[FFmpeg-devel] [PATCH][WIP][RFC] avfilter: add opencl v360 filter
Signed-off-by: Paul B Mahol
---
For now just equirectangular to flat conversion, but could be
with some effort extended with all formats supported by v360 filter,
minus non-padded stuff, but that is not present in normal usecases.
---
libavfilter/Makefile | 2 +
libavfilter/allfilters.c | 1 +
libavfilter/opencl/v360.cl | 158 +++
libavfilter/opencl_source.h | 1 +
libavfilter/vf_v360_opencl.c | 505 +++
5 files changed, 667 insertions(+)
create mode 100644 libavfilter/opencl/v360.cl
create mode 100644 libavfilter/vf_v360_opencl.c
diff --git a/libavfilter/Makefile b/libavfilter/Makefile
index 358f121cb4..eb5365a739 100644
--- a/libavfilter/Makefile
+++ b/libavfilter/Makefile
@@ -485,6 +485,8 @@ OBJS-$(CONFIG_UNSHARP_OPENCL_FILTER) +=
vf_unsharp_opencl.o opencl.o \
OBJS-$(CONFIG_UNTILE_FILTER) += vf_untile.o
OBJS-$(CONFIG_USPP_FILTER) += vf_uspp.o qp_table.o
OBJS-$(CONFIG_V360_FILTER) += vf_v360.o
+OBJS-$(CONFIG_V360_OPENCL_FILTER)+= vf_v360_opencl.o opencl.o \
+opencl/v360.o
OBJS-$(CONFIG_VAGUEDENOISER_FILTER) += vf_vaguedenoiser.o
OBJS-$(CONFIG_VARBLUR_FILTER)+= vf_varblur.o framesync.o
OBJS-$(CONFIG_VECTORSCOPE_FILTER)+= vf_vectorscope.o
diff --git a/libavfilter/allfilters.c b/libavfilter/allfilters.c
index 409ab5d3c4..04f1925c14 100644
--- a/libavfilter/allfilters.c
+++ b/libavfilter/allfilters.c
@@ -462,6 +462,7 @@ extern const AVFilter ff_vf_unsharp_opencl;
extern const AVFilter ff_vf_untile;
extern const AVFilter ff_vf_uspp;
extern const AVFilter ff_vf_v360;
+extern const AVFilter ff_vf_v360_opencl;
extern const AVFilter ff_vf_vaguedenoiser;
extern const AVFilter ff_vf_varblur;
extern const AVFilter ff_vf_vectorscope;
diff --git a/libavfilter/opencl/v360.cl b/libavfilter/opencl/v360.cl
new file mode 100644
index 00..003c188249
--- /dev/null
+++ b/libavfilter/opencl/v360.cl
@@ -0,0 +1,158 @@
+/*
+ * Copyright (c) 2021 Paul B Mahol
+ *
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE | CLK_FILTER_LINEAR);
+
+static float scale(float x, float s)
+{
+return (0.5 * x + 0.5) * (s - 1.);
+}
+
+static float rescale(int x, float s)
+{
+return (2. * x + 1.) / s - 1.;
+}
+
+__kernel void equirect_to_xyz(__write_only image2d_t dst)
+{
+int2 p = (int2)(get_global_id(0), get_global_id(1));
+int2 size = (int2)(get_global_size(0), get_global_size(1));
+float2 f;
+
+f.x = rescale(p.x, size.x);
+f.y = rescale(p.y, size.y);
+
+float sin_phi = sin(f.x);
+float cos_phi = cos(f.x);
+float sin_theta = sin(f.y);
+float cos_theta = cos(f.y);
+
+float4 vec;
+
+vec.x = cos_theta * sin_phi;
+vec.y = sin_theta;
+vec.z = cos_theta * cos_phi;
+
+write_imagef(dst, p, vec);
+}
+
+__kernel void flat_to_xyz(global float3 *dst,
+ float2 flat_range)
+{
+int2 p = (int2)(get_global_id(0), get_global_id(1));
+int2 size = (int2)(get_global_size(0), get_global_size(1));
+float2 f;
+
+f.x = flat_range.x * rescale(p.x, size.x);
+f.y = flat_range.y * rescale(p.y, size.y);
+
+float3 vec;
+
+vec.x = f.x;
+vec.y = f.y;
+vec.z = 1.0;
+
+vec = normalize(vec);
+
+dst[p.y * size.x + p.x] = vec;
+}
+
+__kernel void xyz_to_equirect(global float2 *dst,
+ float2 iflat_range,
+ global float3 *m,
+ __read_only image2d_t src)
+{
+int2 p = (int2)(get_global_id(0), get_global_id(1));
+int2 size = (int2)(get_global_size(0), get_global_size(1));
+
+float3 vec = m[p.x + size.x * p.y];
+
+const float phi = atan2(vec.x, vec.z) / iflat_range.x;
+const float theta = asin(vec.y) / iflat_range.y;
+
+float2 uv;
+
+uv.x = scale(phi, size.x);
+uv.y = scale(theta, size.y);
+
+dst[p.x + p.y * size.x] = uv;
+}
+
+__kernel void remap(__write_only image2d_t dst,
+__read_only image2d_t src,
+global float2 *remap)
+{
+int2 p = (int2)(get_global_id(0), get_global_id(1)
Re: [FFmpeg-devel] [PATCH 1/8] avfilter/af_replaygain: use fabsf() instead of fabs()
On 10/14/2021 10:08 AM, lance.lmw...@gmail.com wrote:
From: Limin Wang
Signed-off-by: Limin Wang
---
libavfilter/af_replaygain.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/libavfilter/af_replaygain.c b/libavfilter/af_replaygain.c
index 4bf2763..da5c196 100644
--- a/libavfilter/af_replaygain.c
+++ b/libavfilter/af_replaygain.c
@@ -428,7 +428,7 @@ static void butter_filter_stereo_samples(ReplayGainContext
*s,
// (slowing us down).
for (j = -4; j < 0; ++j)
-if (fabs(hist_a[i + j]) > 1e-10 || fabs(hist_b[i + j]) > 1e-10)
+if (fabsf(hist_a[i + j]) > 1e-10 || fabsf(hist_b[i + j]) > 1e-10)
Shouldn't the constants also be made into floats? Otherwise the compiler
will probably do a conversion to double.
break;
if (!j) {
@@ -477,7 +477,7 @@ static void yule_filter_stereo_samples(ReplayGainContext
*s, const float *src,
// (slowing us down).
for (j = -20; j < 0; ++j)
-if (fabs(hist_a[i + j]) > 1e-10 || fabs(hist_b[i + j]) > 1e-10)
+if (fabsf(hist_a[i + j]) > 1e-10 || fabsf(hist_b[i + j]) > 1e-10)
break;
if (!j) {
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
To unsubscribe, visit link above, or email
ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
[FFmpeg-devel] [PATCH 2/2] Revert "arm: hevc_qpel: Fix the assembly to work with non-multiple of 8 widths"
This reverts commit 2589060b92eeeb944c6e2b50e38412c0c5fabcf4.
Signed-off-by: J. Dekker
---
libavcodec/arm/hevcdsp_qpel_neon.S | 18 +-
1 file changed, 9 insertions(+), 9 deletions(-)
diff --git a/libavcodec/arm/hevcdsp_qpel_neon.S
b/libavcodec/arm/hevcdsp_qpel_neon.S
index f71bec05ed..caa6efa766 100644
--- a/libavcodec/arm/hevcdsp_qpel_neon.S
+++ b/libavcodec/arm/hevcdsp_qpel_neon.S
@@ -237,7 +237,7 @@
vld1.8{d23}, [r2], r3
bne 8b
subs r5, #8
-ble 99f
+beq 99f
mov r4, r12
add r6, #16
mov r0, r6
@@ -280,7 +280,7 @@
vld1.8{d23}, [r2], r3
bne 8b
subs r5, #8
-ble 99f
+beq 99f
mov r4, r12
add r6, #8
mov r0, r6
@@ -310,7 +310,7 @@
vld1.8{d23}, [r2], r3
bne 8b
subs r5, #8
-ble 99f
+beq 99f
mov r4, r12
add r6, #8
mov r0, r6
@@ -377,7 +377,7 @@ endfunc
vst1.16 {q7}, [r0], r1
bne 8b
subs r5, #8
-ble 99f
+beq 99f
mov r4, r12
add r6, #16
mov r0, r6
@@ -417,7 +417,7 @@ endfunc
vst1.8d0, [r0], r1
bne 8b
subs r5, #8
-ble 99f
+beq 99f
mov r4, r12
add r6, #8
mov r0, r6
@@ -446,7 +446,7 @@ endfunc
vst1.8 d0, [r0], r1
bne 8b
subs r5, #8
-ble 99f
+beq 99f
mov r4, r12
add r6, #8
add r10, #16
@@ -533,7 +533,7 @@ endfunc
\filterh q7
bne 8b
subs r5, #8
-ble 99f
+beq 99f
mov r4, r12
add r6, #16
mov r0, r6
@@ -594,7 +594,7 @@ endfunc
\filterh q7
bne 8b
subs r5, #8
-ble 99f
+beq 99f
mov r4, r12
add r6, #8
mov r0, r6
@@ -641,7 +641,7 @@ endfunc
\filterh q7
bne 8b
subs r5, #8
-ble 99f
+beq 99f
mov r4, r12
add r6, #8
mov r0, r6
--
2.30.1 (Apple Git-130)
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
To unsubscribe, visit link above, or email
ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
[FFmpeg-devel] [PATCH 1/2] lavc/arm: dont assign hevc_qpel non-multiple of 8 width stubs
Signed-off-by: J. Dekker
---
libavcodec/arm/hevcdsp_init_neon.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/libavcodec/arm/hevcdsp_init_neon.c
b/libavcodec/arm/hevcdsp_init_neon.c
index 201a088dac..112edb5edd 100644
--- a/libavcodec/arm/hevcdsp_init_neon.c
+++ b/libavcodec/arm/hevcdsp_init_neon.c
@@ -270,7 +270,8 @@ av_cold void ff_hevc_dsp_init_neon(HEVCDSPContext *c, const
int bit_depth)
put_hevc_qpel_uw_neon[3][1] = ff_hevc_put_qpel_uw_h1v3_neon_8;
put_hevc_qpel_uw_neon[3][2] = ff_hevc_put_qpel_uw_h2v3_neon_8;
put_hevc_qpel_uw_neon[3][3] = ff_hevc_put_qpel_uw_h3v3_neon_8;
-for (x = 0; x < 10; x++) {
+for (x = 3; x < 10; x++) {
+if (x == 4) continue;
c->put_hevc_qpel[x][1][0] = ff_hevc_put_qpel_neon_wrapper;
c->put_hevc_qpel[x][0][1] = ff_hevc_put_qpel_neon_wrapper;
c->put_hevc_qpel[x][1][1] = ff_hevc_put_qpel_neon_wrapper;
--
2.30.1 (Apple Git-130)
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
To unsubscribe, visit link above, or email
ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
[FFmpeg-devel] [PATCH][WIP] avfilter: add opencl v360 filter
Signed-off-by: Paul B Mahol
---
libavfilter/Makefile | 2 +
libavfilter/allfilters.c | 1 +
libavfilter/opencl/v360.cl | 419
libavfilter/opencl_source.h | 1 +
libavfilter/vf_v360_opencl.c | 719 +++
5 files changed, 1142 insertions(+)
create mode 100644 libavfilter/opencl/v360.cl
create mode 100644 libavfilter/vf_v360_opencl.c
diff --git a/libavfilter/Makefile b/libavfilter/Makefile
index 358f121cb4..eb5365a739 100644
--- a/libavfilter/Makefile
+++ b/libavfilter/Makefile
@@ -485,6 +485,8 @@ OBJS-$(CONFIG_UNSHARP_OPENCL_FILTER) +=
vf_unsharp_opencl.o opencl.o \
OBJS-$(CONFIG_UNTILE_FILTER) += vf_untile.o
OBJS-$(CONFIG_USPP_FILTER) += vf_uspp.o qp_table.o
OBJS-$(CONFIG_V360_FILTER) += vf_v360.o
+OBJS-$(CONFIG_V360_OPENCL_FILTER)+= vf_v360_opencl.o opencl.o \
+opencl/v360.o
OBJS-$(CONFIG_VAGUEDENOISER_FILTER) += vf_vaguedenoiser.o
OBJS-$(CONFIG_VARBLUR_FILTER)+= vf_varblur.o framesync.o
OBJS-$(CONFIG_VECTORSCOPE_FILTER)+= vf_vectorscope.o
diff --git a/libavfilter/allfilters.c b/libavfilter/allfilters.c
index 409ab5d3c4..04f1925c14 100644
--- a/libavfilter/allfilters.c
+++ b/libavfilter/allfilters.c
@@ -462,6 +462,7 @@ extern const AVFilter ff_vf_unsharp_opencl;
extern const AVFilter ff_vf_untile;
extern const AVFilter ff_vf_uspp;
extern const AVFilter ff_vf_v360;
+extern const AVFilter ff_vf_v360_opencl;
extern const AVFilter ff_vf_vaguedenoiser;
extern const AVFilter ff_vf_varblur;
extern const AVFilter ff_vf_vectorscope;
diff --git a/libavfilter/opencl/v360.cl b/libavfilter/opencl/v360.cl
new file mode 100644
index 00..4acb5a025a
--- /dev/null
+++ b/libavfilter/opencl/v360.cl
@@ -0,0 +1,419 @@
+/*
+ * Copyright (c) 2021 Paul B Mahol
+ *
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE | CLK_FILTER_LINEAR);
+
+static float scale(float x, float s)
+{
+return (0.5 * x + 0.5) * (s - 1.);
+}
+
+static float rescale(int x, float s)
+{
+return (2. * x + 1.) / s - 1.;
+}
+
+__kernel void equirect_to_xyz(global float3 *dst,
+ float2 flat_range)
+{
+int2 p = (int2)(get_global_id(0), get_global_id(1));
+int2 size = (int2)(get_global_size(0), get_global_size(1));
+float2 f;
+
+f.x = flat_range.x * rescale(p.x, size.x);
+f.y = flat_range.y * rescale(p.y, size.y);
+
+float sin_phi = sin(f.x);
+float cos_phi = cos(f.x);
+float sin_theta = sin(f.y);
+float cos_theta = cos(f.y);
+
+float3 vec;
+
+vec.x = cos_theta * sin_phi;
+vec.y = sin_theta;
+vec.z = cos_theta * cos_phi;
+
+dst[p.y * size.x + p.x] = vec;
+}
+
+__kernel void flat_to_xyz(global float3 *dst,
+ float2 flat_range)
+{
+int2 p = (int2)(get_global_id(0), get_global_id(1));
+int2 size = (int2)(get_global_size(0), get_global_size(1));
+float2 f;
+
+f.x = flat_range.x * rescale(p.x, size.x);
+f.y = flat_range.y * rescale(p.y, size.y);
+
+float3 vec;
+
+vec.x = f.x;
+vec.y = f.y;
+vec.z = 1.0;
+
+vec = normalize(vec);
+
+dst[p.y * size.x + p.x] = vec;
+}
+
+__kernel void fisheye_to_xyz(global float3 *dst,
+ float2 flat_range)
+{
+int2 p = (int2)(get_global_id(0), get_global_id(1));
+int2 size = (int2)(get_global_size(0), get_global_size(1));
+float2 uv;
+
+uv.x = flat_range.x * rescale(p.x, size.x);
+uv.y = flat_range.y * rescale(p.y, size.y);
+
+float phi = atan2(uv.y, uv.x);
+float theta = M_PI_2 * (1.f - hypot(uv.x, uv.y));
+
+float sin_phi = sin(phi);
+float cos_phi = cos(phi);
+float sin_theta = sin(theta);
+float cos_theta = cos(theta);
+
+float3 vec;
+
+vec.x = cos_theta * cos_phi;
+vec.y = cos_theta * sin_phi;
+vec.z = sin_theta;
+
+vec = normalize(vec);
+
+dst[p.y * size.x + p.x] = vec;
+}
+
+__kernel void xyz_to_flat(global float2 *dst,
+ float2 iflat_range,
+ global float3 *m,
+ __read_only imag
[FFmpeg-devel] [PATCH 1/1] swscale/swscale: check SWS_PRINT_INFO flag for printing alignment warnings
This makes output consistent with a similar warning just few
lines above where this flag is checked in the same way.
Signed-off-by: softworkz
---
libswscale/swscale.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/libswscale/swscale.c b/libswscale/swscale.c
index 75cde31f4f..feef3482d2 100644
--- a/libswscale/swscale.c
+++ b/libswscale/swscale.c
@@ -329,7 +329,7 @@ static int swscale(SwsContext *c, const uint8_t *src[],
) {
SwsContext *const ctx = c->parent ? c->parent : c;
int cpu_flags = av_get_cpu_flags();
-if (HAVE_MMXEXT && (cpu_flags & AV_CPU_FLAG_SSE2) &&
+if (flags & SWS_PRINT_INFO && HAVE_MMXEXT && (cpu_flags &
AV_CPU_FLAG_SSE2) &&
!atomic_exchange_explicit(&ctx->stride_unaligned_warned,1,
memory_order_relaxed)) {
av_log(c, AV_LOG_WARNING, "Warning: data is not aligned! This can
lead to a speed loss\n");
}
--
2.30.2.windows.1
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
To unsubscribe, visit link above, or email
ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-devel] [PATCH v5 1/9] cbs_av1: fix incorrect data type
On 10/14/2021 11:04 PM, Wang, Fei W wrote: On Tue, 2021-10-12 at 16:23 +0800, Fei Wang wrote: Since order_hint_bits_minus_1 range is 0~7, cur_frame_hint can be most 128. And similar return value for cbs_av1_get_relative_dist. So if plus them and use int8_t for the result may lose its precision. Signed-off-by: Fei Wang --- update: 1. move additional film grain frame from av1dec.c to vaapi_av1.c 2. patch 3~5 can fix clip: https://drive.google.com/file/d/1Qdx_18_BFcFf_5_XXSZOVohLpaAb-yIw/view?usp=sharing libavcodec/cbs_av1_syntax_template.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/libavcodec/cbs_av1_syntax_template.c b/libavcodec/cbs_av1_syntax_template.c index 6fe6e9a4f3..d98d3d42de 100644 --- a/libavcodec/cbs_av1_syntax_template.c +++ b/libavcodec/cbs_av1_syntax_template.c @@ -355,7 +355,7 @@ static int FUNC(set_frame_refs)(CodedBitstreamContext *ctx, RWContext *rw, AV1_REF_FRAME_ALTREF2, AV1_REF_FRAME_ALTREF }; int8_t ref_frame_idx[AV1_REFS_PER_FRAME], used_frame[AV1_NUM_REF_FRAMES]; -int8_t shifted_order_hints[AV1_NUM_REF_FRAMES]; +int16_t shifted_order_hints[AV1_NUM_REF_FRAMES]; int cur_frame_hint, latest_order_hint, earliest_order_hint, ref; int i, j; Hi James, Could you help to review this version? This fixed your comments in V4. And we tested the 3th patch with DXVA, which can also fix the clip I shared. Fei Thanks Set applied. ___ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
