter/vf_unsharp.c
+3 -3
View File
@@ -157,9 +157,9 @@ static int name##_##nbits(AVFilterContext *ctx, void *arg, int jobnr, int nb_job
const uint##nbits##_t *srx = src - steps_y * src_stride + x - steps_x; \
uint##nbits##_t *dsx = dst - steps_y * dst_stride + x - steps_x; \
\
res = (int32_t)*srx + ((((int32_t) * srx - \
(int32_t)((tmp1 + halfscale) >> scalebits)) * amount) >> (8+nbits)); \
*dsx = av_clip_uint##nbits(res); \
res = (int32_t)*srx + (int32_t)(((int64_t)((int32_t)*srx - \
(int32_t)((tmp1 + halfscale) >> scalebits)) * amount) >> 16); \
*dsx = av_clip_uintp2(res, s->bitdepth); \
} \
} \
if (y >= 0) { \
tests/ref/fate/filter-unsharp-yuv420p10
+20 -20
View File
@@ -3,23 +3,23 @@
#codec_id 0: rawvideo
#dimensions 0: 320x240
#sar 0: 1/1
0, 0, 0, 1, 230400, 0x5166b2b0
0, 1, 1, 1, 230400, 0x1fcc8e0e
0, 2, 2, 1, 230400, 0xce04db47
0, 3, 3, 1, 230400, 0xc25661af
0, 4, 4, 1, 230400, 0x727f06aa
0, 5, 5, 1, 230400, 0x2e0f9cc5
0, 6, 6, 1, 230400, 0x407fb93b
0, 7, 7, 1, 230400, 0xc19c2087
0, 8, 8, 1, 230400, 0xacacb223
0, 9, 9, 1, 230400, 0x1277e355
0, 10, 10, 1, 230400, 0xcd36c65d
0, 11, 11, 1, 230400, 0x6c530182
0, 12, 12, 1, 230400, 0x803b2da3
0, 13, 13, 1, 230400, 0x82638dc2
0, 14, 14, 1, 230400, 0x2064904b
0, 15, 15, 1, 230400, 0xce3a7092
0, 16, 16, 1, 230400, 0x374abb39
0, 17, 17, 1, 230400, 0xf8b37d16
0, 18, 18, 1, 230400, 0xb3d1c642
0, 19, 19, 1, 230400, 0xc9b392d1
0, 0, 0, 1, 230400, 0xccb5ffac
0, 1, 1, 1, 230400, 0x6aed4ef7
0, 2, 2, 1, 230400, 0xc7bc3502
0, 3, 3, 1, 230400, 0x2a7b9fa5
0, 4, 4, 1, 230400, 0x29cf1fe7
0, 5, 5, 1, 230400, 0x1b457849
0, 6, 6, 1, 230400, 0x42fd4a9e
0, 7, 7, 1, 230400, 0xccec0881
0, 8, 8, 1, 230400, 0x9b6cca80
0, 9, 9, 1, 230400, 0xee3d5ff1
0, 10, 10, 1, 230400, 0x794d9319
0, 11, 11, 1, 230400, 0x5196b793
0, 12, 12, 1, 230400, 0x239d7933
0, 13, 13, 1, 230400, 0x02ea92ee
0, 14, 14, 1, 230400, 0x111b2c9b
0, 15, 15, 1, 230400, 0x5fbf8810
0, 16, 16, 1, 230400, 0x96afa114
0, 17, 17, 1, 230400, 0x3b587382
0, 18, 18, 1, 230400, 0xb6aa9d42
0, 19, 19, 1, 230400, 0xd463c17e