diff mbox series

[FFmpeg-devel,8/9] avfilter/scale_amf: Add AMF HW scaler & color converter

Message ID 20240214015515.1027-8-ovchinnikov.dmitrii@gmail.com
State New
Headers show
Series [FFmpeg-devel,1/9] libavutil: add hwcontext_amf. | expand

Checks

Context Check Description
yinshiyou/make_loongarch64 fail Make failed
andriy/make_x86 fail Make failed

Commit Message

Dmitrii Ovchinnikov Feb. 14, 2024, 1:55 a.m. UTC
From: Evgeny Pavlov <lucenticus@gmail.com>

This commit adds two AMF filters: scale_amf & scale_amf_hq.
Both filters are using AMF hardware acceleration.
scale_amf supports simple scaling algorithms & color conversion.
scale_amf_hq supports advanced scaling algorithms & might be used
for upscaling only.

Signed-off-by: Evgeny Pavlov <lucenticus@gmail.com>
---
 configure                         |   1 +
 libavfilter/Makefile              |   2 +
 libavfilter/allfilters.c          |   2 +
 libavfilter/vf_scale_amf.c        | 266 +++++++++++++++
 libavfilter/vf_scale_amf_common.c | 515 ++++++++++++++++++++++++++++++
 libavfilter/vf_scale_amf_common.h |  71 ++++
 libavfilter/vf_scale_amf_hq.c     | 191 +++++++++++
 7 files changed, 1048 insertions(+)
 create mode 100644 libavfilter/vf_scale_amf.c
 create mode 100644 libavfilter/vf_scale_amf_common.c
 create mode 100644 libavfilter/vf_scale_amf_common.h
 create mode 100644 libavfilter/vf_scale_amf_hq.c

Comments

Timo Rothenpieler Feb. 14, 2024, 3:08 p.m. UTC | #1
On 14/02/2024 02:55, Dmitrii Ovchinnikov wrote:
> From: Evgeny Pavlov <lucenticus@gmail.com>
> 
> This commit adds two AMF filters: scale_amf & scale_amf_hq.
> Both filters are using AMF hardware acceleration.
> scale_amf supports simple scaling algorithms & color conversion.
> scale_amf_hq supports advanced scaling algorithms & might be used
> for upscaling only.

Haven't looked at the patch yet, but can't this be one filter, and it 
picks the best possible method depending on options/inputs/whatever?
Evgeny Pavlov Feb. 14, 2024, 3:27 p.m. UTC | #2
On Wed, Feb 14, 2024 at 4:08 PM Timo Rothenpieler <timo@rothenpieler.org>
wrote:

> On 14/02/2024 02:55, Dmitrii Ovchinnikov wrote:
> > From: Evgeny Pavlov <lucenticus@gmail.com>
> >
> > This commit adds two AMF filters: scale_amf & scale_amf_hq.
> > Both filters are using AMF hardware acceleration.
> > scale_amf supports simple scaling algorithms & color conversion.
> > scale_amf_hq supports advanced scaling algorithms & might be used
> > for upscaling only.
>
> Haven't looked at the patch yet, but can't this be one filter, and it
> picks the best possible method depending on options/inputs/whatever?
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel@ffmpeg.org
> https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>
> To unsubscribe, visit link above, or email
> ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".
>

AMF has 2 separate components for color conversion + simple scaling
(VideoConverter)  and for advanced scaling (HQScaler).
We've got a recommendation from the AMD AMF team to implement these
components as separate ffmpeg filters.
Dennis Mungai Feb. 14, 2024, 4:26 p.m. UTC | #3
On Wed, 14 Feb 2024, 18:28 Evgeny Pavlov, <lucenticus@gmail.com> wrote:

> On Wed, Feb 14, 2024 at 4:08 PM Timo Rothenpieler <timo@rothenpieler.org>
> wrote:
>
> > On 14/02/2024 02:55, Dmitrii Ovchinnikov wrote:
> > > From: Evgeny Pavlov <lucenticus@gmail.com>
> > >
> > > This commit adds two AMF filters: scale_amf & scale_amf_hq.
> > > Both filters are using AMF hardware acceleration.
> > > scale_amf supports simple scaling algorithms & color conversion.
> > > scale_amf_hq supports advanced scaling algorithms & might be used
> > > for upscaling only.
> >
> > Haven't looked at the patch yet, but can't this be one filter, and it
> > picks the best possible method depending on options/inputs/whatever?
> > _______________________________________________
> > ffmpeg-devel mailing list
> > ffmpeg-devel@ffmpeg.org
> > https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
> >
> > To unsubscribe, visit link above, or email
> > ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".
> >
>
> AMF has 2 separate components for color conversion + simple scaling
> (VideoConverter)  and for advanced scaling (HQScaler).
> We've got a recommendation from the AMD AMF team to implement these
> components as separate ffmpeg filters.
>


Still, this should be a single VIdeo Post-Processing (VPP) -style filter,
exposing these scaling and video post processing options as tunables
therein.

A perfect example of such an implementation that excels in such an
abstraction is intel's vpp_qsv filter, from which multiple compute, color
space conversion methods, tonemapping, etc are made available through
tunables.

Another benefit of such an abstraction would be that re-using this filter
on other GPU derivatives of the discrete silicon, eg in smaller IGPs
sharing these offload blocks would be auto-detecting logic to ensure that
even with defaults, the filter chains run.

Taking another example from Intel, they have a full H/W path for low power
encode and Post-Processing that can be automatically toggled on by specific
filter options without user intervention, guaranteeing runtime safety for
the same command(s) even on newer GPUs.

Food for thought.

>
Evgeny Pavlov Feb. 19, 2024, 11:18 a.m. UTC | #4
On Wed, Feb 14, 2024 at 5:26 PM Dennis Mungai <dmngaie@gmail.com> wrote:

> On Wed, 14 Feb 2024, 18:28 Evgeny Pavlov, <lucenticus@gmail.com> wrote:
>
> > On Wed, Feb 14, 2024 at 4:08 PM Timo Rothenpieler <timo@rothenpieler.org
> >
> > wrote:
> >
> > > On 14/02/2024 02:55, Dmitrii Ovchinnikov wrote:
> > > > From: Evgeny Pavlov <lucenticus@gmail.com>
> > > >
> > > > This commit adds two AMF filters: scale_amf & scale_amf_hq.
> > > > Both filters are using AMF hardware acceleration.
> > > > scale_amf supports simple scaling algorithms & color conversion.
> > > > scale_amf_hq supports advanced scaling algorithms & might be used
> > > > for upscaling only.
> > >
> > > Haven't looked at the patch yet, but can't this be one filter, and it
> > > picks the best possible method depending on options/inputs/whatever?
> > > _______________________________________________
> > > ffmpeg-devel mailing list
> > > ffmpeg-devel@ffmpeg.org
> > > https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
> > >
> > > To unsubscribe, visit link above, or email
> > > ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".
> > >
> >
> > AMF has 2 separate components for color conversion + simple scaling
> > (VideoConverter)  and for advanced scaling (HQScaler).
> > We've got a recommendation from the AMD AMF team to implement these
> > components as separate ffmpeg filters.
> >
>
>
> Still, this should be a single VIdeo Post-Processing (VPP) -style filter,
> exposing these scaling and video post processing options as tunables
> therein.
>
> A perfect example of such an implementation that excels in such an
> abstraction is intel's vpp_qsv filter, from which multiple compute, color
> space conversion methods, tonemapping, etc are made available through
> tunables.
>
> Another benefit of such an abstraction would be that re-using this filter
> on other GPU derivatives of the discrete silicon, eg in smaller IGPs
> sharing these offload blocks would be auto-detecting logic to ensure that
> even with defaults, the filter chains run.
>
> Taking another example from Intel, they have a full H/W path for low power
> encode and Post-Processing that can be automatically toggled on by specific
> filter options without user intervention, guaranteeing runtime safety for
> the same command(s) even on newer GPUs.
>
> Food for thought.
>
> >
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel@ffmpeg.org
> https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>
> To unsubscribe, visit link above, or email
> ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".
>

AMF Video Converter (which is used in scale_amf) is the full analog of
Intel's VPP. And in the future, it may have an option for dedicated HW
block in GPU. It has scaling, color conversion, tonemapping, HDR <> SDR
conversion features.
But HQ Scaler brings advanced scaling algorithms that are similar to AMD
FSR (https://www.amd.com/en/technologies/fidelityfx-super-resolution ) and
don’t have color conversion and cannot be mapped to HW blocks.
Do you think that it would be better to rename scale_amf to vpp_amf or some
other name? Maybe we should rename scale_amf_hq as well for better
usability (e.g. sr_amf)?
Dennis Mungai Feb. 19, 2024, 2:43 p.m. UTC | #5
On Mon, 19 Feb 2024, 14:18 Evgeny Pavlov, <lucenticus@gmail.com> wrote:

> On Wed, Feb 14, 2024 at 5:26 PM Dennis Mungai <dmngaie@gmail.com> wrote:
>
> > On Wed, 14 Feb 2024, 18:28 Evgeny Pavlov, <lucenticus@gmail.com> wrote:
> >
> > > On Wed, Feb 14, 2024 at 4:08 PM Timo Rothenpieler <
> timo@rothenpieler.org
> > >
> > > wrote:
> > >
> > > > On 14/02/2024 02:55, Dmitrii Ovchinnikov wrote:
> > > > > From: Evgeny Pavlov <lucenticus@gmail.com>
> > > > >
> > > > > This commit adds two AMF filters: scale_amf & scale_amf_hq.
> > > > > Both filters are using AMF hardware acceleration.
> > > > > scale_amf supports simple scaling algorithms & color conversion.
> > > > > scale_amf_hq supports advanced scaling algorithms & might be used
> > > > > for upscaling only.
> > > >
> > > > Haven't looked at the patch yet, but can't this be one filter, and it
> > > > picks the best possible method depending on options/inputs/whatever?
> > > > _______________________________________________
> > > > ffmpeg-devel mailing list
> > > > ffmpeg-devel@ffmpeg.org
> > > > https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
> > > >
> > > > To unsubscribe, visit link above, or email
> > > > ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".
> > > >
> > >
> > > AMF has 2 separate components for color conversion + simple scaling
> > > (VideoConverter)  and for advanced scaling (HQScaler).
> > > We've got a recommendation from the AMD AMF team to implement these
> > > components as separate ffmpeg filters.
> > >
> >
> >
> > Still, this should be a single VIdeo Post-Processing (VPP) -style filter,
> > exposing these scaling and video post processing options as tunables
> > therein.
> >
> > A perfect example of such an implementation that excels in such an
> > abstraction is intel's vpp_qsv filter, from which multiple compute, color
> > space conversion methods, tonemapping, etc are made available through
> > tunables.
> >
> > Another benefit of such an abstraction would be that re-using this filter
> > on other GPU derivatives of the discrete silicon, eg in smaller IGPs
> > sharing these offload blocks would be auto-detecting logic to ensure that
> > even with defaults, the filter chains run.
> >
> > Taking another example from Intel, they have a full H/W path for low
> power
> > encode and Post-Processing that can be automatically toggled on by
> specific
> > filter options without user intervention, guaranteeing runtime safety for
> > the same command(s) even on newer GPUs.
> >
> > Food for thought.
> >
> > >
> > _______________________________________________
> > ffmpeg-devel mailing list
> > ffmpeg-devel@ffmpeg.org
> > https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
> >
> > To unsubscribe, visit link above, or email
> > ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".
> >
>
> AMF Video Converter (which is used in scale_amf) is the full analog of
> Intel's VPP. And in the future, it may have an option for dedicated HW
> block in GPU. It has scaling, color conversion, tonemapping, HDR <> SDR
> conversion features.
> But HQ Scaler brings advanced scaling algorithms that are similar to AMD
> FSR (https://www.amd.com/en/technologies/fidelityfx-super-resolution ) and
> don’t have color conversion and cannot be mapped to HW blocks.
> Do you think that it would be better to rename scale_amf to vpp_amf or some
> other name? Maybe we should rename scale_amf_hq as well for better
> usability (e.g. sr_amf)?
>


Evgeny,

That's an excellent proposal.

With the renaming of scale_amf to vpp_amf and the extra filter abstracting
FSR style upscaling stuff to sr_amf,  it becomes absolutely clear what each
filter is meant to accomplish.

>
diff mbox series

Patch

diff --git a/configure b/configure
index f72533b7d2..3d1c44d7ae 100755
--- a/configure
+++ b/configure
@@ -3826,6 +3826,7 @@  rubberband_filter_deps="librubberband"
 sab_filter_deps="gpl swscale"
 scale2ref_filter_deps="swscale"
 scale_filter_deps="swscale"
+scale_amf_filter_deps="amf"
 scale_qsv_filter_deps="libmfx"
 scale_qsv_filter_select="qsvvpp"
 scdet_filter_select="scene_sad"
diff --git a/libavfilter/Makefile b/libavfilter/Makefile
index f6c1d641d6..a87e519b85 100644
--- a/libavfilter/Makefile
+++ b/libavfilter/Makefile
@@ -456,6 +456,8 @@  OBJS-$(CONFIG_ROBERTS_OPENCL_FILTER)         += vf_convolution_opencl.o opencl.o
 OBJS-$(CONFIG_ROTATE_FILTER)                 += vf_rotate.o
 OBJS-$(CONFIG_SAB_FILTER)                    += vf_sab.o
 OBJS-$(CONFIG_SCALE_FILTER)                  += vf_scale.o scale_eval.o
+OBJS-$(CONFIG_SCALE_AMF_FILTER)              += vf_scale_amf.o scale_eval.o vf_scale_amf_common.o
+OBJS-$(CONFIG_SCALE_AMF_FILTER)              += vf_scale_amf_hq.o scale_eval.o vf_scale_amf_common.o
 OBJS-$(CONFIG_SCALE_CUDA_FILTER)             += vf_scale_cuda.o scale_eval.o \
                                                 vf_scale_cuda.ptx.o cuda/load_helper.o
 OBJS-$(CONFIG_SCALE_NPP_FILTER)              += vf_scale_npp.o scale_eval.o
diff --git a/libavfilter/allfilters.c b/libavfilter/allfilters.c
index 149bf50997..299cfb148b 100644
--- a/libavfilter/allfilters.c
+++ b/libavfilter/allfilters.c
@@ -430,6 +430,8 @@  extern const AVFilter ff_vf_roberts_opencl;
 extern const AVFilter ff_vf_rotate;
 extern const AVFilter ff_vf_sab;
 extern const AVFilter ff_vf_scale;
+extern const AVFilter ff_vf_scale_amf;
+extern const AVFilter ff_vf_scale_amf_hq;
 extern const AVFilter ff_vf_scale_cuda;
 extern const AVFilter ff_vf_scale_npp;
 extern const AVFilter ff_vf_scale_qsv;
diff --git a/libavfilter/vf_scale_amf.c b/libavfilter/vf_scale_amf.c
new file mode 100644
index 0000000000..6d2c211d83
--- /dev/null
+++ b/libavfilter/vf_scale_amf.c
@@ -0,0 +1,266 @@ 
+/*
+ * 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
+ */
+
+/**
+ * @file
+ * scale video filter - AMF
+ */
+
+#include <stdio.h>
+#include <string.h>
+
+#include "libavutil/avassert.h"
+#include "libavutil/imgutils.h"
+#include "libavutil/opt.h"
+#include "libavutil/pixdesc.h"
+#include "libavutil/time.h"
+
+#include "libavutil/hwcontext.h"
+#include "libavutil/hwcontext_amf.h"
+
+#include "AMF/components/VideoConverter.h"
+#include "vf_scale_amf_common.h"
+
+#include "avfilter.h"
+#include "formats.h"
+#include "video.h"
+#include "scale_eval.h"
+#include "internal.h"
+
+#if CONFIG_DXVA2
+#include <d3d9.h>
+#endif
+
+#if CONFIG_D3D11VA
+#include <d3d11.h>
+#endif
+
+static int amf_scale_query_formats(AVFilterContext *avctx)
+{
+    const enum AVPixelFormat *output_pix_fmts;
+    static const enum AVPixelFormat input_pix_fmts[] = {
+        AV_PIX_FMT_NV12,
+        AV_PIX_FMT_P010,
+        AV_PIX_FMT_0RGB,
+        AV_PIX_FMT_BGR0,
+        AV_PIX_FMT_BGRA,
+        AV_PIX_FMT_RGB0,
+        AV_PIX_FMT_RGBA,
+        AV_PIX_FMT_GRAY8,
+        AV_PIX_FMT_YUV420P,
+        AV_PIX_FMT_YUV420P10,
+        AV_PIX_FMT_YUYV422,
+        AV_PIX_FMT_AMF,
+        AV_PIX_FMT_NONE,
+    };
+    static const enum AVPixelFormat output_pix_fmts_default[] = {
+        AV_PIX_FMT_AMF,
+        AV_PIX_FMT_D3D11,
+        AV_PIX_FMT_DXVA2_VLD,
+        AV_PIX_FMT_NV12,
+        AV_PIX_FMT_BGRA,
+        AV_PIX_FMT_YUV420P,
+        AV_PIX_FMT_NONE,
+    };
+    output_pix_fmts = output_pix_fmts_default;
+
+    return amf_setup_input_output_formats(avctx, input_pix_fmts, output_pix_fmts);
+}
+
+static int amf_scale_config_output(AVFilterLink *outlink)
+{
+    AVFilterContext *avctx = outlink->src;
+    AVFilterLink   *inlink = avctx->inputs[0];
+    AMFScaleContext  *ctx = avctx->priv;
+    AVAMFDeviceContextInternal * internal = NULL;
+    AVHWFramesContext *hwframes_out = NULL;
+    AMFSize out_size;
+    int err;
+    AMF_RESULT res;
+    enum AMF_VIDEO_CONVERTER_COLOR_PROFILE_ENUM amf_color_profile;
+    enum AVPixelFormat in_format;
+
+    err = amf_init_scale_config(outlink, &in_format);
+    if (err < 0)
+        return err;
+    // FIXME: add checks whether we have HW context
+    hwframes_out = (AVHWFramesContext*)ctx->hwframes_out_ref->data;
+
+    internal = (AVAMFDeviceContextInternal * )ctx->amf_device_ctx_internal->data;
+    res = internal->factory->pVtbl->CreateComponent(internal->factory, internal->context, AMFVideoConverter, &ctx->scaler);
+    AMF_RETURN_IF_FALSE(ctx, res == AMF_OK, AVERROR_FILTER_NOT_FOUND, "CreateComponent(%ls) failed with error %d\n", AMFVideoConverter, res);
+    // FIXME: add checks whether we have HW context
+    AMF_ASSIGN_PROPERTY_INT64(res, ctx->scaler, AMF_VIDEO_CONVERTER_OUTPUT_FORMAT, (amf_int32)av_amf_av_to_amf_format(hwframes_out->sw_format));
+    AMF_RETURN_IF_FALSE(avctx, res == AMF_OK, AVERROR_UNKNOWN, "AMFConverter-SetProperty() failed with error %d\n", res);
+
+    out_size.width = outlink->w;
+    out_size.height = outlink->h;
+    AMF_ASSIGN_PROPERTY_SIZE(res, ctx->scaler, AMF_VIDEO_CONVERTER_OUTPUT_SIZE, out_size);
+    AMF_RETURN_IF_FALSE(avctx, res == AMF_OK, AVERROR_UNKNOWN, "AMFConverter-SetProperty() failed with error %d\n", res);
+
+    AMF_ASSIGN_PROPERTY_INT64(res, ctx->scaler, AMF_VIDEO_CONVERTER_SCALE, (amf_int32)ctx->scale_type);
+    AMF_RETURN_IF_FALSE(avctx, res == AMF_OK, AVERROR_UNKNOWN, "AMFConverter-SetProperty() failed with error %d\n", res);
+
+    amf_color_profile = AMF_VIDEO_CONVERTER_COLOR_PROFILE_UNKNOWN;
+
+    switch(ctx->color_profile) {
+    case AMF_VIDEO_CONVERTER_COLOR_PROFILE_601:
+        if (ctx->color_range == AMF_COLOR_RANGE_FULL) {
+            amf_color_profile = AMF_VIDEO_CONVERTER_COLOR_PROFILE_FULL_601;
+        } else {
+            amf_color_profile = AMF_VIDEO_CONVERTER_COLOR_PROFILE_601;
+        }
+        break;
+    case AMF_VIDEO_CONVERTER_COLOR_PROFILE_709:
+        if (ctx->color_range == AMF_COLOR_RANGE_FULL) {
+            amf_color_profile = AMF_VIDEO_CONVERTER_COLOR_PROFILE_FULL_709;
+        } else {
+            amf_color_profile = AMF_VIDEO_CONVERTER_COLOR_PROFILE_709;
+        }
+        break;
+    case AMF_VIDEO_CONVERTER_COLOR_PROFILE_2020:
+        if (ctx->color_range == AMF_COLOR_RANGE_FULL) {
+            amf_color_profile = AMF_VIDEO_CONVERTER_COLOR_PROFILE_FULL_2020;
+        } else {
+            amf_color_profile = AMF_VIDEO_CONVERTER_COLOR_PROFILE_2020;
+        }
+        break;
+    default:
+        amf_color_profile = AMF_VIDEO_CONVERTER_COLOR_PROFILE_UNKNOWN;
+        break;
+    }
+
+    if (amf_color_profile != AMF_VIDEO_CONVERTER_COLOR_PROFILE_UNKNOWN) {
+        AMF_ASSIGN_PROPERTY_INT64(res, ctx->scaler, AMF_VIDEO_CONVERTER_COLOR_PROFILE, amf_color_profile);
+    }
+
+    if (ctx->color_range != AMF_COLOR_RANGE_UNDEFINED) {
+        AMF_ASSIGN_PROPERTY_INT64(res, ctx->scaler, AMF_VIDEO_CONVERTER_OUTPUT_COLOR_RANGE, ctx->color_range);
+    }
+
+    if (ctx->primaries != AMF_COLOR_PRIMARIES_UNDEFINED) {
+        AMF_ASSIGN_PROPERTY_INT64(res, ctx->scaler, AMF_VIDEO_CONVERTER_OUTPUT_COLOR_PRIMARIES, ctx->primaries);
+    }
+
+    if (ctx->trc != AMF_COLOR_TRANSFER_CHARACTERISTIC_UNDEFINED) {
+        AMF_ASSIGN_PROPERTY_INT64(res, ctx->scaler, AMF_VIDEO_CONVERTER_OUTPUT_TRANSFER_CHARACTERISTIC, ctx->trc);
+    }
+
+    res = ctx->scaler->pVtbl->Init(ctx->scaler, av_amf_av_to_amf_format(in_format), inlink->w, inlink->h);
+    AMF_RETURN_IF_FALSE(avctx, res == AMF_OK, AVERROR_UNKNOWN, "AMFConverter-Init() failed with error %d\n", res);
+
+    return 0;
+}
+
+#define OFFSET(x) offsetof(AMFScaleContext, x)
+#define FLAGS AV_OPT_FLAG_VIDEO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
+static const AVOption scale_amf_options[] = {
+    { "w",              "Output video width",   OFFSET(w_expr),     AV_OPT_TYPE_STRING, { .str = "iw"   }, .flags = FLAGS },
+    { "h",              "Output video height",  OFFSET(h_expr),     AV_OPT_TYPE_STRING, { .str = "ih"   }, .flags = FLAGS },
+    { "format",         "Output pixel format",  OFFSET(format_str), AV_OPT_TYPE_STRING, { .str = "same" }, .flags = FLAGS },
+
+    { "scale_type",     "Scale type",           OFFSET(scale_type),      AV_OPT_TYPE_INT,   { .i64 = AMF_VIDEO_CONVERTER_SCALE_BILINEAR }, AMF_VIDEO_CONVERTER_SCALE_BILINEAR, AMF_VIDEO_CONVERTER_SCALE_BICUBIC, FLAGS, "scale_type" },
+    { "bilinear",       "Bilinear",         0,  AV_OPT_TYPE_CONST, { .i64 = AMF_VIDEO_CONVERTER_SCALE_BILINEAR }, 0, 0, FLAGS, "scale_type" },
+    { "bicubic",        "Bicubic",          0,  AV_OPT_TYPE_CONST, { .i64 = AMF_VIDEO_CONVERTER_SCALE_BICUBIC },  0, 0, FLAGS, "scale_type" },
+
+    { "color_profile",  "Color profile",        OFFSET(color_profile), AV_OPT_TYPE_INT,   { .i64 = AMF_VIDEO_CONVERTER_COLOR_PROFILE_UNKNOWN }, AMF_VIDEO_CONVERTER_COLOR_PROFILE_UNKNOWN, AMF_VIDEO_CONVERTER_COLOR_PROFILE_FULL_2020, FLAGS, "color_profile" },
+    { "bt601",          "BT.601",           0,  AV_OPT_TYPE_CONST, { .i64 = AMF_VIDEO_CONVERTER_COLOR_PROFILE_601 }, 0, 0, FLAGS, "color_profile" },
+    { "bt709",          "BT.709",           0,  AV_OPT_TYPE_CONST, { .i64 = AMF_VIDEO_CONVERTER_COLOR_PROFILE_709 },  0, 0, FLAGS, "color_profile" },
+    { "bt2020",         "BT.2020",          0,  AV_OPT_TYPE_CONST, { .i64 = AMF_VIDEO_CONVERTER_COLOR_PROFILE_2020 },  0, 0, FLAGS, "color_profile" },
+
+    { "color_range",    "Color range",          OFFSET(color_range),      AV_OPT_TYPE_INT,   { .i64 = AMF_COLOR_RANGE_UNDEFINED }, AMF_COLOR_RANGE_UNDEFINED, AMF_COLOR_RANGE_FULL, FLAGS, "color_range" },
+    { "studio",         "Studio",                   0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_RANGE_STUDIO }, 0, 0, FLAGS, "color_range" },
+    { "full",           "Full",                     0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_RANGE_FULL }, 0, 0, FLAGS, "color_range" },
+
+    { "primaries",      "Output color primaries",   OFFSET(primaries),  AV_OPT_TYPE_INT,   { .i64 = AMF_COLOR_PRIMARIES_UNDEFINED }, AMF_COLOR_PRIMARIES_UNDEFINED, AMF_COLOR_PRIMARIES_JEDEC_P22, FLAGS, "primaries" },
+    { "bt709",          "BT.709",                   0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_PRIMARIES_BT709 }, 0, 0, FLAGS, "primaries" },
+    { "bt470m",         "BT.470M",                  0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_PRIMARIES_BT470M }, 0, 0, FLAGS, "primaries" },
+    { "bt470bg",        "BT.470BG",                 0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_PRIMARIES_BT470BG }, 0, 0, FLAGS, "primaries" },
+    { "smpte170m",      "SMPTE170M",                0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_PRIMARIES_SMPTE170M }, 0, 0, FLAGS, "primaries" },
+    { "smpte240m",      "SMPTE240M",                0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_PRIMARIES_SMPTE240M }, 0, 0, FLAGS, "primaries" },
+    { "film",           "FILM",                     0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_PRIMARIES_FILM }, 0, 0, FLAGS, "primaries" },
+    { "bt2020",         "BT2020",                   0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_PRIMARIES_BT2020 }, 0, 0, FLAGS, "primaries" },
+    { "smpte428",       "SMPTE428",                 0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_PRIMARIES_SMPTE428 }, 0, 0, FLAGS, "primaries" },
+    { "smpte431",       "SMPTE431",                 0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_PRIMARIES_SMPTE431 }, 0, 0, FLAGS, "primaries" },
+    { "smpte432",       "SMPTE432",                 0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_PRIMARIES_SMPTE432 }, 0, 0, FLAGS, "primaries" },
+    { "jedec-p22",      "JEDEC_P22",                0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_PRIMARIES_JEDEC_P22 }, 0, 0, FLAGS, "primaries" },
+
+    { "trc",            "Output transfer characteristics",  OFFSET(trc),  AV_OPT_TYPE_INT,   { .i64 = AMF_COLOR_TRANSFER_CHARACTERISTIC_UNDEFINED }, AMF_COLOR_TRANSFER_CHARACTERISTIC_UNDEFINED, AMF_COLOR_TRANSFER_CHARACTERISTIC_ARIB_STD_B67, FLAGS, "trc" },
+    { "bt709",          "BT.709",                   0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_TRANSFER_CHARACTERISTIC_BT709 }, 0, 0, FLAGS, "trc" },
+    { "gamma22",        "GAMMA22",                  0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_TRANSFER_CHARACTERISTIC_GAMMA22 }, 0, 0, FLAGS, "trc" },
+    { "gamma28",        "GAMMA28",                  0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_TRANSFER_CHARACTERISTIC_GAMMA28 }, 0, 0, FLAGS, "trc" },
+    { "smpte170m",      "SMPTE170M",                0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_TRANSFER_CHARACTERISTIC_SMPTE170M }, 0, 0, FLAGS, "trc" },
+    { "smpte240m",      "SMPTE240M",                0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_TRANSFER_CHARACTERISTIC_SMPTE240M }, 0, 0, FLAGS, "trc" },
+    { "linear",         "Linear",                   0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_TRANSFER_CHARACTERISTIC_LINEAR }, 0, 0, FLAGS, "trc" },
+    { "log",            "LOG",                      0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_TRANSFER_CHARACTERISTIC_LOG }, 0, 0, FLAGS, "trc" },
+    { "log-sqrt",       "LOG_SQRT",                 0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_TRANSFER_CHARACTERISTIC_LOG_SQRT }, 0, 0, FLAGS, "trc" },
+    { "iec61966-2-4",   "IEC61966_2_4",             0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_TRANSFER_CHARACTERISTIC_IEC61966_2_4 }, 0, 0, FLAGS, "trc" },
+    { "bt1361-ecg",     "BT1361_ECG",               0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_TRANSFER_CHARACTERISTIC_BT1361_ECG }, 0, 0, FLAGS, "trc" },
+    { "iec61966-2-1",   "IEC61966_2_1",             0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_TRANSFER_CHARACTERISTIC_IEC61966_2_1 }, 0, 0, FLAGS, "trc" },
+    { "bt2020-10",      "BT.2020_10",               0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_TRANSFER_CHARACTERISTIC_BT2020_10 }, 0, 0, FLAGS, "trc" },
+    { "bt2020-12",      "BT.2020-12",               0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_TRANSFER_CHARACTERISTIC_BT2020_12 }, 0, 0, FLAGS, "trc" },
+    { "smpte2084",      "SMPTE2084",                0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_TRANSFER_CHARACTERISTIC_SMPTE2084 }, 0, 0, FLAGS, "trc" },
+    { "smpte428",       "SMPTE428",                 0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_TRANSFER_CHARACTERISTIC_SMPTE428 }, 0, 0, FLAGS, "trc" },
+    { "arib-std-b67",   "ARIB_STD_B67",             0,  AV_OPT_TYPE_CONST, { .i64 = AMF_COLOR_TRANSFER_CHARACTERISTIC_ARIB_STD_B67 }, 0, 0, FLAGS, "trc" },
+
+    { "force_original_aspect_ratio", "decrease or increase w/h if necessary to keep the original AR", OFFSET(force_original_aspect_ratio), AV_OPT_TYPE_INT, { .i64 = 0}, 0, 2, FLAGS, "force_oar" },
+    { "disable",  NULL, 0, AV_OPT_TYPE_CONST, {.i64 = 0 }, 0, 0, FLAGS, "force_oar" },
+    { "decrease", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = 1 }, 0, 0, FLAGS, "force_oar" },
+    { "increase", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = 2 }, 0, 0, FLAGS, "force_oar" },
+    { "force_divisible_by", "enforce that the output resolution is divisible by a defined integer when force_original_aspect_ratio is used", OFFSET(force_divisible_by), AV_OPT_TYPE_INT, { .i64 = 1}, 1, 256, FLAGS },
+
+    { NULL },
+};
+
+
+AVFILTER_DEFINE_CLASS(scale_amf);
+
+static const AVFilterPad amf_scale_inputs[] = {
+    {
+        .name         = "default",
+        .type         = AVMEDIA_TYPE_VIDEO,
+        .filter_frame = amf_scale_filter_frame,
+    }
+};
+
+static const AVFilterPad amf_scale_outputs[] = {
+    {
+        .name         = "default",
+        .type         = AVMEDIA_TYPE_VIDEO,
+        .config_props = amf_scale_config_output,
+    }
+};
+
+AVFilter ff_vf_scale_amf = {
+    .name      = "scale_amf",
+    .description = NULL_IF_CONFIG_SMALL("AMF video scaling and format conversion"),
+
+    .init          = amf_scale_init,
+    .uninit        = amf_scale_uninit,
+    FILTER_QUERY_FUNC(&amf_scale_query_formats),
+
+    .priv_size = sizeof(AMFScaleContext),
+    .priv_class = &scale_amf_class,
+
+    FILTER_INPUTS(amf_scale_inputs),
+    FILTER_OUTPUTS(amf_scale_outputs),
+    FILTER_SINGLE_PIXFMT(AV_PIX_FMT_AMF),
+
+    .flags_internal = FF_FILTER_FLAG_HWFRAME_AWARE,
+    .flags          = AVFILTER_FLAG_HWDEVICE,
+};
diff --git a/libavfilter/vf_scale_amf_common.c b/libavfilter/vf_scale_amf_common.c
new file mode 100644
index 0000000000..5c26ebd2c9
--- /dev/null
+++ b/libavfilter/vf_scale_amf_common.c
@@ -0,0 +1,515 @@ 
+/*
+ * 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
+ */
+
+#include "vf_scale_amf_common.h"
+
+#include "libavutil/avassert.h"
+#include "avfilter.h"
+#include "internal.h"
+#include "formats.h"
+#include "libavutil/imgutils.h"
+
+#include "libavutil/hwcontext_amf.h"
+#include "AMF/components/ColorSpace.h"
+#include "scale_eval.h"
+
+#if CONFIG_DXVA2
+#include <d3d9.h>
+#endif
+
+#if CONFIG_D3D11VA
+#include <d3d11.h>
+#endif
+
+int amf_scale_init(AVFilterContext *avctx)
+{
+    AMFScaleContext     *ctx = avctx->priv;
+
+    if (!strcmp(ctx->format_str, "same")) {
+        ctx->format = AV_PIX_FMT_NONE;
+    } else {
+        ctx->format = av_get_pix_fmt(ctx->format_str);
+        if (ctx->format == AV_PIX_FMT_NONE) {
+            av_log(avctx, AV_LOG_ERROR, "Unrecognized pixel format: %s\n", ctx->format_str);
+            return AVERROR(EINVAL);
+        }
+    }
+
+    return 0;
+}
+
+void amf_scale_uninit(AVFilterContext *avctx)
+{
+    AMFScaleContext *ctx = avctx->priv;
+
+    if (ctx->scaler) {
+        ctx->scaler->pVtbl->Terminate(ctx->scaler);
+        ctx->scaler->pVtbl->Release(ctx->scaler);
+        ctx->scaler = NULL;
+    }
+
+    av_buffer_unref(&ctx->amf_device_ref);
+    av_buffer_unref(&ctx->hwdevice_ref);
+    av_buffer_unref(&ctx->hwframes_in_ref);
+    av_buffer_unref(&ctx->hwframes_out_ref);
+}
+
+int amf_scale_filter_frame(AVFilterLink *inlink, AVFrame *in)
+{
+    AVFilterContext             *avctx = inlink->dst;
+    AMFScaleContext             *ctx = avctx->priv;
+    AVFilterLink                *outlink = avctx->outputs[0];
+    AMF_RESULT  res;
+    AMFSurface *surface_in;
+    AMFSurface *surface_out;
+    AMFData *data_out = NULL;
+    enum AVColorSpace out_colorspace;
+    enum AVColorRange out_color_range;
+
+    AVFrame *out = NULL;
+    int ret = 0;
+
+    if (!ctx->scaler)
+        return AVERROR(EINVAL);
+
+    ret = amf_avframe_to_amfsurface(avctx, in, &surface_in);
+    if (ret < 0)
+        goto fail;
+
+    res = ctx->scaler->pVtbl->SubmitInput(ctx->scaler, (AMFData*)surface_in);
+    AMF_GOTO_FAIL_IF_FALSE(avctx, res == AMF_OK, AVERROR_UNKNOWN, "SubmitInput() failed with error %d\n", res);
+    res = ctx->scaler->pVtbl->QueryOutput(ctx->scaler, &data_out);
+    AMF_GOTO_FAIL_IF_FALSE(avctx, res == AMF_OK, AVERROR_UNKNOWN, "QueryOutput() failed with error %d\n", res);
+
+    if (data_out) {
+        AMFGuid guid = IID_AMFSurface();
+        data_out->pVtbl->QueryInterface(data_out, &guid, (void**)&surface_out); // query for buffer interface
+        data_out->pVtbl->Release(data_out);
+    }
+
+    out = amf_amfsurface_to_avframe(avctx, surface_out);
+
+    ret = av_frame_copy_props(out, in);
+    av_frame_unref(in);
+
+    out_colorspace = AVCOL_SPC_UNSPECIFIED;
+
+    if (ctx->color_profile != AMF_VIDEO_CONVERTER_COLOR_PROFILE_UNKNOWN) {
+        switch(ctx->color_profile) {
+        case AMF_VIDEO_CONVERTER_COLOR_PROFILE_601:
+            out_colorspace = AVCOL_SPC_SMPTE170M;
+        break;
+        case AMF_VIDEO_CONVERTER_COLOR_PROFILE_709:
+            out_colorspace = AVCOL_SPC_BT709;
+        break;
+        case AMF_VIDEO_CONVERTER_COLOR_PROFILE_2020:
+            out_colorspace = AVCOL_SPC_BT2020_NCL;
+        break;
+        case AMF_VIDEO_CONVERTER_COLOR_PROFILE_JPEG:
+            out_colorspace = AVCOL_SPC_RGB;
+        break;
+        default:
+            out_colorspace = AVCOL_SPC_UNSPECIFIED;
+        break;
+        }
+        out->colorspace = out_colorspace;
+    }
+
+    out_color_range = AVCOL_RANGE_UNSPECIFIED;
+    if (ctx->color_range == AMF_COLOR_RANGE_FULL)
+        out_color_range = AVCOL_RANGE_JPEG;
+    else if (ctx->color_range == AMF_COLOR_RANGE_STUDIO)
+        out_color_range = AVCOL_RANGE_MPEG;
+
+    if (ctx->color_range != AMF_COLOR_RANGE_UNDEFINED)
+        out->color_range = out_color_range;
+
+    if (ctx->primaries != AMF_COLOR_PRIMARIES_UNDEFINED)
+        out->color_primaries = ctx->primaries;
+
+    if (ctx->trc != AMF_COLOR_TRANSFER_CHARACTERISTIC_UNDEFINED)
+        out->color_trc = ctx->trc;
+
+
+    if (ret < 0)
+        goto fail;
+
+    out->format = outlink->format;
+    out->width  = outlink->w;
+    out->height = outlink->h;
+
+    out->hw_frames_ctx = av_buffer_ref(ctx->hwframes_out_ref);
+    if (!out->hw_frames_ctx) {
+        ret = AVERROR(ENOMEM);
+        goto fail;
+    }
+
+    if (inlink->sample_aspect_ratio.num) {
+        outlink->sample_aspect_ratio = av_mul_q((AVRational){outlink->h * inlink->w, outlink->w * inlink->h}, inlink->sample_aspect_ratio);
+    } else
+        outlink->sample_aspect_ratio = inlink->sample_aspect_ratio;
+
+    av_frame_free(&in);
+    return ff_filter_frame(outlink, out);
+fail:
+    av_frame_free(&in);
+    av_frame_free(&out);
+    return ret;
+}
+
+
+
+int amf_setup_input_output_formats(AVFilterContext *avctx,
+                                    const enum AVPixelFormat *input_pix_fmts,
+                                    const enum AVPixelFormat *output_pix_fmts)
+{
+    int err;
+    int i;
+    AVFilterFormats *input_formats;
+
+    //in case if hw_device_ctx is set to DXVA2 we change order of pixel formats to set DXVA2 be choosen by default
+    //The order is ignored if hw_frames_ctx is not NULL on the config_output stage
+    if (avctx->hw_device_ctx) {
+        AVHWDeviceContext *device_ctx = (AVHWDeviceContext*)avctx->hw_device_ctx->data;
+
+        switch (device_ctx->type) {
+    #if CONFIG_D3D11VA
+        case AV_HWDEVICE_TYPE_D3D11VA:
+            {
+                static const enum AVPixelFormat output_pix_fmts_d3d11[] = {
+                    AV_PIX_FMT_D3D11,
+                    AV_PIX_FMT_NONE,
+                };
+                output_pix_fmts = output_pix_fmts_d3d11;
+            }
+            break;
+    #endif
+    #if CONFIG_DXVA2
+        case AV_HWDEVICE_TYPE_DXVA2:
+            {
+                static const enum AVPixelFormat output_pix_fmts_dxva2[] = {
+                    AV_PIX_FMT_DXVA2_VLD,
+                    AV_PIX_FMT_NONE,
+                };
+                output_pix_fmts = output_pix_fmts_dxva2;
+            }
+            break;
+    #endif
+        default:
+            {
+                av_log(avctx, AV_LOG_ERROR, "Unsupported device : %s\n", av_hwdevice_get_type_name(device_ctx->type));
+                return AVERROR(EINVAL);
+            }
+            break;
+        }
+    }
+
+    input_formats = ff_make_format_list(output_pix_fmts);
+    if (!input_formats) {
+        return AVERROR(ENOMEM);
+    }
+
+    for (i = 0; input_pix_fmts[i] != AV_PIX_FMT_NONE; i++) {
+        err = ff_add_format(&input_formats, input_pix_fmts[i]);
+        if (err < 0)
+            return err;
+    }
+
+    if ((err = ff_formats_ref(input_formats, &avctx->inputs[0]->outcfg.formats)) < 0 ||
+        (err = ff_formats_ref(ff_make_format_list(output_pix_fmts),
+                              &avctx->outputs[0]->incfg.formats)) < 0)
+        return err;
+    return 0;
+}
+
+int amf_copy_surface(AVFilterContext *avctx, const AVFrame *frame,
+    AMFSurface* surface)
+{
+    AMFPlane *plane;
+    uint8_t  *dst_data[4];
+    int       dst_linesize[4];
+    int       planes;
+    int       i;
+
+    planes = surface->pVtbl->GetPlanesCount(surface);
+    av_assert0(planes < FF_ARRAY_ELEMS(dst_data));
+
+    for (i = 0; i < planes; i++) {
+        plane = surface->pVtbl->GetPlaneAt(surface, i);
+        dst_data[i] = plane->pVtbl->GetNative(plane);
+        dst_linesize[i] = plane->pVtbl->GetHPitch(plane);
+    }
+    av_image_copy(dst_data, dst_linesize,
+        (const uint8_t**)frame->data, frame->linesize, frame->format,
+        frame->width, frame->height);
+
+    return 0;
+}
+
+int amf_init_scale_config(AVFilterLink *outlink, enum AVPixelFormat *in_format)
+{
+    AVFilterContext *avctx = outlink->src;
+    AVFilterLink   *inlink = avctx->inputs[0];
+    AMFScaleContext  *ctx = avctx->priv;
+    AVHWFramesContext *hwframes_out;
+    int err;
+    AMF_RESULT res;
+
+    if ((err = ff_scale_eval_dimensions(avctx,
+                                        ctx->w_expr, ctx->h_expr,
+                                        inlink, outlink,
+                                        &ctx->width, &ctx->height)) < 0)
+        return err;
+
+    ff_scale_adjust_dimensions(inlink, &ctx->width, &ctx->height,
+                               ctx->force_original_aspect_ratio, ctx->force_divisible_by);
+
+    av_buffer_unref(&ctx->amf_device_ref);
+    av_buffer_unref(&ctx->hwframes_in_ref);
+    av_buffer_unref(&ctx->hwframes_out_ref);
+
+    if (inlink->hw_frames_ctx) {
+        AVHWFramesContext *frames_ctx = (AVHWFramesContext*)inlink->hw_frames_ctx->data;
+        if (frames_ctx->device_ctx->type == AV_HWDEVICE_TYPE_AMF) {
+            AVAMFDeviceContext * amf_ctx =  frames_ctx->device_ctx->hwctx;
+            ctx->amf_device_ctx_internal = av_buffer_ref(amf_ctx->internal);
+        }
+        if (av_amf_av_to_amf_format(frames_ctx->sw_format) == AMF_SURFACE_UNKNOWN) {
+            av_log(avctx, AV_LOG_ERROR, "Format of input frames context (%s) is not supported by AMF.\n",
+                   av_get_pix_fmt_name(frames_ctx->sw_format));
+            return AVERROR(EINVAL);
+        }
+
+        err = av_hwdevice_ctx_create_derived(&ctx->amf_device_ref, AV_HWDEVICE_TYPE_AMF, frames_ctx->device_ref, 0);
+        if (err < 0)
+            return err;
+
+        ctx->hwframes_in_ref = av_buffer_ref(inlink->hw_frames_ctx);
+        if (!ctx->hwframes_in_ref)
+            return AVERROR(ENOMEM);
+
+        ctx->hwframes_out_ref = av_hwframe_ctx_alloc(frames_ctx->device_ref);
+        if (!ctx->hwframes_out_ref)
+            return AVERROR(ENOMEM);
+
+        hwframes_out = (AVHWFramesContext*)ctx->hwframes_out_ref->data;
+        hwframes_out->format    = outlink->format;
+        hwframes_out->sw_format = frames_ctx->sw_format;
+    } else if (avctx->hw_device_ctx) {
+        AVHWDeviceContext   *hwdev_ctx;
+        err = av_hwdevice_ctx_create_derived(&ctx->amf_device_ref, AV_HWDEVICE_TYPE_AMF, avctx->hw_device_ctx, 0);
+        if (err < 0)
+            return err;
+        hwdev_ctx = (AVHWDeviceContext*)avctx->hw_device_ctx->data;
+        if (hwdev_ctx->type == AV_HWDEVICE_TYPE_AMF)
+        {
+            AVAMFDeviceContext * amf_ctx =  hwdev_ctx->hwctx;
+            ctx->amf_device_ctx_internal = av_buffer_ref(amf_ctx->internal);
+        }
+        ctx->hwdevice_ref = av_buffer_ref(avctx->hw_device_ctx);
+        if (!ctx->hwdevice_ref)
+            return AVERROR(ENOMEM);
+
+        ctx->hwframes_out_ref = av_hwframe_ctx_alloc(ctx->hwdevice_ref);
+        if (!ctx->hwframes_out_ref)
+            return AVERROR(ENOMEM);
+
+        hwframes_out = (AVHWFramesContext*)ctx->hwframes_out_ref->data;
+        hwframes_out->format    = AV_PIX_FMT_AMF;
+        hwframes_out->sw_format = outlink->format;
+    } else {
+        AVAMFDeviceContextInternal *wrapped = av_mallocz(sizeof(*wrapped));
+        ctx->amf_device_ctx_internal = av_buffer_create((uint8_t *)wrapped, sizeof(*wrapped),
+                                                av_amf_context_internal_free, NULL, 0);
+        if ((res == av_amf_context_internal_create((AVAMFDeviceContextInternal *)ctx->amf_device_ctx_internal->data, avctx, "", NULL, 0)) != 0) {
+            return res;
+        }
+        ctx->hwframes_out_ref = av_hwframe_ctx_alloc(ctx->amf_device_ref);
+        if (!ctx->hwframes_out_ref)
+            return AVERROR(ENOMEM);
+
+        hwframes_out = (AVHWFramesContext*)ctx->hwframes_out_ref->data;
+        hwframes_out->format    = outlink->format;
+        hwframes_out->sw_format = inlink->format;
+    }
+
+    if (ctx->format != AV_PIX_FMT_NONE) {
+        hwframes_out->sw_format = ctx->format;
+    }
+
+    if (inlink->format == AV_PIX_FMT_AMF) {
+        if (!inlink->hw_frames_ctx || !inlink->hw_frames_ctx->data)
+            return AVERROR(EINVAL);
+        else
+            *in_format = ((AVHWFramesContext*)inlink->hw_frames_ctx->data)->sw_format;
+    } else
+        *in_format = inlink->format;
+
+    outlink->w = ctx->width;
+    outlink->h = ctx->height;
+
+    hwframes_out->width = outlink->w;
+    hwframes_out->height = outlink->h;
+
+    err = av_hwframe_ctx_init(ctx->hwframes_out_ref);
+    if (err < 0)
+        return err;
+
+    outlink->hw_frames_ctx = av_buffer_ref(ctx->hwframes_out_ref);
+    if (!outlink->hw_frames_ctx) {
+        return AVERROR(ENOMEM);
+    }
+    return 0;
+}
+
+void amf_free_amfsurface(void *opaque, uint8_t *data)
+{
+    AMFSurface *surface = (AMFSurface*)data;
+    surface->pVtbl->Release(surface);
+}
+
+AVFrame *amf_amfsurface_to_avframe(AVFilterContext *avctx, AMFSurface* pSurface)
+{
+    AVFrame *frame = av_frame_alloc();
+    AMFScaleContext  *ctx = avctx->priv;
+
+    if (!frame)
+        return NULL;
+
+    if (ctx->hwframes_out_ref) {
+        AVHWFramesContext *hwframes_out = (AVHWFramesContext *)ctx->hwframes_out_ref->data;
+        if (hwframes_out->format == AV_PIX_FMT_AMF) {
+            int ret = av_hwframe_get_buffer(ctx->hwframes_out_ref, frame, 0);
+            if (ret < 0) {
+                av_log(avctx, AV_LOG_ERROR, "Get hw frame failed.\n");
+                av_frame_free(&frame);
+                return NULL;
+            }
+            frame->data[3] = (uint8_t *)pSurface;
+            frame->buf[1] = av_buffer_create((uint8_t *)pSurface, sizeof(AMFSurface),
+                                            amf_free_amfsurface,
+                                            (void*)avctx,
+                                            AV_BUFFER_FLAG_READONLY);
+        } else { // FIXME: add processing of other hw formats
+            av_log(ctx, AV_LOG_ERROR, "Unknown pixel format\n");
+            return NULL;
+        }
+    } else {
+
+        switch (pSurface->pVtbl->GetMemoryType(pSurface))
+        {
+    #if CONFIG_D3D11VA
+            case AMF_MEMORY_DX11:
+            {
+                AMFPlane *plane0 = pSurface->pVtbl->GetPlaneAt(pSurface, 0);
+                frame->data[0] = plane0->pVtbl->GetNative(plane0);
+                frame->data[1] = (uint8_t*)(intptr_t)0;
+
+                frame->buf[0] = av_buffer_create(NULL,
+                                        0,
+                                        amf_free_amfsurface,
+                                        pSurface,
+                                        AV_BUFFER_FLAG_READONLY);
+            }
+            break;
+    #endif
+    #if CONFIG_DXVA2
+            case AMF_MEMORY_DX9:
+            {
+                AMFPlane *plane0 = pSurface->pVtbl->GetPlaneAt(pSurface, 0);
+                frame->data[3] = plane0->pVtbl->GetNative(plane0);
+
+                frame->buf[0] = av_buffer_create(NULL,
+                                        0,
+                                        amf_free_amfsurface,
+                                        pSurface,
+                                        AV_BUFFER_FLAG_READONLY);
+            }
+            break;
+    #endif
+        default:
+            {
+                av_log(avctx, AV_LOG_ERROR, "Unsupported memory type : %d\n", pSurface->pVtbl->GetMemoryType(pSurface));
+                return NULL;
+            }
+        }
+    }
+
+    return frame;
+}
+
+int amf_avframe_to_amfsurface(AVFilterContext *avctx, const AVFrame *frame, AMFSurface** ppSurface)
+{
+    AMFScaleContext *ctx = avctx->priv;
+    AVAMFDeviceContextInternal* internal = (AVAMFDeviceContextInternal *)ctx->amf_device_ctx_internal->data;
+    AMFSurface *surface;
+    AMF_RESULT  res;
+    int hw_surface = 0;
+
+    switch (frame->format) {
+#if CONFIG_D3D11VA
+    case AV_PIX_FMT_D3D11:
+        {
+            static const GUID AMFTextureArrayIndexGUID = { 0x28115527, 0xe7c3, 0x4b66, { 0x99, 0xd3, 0x4f, 0x2a, 0xe6, 0xb4, 0x7f, 0xaf } };
+            ID3D11Texture2D *texture = (ID3D11Texture2D*)frame->data[0]; // actual texture
+            int index = (intptr_t)frame->data[1]; // index is a slice in texture array is - set to tell AMF which slice to use
+            texture->lpVtbl->SetPrivateData(texture, &AMFTextureArrayIndexGUID, sizeof(index), &index);
+
+            res = internal->context->pVtbl->CreateSurfaceFromDX11Native(internal->context, texture, &surface, NULL); // wrap to AMF surface
+            AMF_RETURN_IF_FALSE(avctx, res == AMF_OK, AVERROR(ENOMEM), "CreateSurfaceFromDX11Native() failed  with error %d\n", res);
+            hw_surface = 1;
+        }
+        break;
+#endif
+    case AV_PIX_FMT_AMF:
+        {
+            surface = (AMFSurface*)frame->data[3]; // actual surface
+            hw_surface = 1;
+        }
+        break;
+
+#if CONFIG_DXVA2
+    case AV_PIX_FMT_DXVA2_VLD:
+        {
+            IDirect3DSurface9 *texture = (IDirect3DSurface9 *)frame->data[3]; // actual texture
+
+            res = internal->context->pVtbl->CreateSurfaceFromDX9Native(internal->context, texture, &surface, NULL); // wrap to AMF surface
+            AMF_RETURN_IF_FALSE(avctx, res == AMF_OK, AVERROR(ENOMEM), "CreateSurfaceFromDX9Native() failed  with error %d\n", res);
+            hw_surface = 1;
+        }
+        break;
+#endif
+    default:
+        {
+            AMF_SURFACE_FORMAT amf_fmt = av_amf_av_to_amf_format(frame->format);
+            res = internal->context->pVtbl->AllocSurface(internal->context, AMF_MEMORY_HOST, amf_fmt, frame->width, frame->height, &surface);
+            AMF_RETURN_IF_FALSE(avctx, res == AMF_OK, AVERROR(ENOMEM), "AllocSurface() failed  with error %d\n", res);
+            amf_copy_surface(avctx, frame, surface);
+        }
+        break;
+    }
+
+    if (hw_surface) {
+        // input HW surfaces can be vertically aligned by 16; tell AMF the real size
+        surface->pVtbl->SetCrop(surface, 0, 0, frame->width, frame->height);
+    }
+
+    surface->pVtbl->SetPts(surface, frame->pts);
+    *ppSurface = surface;
+    return 0;
+}
diff --git a/libavfilter/vf_scale_amf_common.h b/libavfilter/vf_scale_amf_common.h
new file mode 100644
index 0000000000..a43c7602cb
--- /dev/null
+++ b/libavfilter/vf_scale_amf_common.h
@@ -0,0 +1,71 @@ 
+/*
+ * 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
+ */
+
+#ifndef AVFILTER_AMF_COMMON_H
+#define AVFILTER_AMF_COMMON_H
+
+#include "avfilter.h"
+
+#include "AMF/core/Surface.h"
+#include "AMF/components/Component.h"
+
+typedef struct AMFScaleContext {
+    const AVClass *class;
+
+    int width, height;
+    enum AVPixelFormat format;
+    int scale_type;
+    int color_profile;
+    int color_range;
+    int primaries;
+    int trc;
+    int fill;
+    int fill_color;
+    int keep_ratio;
+
+    // HQScaler properties
+    int algorithm;
+    float sharpness;
+
+    char *w_expr;
+    char *h_expr;
+    char *format_str;
+    int force_original_aspect_ratio;
+    int force_divisible_by;
+
+    AMFComponent        *scaler;
+    AVBufferRef         *amf_device_ref;
+
+    AVBufferRef         *hwframes_in_ref;
+    AVBufferRef         *hwframes_out_ref;
+    AVBufferRef         *hwdevice_ref;
+
+    AVBufferRef        *amf_device_ctx_internal;
+} AMFScaleContext;
+
+int amf_scale_init(AVFilterContext *avctx);
+void amf_scale_uninit(AVFilterContext *avctx);
+int amf_init_scale_config(AVFilterLink *outlink, enum AVPixelFormat *in_format);
+int amf_copy_surface(AVFilterContext *avctx, const AVFrame *frame, AMFSurface* surface);
+void amf_free_amfsurface(void *opaque, uint8_t *data);
+AVFrame *amf_amfsurface_to_avframe(AVFilterContext *avctx, AMFSurface* pSurface);
+int amf_avframe_to_amfsurface(AVFilterContext *avctx, const AVFrame *frame, AMFSurface** ppSurface);
+int amf_setup_input_output_formats(AVFilterContext *avctx, const enum AVPixelFormat *input_pix_fmts, const enum AVPixelFormat *output_pix_fmts);
+int amf_scale_filter_frame(AVFilterLink *inlink, AVFrame *in);
+
+#endif /* AVFILTER_AMF_COMMON_H */
diff --git a/libavfilter/vf_scale_amf_hq.c b/libavfilter/vf_scale_amf_hq.c
new file mode 100644
index 0000000000..63c9bd8301
--- /dev/null
+++ b/libavfilter/vf_scale_amf_hq.c
@@ -0,0 +1,191 @@ 
+/*
+ * 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
+ */
+
+/**
+ * @file
+ * scale video filter - AMF
+ */
+
+#include <stdio.h>
+#include <string.h>
+
+#include "libavutil/avassert.h"
+#include "libavutil/imgutils.h"
+#include "libavutil/opt.h"
+#include "libavutil/pixdesc.h"
+#include "libavutil/time.h"
+
+#include "libavutil/hwcontext.h"
+#include "libavutil/hwcontext_amf.h"
+
+#include "AMF/components/HQScaler.h"
+#include "AMF/components/ColorSpace.h"
+#include "vf_scale_amf_common.h"
+
+#include "avfilter.h"
+#include "internal.h"
+#include "formats.h"
+#include "video.h"
+
+#if CONFIG_DXVA2
+#include <d3d9.h>
+#endif
+
+#if CONFIG_D3D11VA
+#include <d3d11.h>
+#endif
+
+
+static int amf_scale_query_formats(AVFilterContext *avctx)
+{
+    const enum AVPixelFormat *output_pix_fmts;
+    static const enum AVPixelFormat input_pix_fmts[] = {
+        AV_PIX_FMT_NV12,
+        AV_PIX_FMT_P010,
+        AV_PIX_FMT_BGRA,
+        AV_PIX_FMT_RGBA,
+        AV_PIX_FMT_AMF,
+        AV_PIX_FMT_RGBAF16,
+        AV_PIX_FMT_NONE,
+    };
+    static const enum AVPixelFormat output_pix_fmts_default[] = {
+        AV_PIX_FMT_NV12,
+        AV_PIX_FMT_P010,
+        AV_PIX_FMT_BGRA,
+        AV_PIX_FMT_RGBA,
+        AV_PIX_FMT_AMF,
+        AV_PIX_FMT_D3D11,
+        AV_PIX_FMT_DXVA2_VLD,
+        AV_PIX_FMT_RGBAF16,
+        AV_PIX_FMT_NONE,
+    };
+    output_pix_fmts = output_pix_fmts_default;
+
+    return amf_setup_input_output_formats(avctx, input_pix_fmts, output_pix_fmts);
+}
+
+static int amf_scale_config_output(AVFilterLink *outlink)
+{
+    AVFilterContext *avctx = outlink->src;
+    AVFilterLink   *inlink = avctx->inputs[0];
+    AMFScaleContext  *ctx = avctx->priv;
+    AVAMFDeviceContextInternal * internal = NULL;
+    AMFSize out_size;
+    int err;
+    AMF_RESULT res;
+    enum AVPixelFormat in_format;
+
+    err = amf_init_scale_config(outlink, &in_format);
+    if (err < 0)
+        return err;
+
+    // HQ scaler should be used for upscaling only
+    if (inlink->w > outlink->w || inlink->h > outlink->h) {
+        av_log(avctx, AV_LOG_ERROR, "AMF HQ scaler should be used for upscaling only.\n");
+        return AVERROR_UNKNOWN;
+    }
+    // FIXME: add checks whether we have HW context
+
+    internal = (AVAMFDeviceContextInternal * )ctx->amf_device_ctx_internal->data;
+    res = internal->factory->pVtbl->CreateComponent(internal->factory, internal->context, AMFHQScaler, &ctx->scaler);
+    AMF_RETURN_IF_FALSE(ctx, res == AMF_OK, AVERROR_FILTER_NOT_FOUND, "CreateComponent(%ls) failed with error %d\n", AMFHQScaler, res);
+
+    out_size.width = outlink->w;
+    out_size.height = outlink->h;
+    AMF_ASSIGN_PROPERTY_SIZE(res, ctx->scaler, AMF_HQ_SCALER_OUTPUT_SIZE, out_size);
+    AMF_RETURN_IF_FALSE(avctx, res == AMF_OK, AVERROR_UNKNOWN, "AMFHQScaler-SetProperty() failed with error %d\n", res);
+
+    if (ctx->algorithm != -1) {
+        AMF_ASSIGN_PROPERTY_INT64(res, ctx->scaler, AMF_HQ_SCALER_ALGORITHM, ctx->algorithm);
+    }
+    if (ctx->sharpness != -1) {
+        AMF_ASSIGN_PROPERTY_DOUBLE(res, ctx->scaler, AMF_HQ_SCALER_SHARPNESS, ctx->sharpness);
+    }
+    AMF_ASSIGN_PROPERTY_BOOL(res, ctx->scaler, AMF_HQ_SCALER_FILL, ctx->fill);
+    AMF_ASSIGN_PROPERTY_BOOL(res, ctx->scaler, AMF_HQ_SCALER_KEEP_ASPECT_RATIO, ctx->keep_ratio);
+    // Setup default options to skip color conversion
+    ctx->color_profile = AMF_VIDEO_CONVERTER_COLOR_PROFILE_UNKNOWN;
+    ctx->color_range = AMF_COLOR_RANGE_UNDEFINED;
+    ctx->primaries = AMF_COLOR_PRIMARIES_UNDEFINED;
+    ctx->trc = AMF_COLOR_TRANSFER_CHARACTERISTIC_UNDEFINED;
+
+    res = ctx->scaler->pVtbl->Init(ctx->scaler, av_amf_av_to_amf_format(in_format), inlink->w, inlink->h);
+    AMF_RETURN_IF_FALSE(avctx, res == AMF_OK, AVERROR_UNKNOWN, "AMFHQScaler-Init() failed with error %d\n", res);
+
+    return 0;
+}
+
+#define OFFSET(x) offsetof(AMFScaleContext, x)
+#define FLAGS AV_OPT_FLAG_VIDEO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
+static const AVOption scale_amf_hq_options[] = {
+    { "w",              "Output video width",   OFFSET(w_expr),     AV_OPT_TYPE_STRING, { .str = "iw"   }, .flags = FLAGS },
+    { "h",              "Output video height",  OFFSET(h_expr),     AV_OPT_TYPE_STRING, { .str = "ih"   }, .flags = FLAGS },
+
+    { "format",         "Output pixel format",  OFFSET(format_str), AV_OPT_TYPE_STRING, { .str = "same" }, .flags = FLAGS },
+    { "sharpness",      "Sharpness",            OFFSET(sharpness),  AV_OPT_TYPE_FLOAT,  { .dbl = -1 }, -1, 2., FLAGS, "sharpness" },
+    { "keep-ratio",     "Keep aspect ratio",    OFFSET(keep_ratio), AV_OPT_TYPE_BOOL, { .i64 = 0  },  0, 1, FLAGS, "keep_ration" },
+    { "fill",           "Fill",                 OFFSET(fill),       AV_OPT_TYPE_BOOL, { .i64 = 0  },  0, 1, FLAGS, "fill" },
+
+    { "algorithm",      "Scaling algorithm",    OFFSET(algorithm),      AV_OPT_TYPE_INT,   { .i64 = -1 }, -1, AMF_HQ_SCALER_ALGORITHM_VIDEOSR1_1, FLAGS, "algorithm" },
+    { "bilinear",       "Bilinear",             0,  AV_OPT_TYPE_CONST, { .i64 = AMF_HQ_SCALER_ALGORITHM_BILINEAR }, 0, 0, FLAGS, "algorithm" },
+    { "bicubic",        "Bicubic",              0,  AV_OPT_TYPE_CONST, { .i64 = AMF_HQ_SCALER_ALGORITHM_BICUBIC }, 0, 0, FLAGS, "algorithm" },
+    { "sr1-0",          "Video SR1.0",          0,  AV_OPT_TYPE_CONST, { .i64 = AMF_HQ_SCALER_ALGORITHM_VIDEOSR1_0 }, 0, 0, FLAGS, "algorithm" },
+    { "point",          "Point",                0,  AV_OPT_TYPE_CONST, { .i64 = AMF_HQ_SCALER_ALGORITHM_POINT }, 0, 0, FLAGS, "algorithm" },
+    { "sr1-1",          "Video SR1.1",          0,  AV_OPT_TYPE_CONST, { .i64 = AMF_HQ_SCALER_ALGORITHM_VIDEOSR1_1 }, 0, 0, FLAGS, "algorithm" },
+
+    { NULL },
+};
+
+
+AVFILTER_DEFINE_CLASS(scale_amf_hq);
+
+static const AVFilterPad amf_scale_inputs[] = {
+    {
+        .name         = "default",
+        .type         = AVMEDIA_TYPE_VIDEO,
+        .filter_frame = amf_scale_filter_frame,
+    }
+};
+
+static const AVFilterPad amf_scale_outputs[] = {
+    {
+        .name         = "default",
+        .type         = AVMEDIA_TYPE_VIDEO,
+        .config_props = amf_scale_config_output,
+    }
+};
+
+AVFilter ff_vf_scale_amf_hq = {
+    .name      = "scale_amf_hq",
+    .description = NULL_IF_CONFIG_SMALL("AMF HQ video upscaling"),
+
+    .init          = amf_scale_init,
+    .uninit        = amf_scale_uninit,
+    FILTER_QUERY_FUNC(&amf_scale_query_formats),
+
+    .priv_size = sizeof(AMFScaleContext),
+    .priv_class = &scale_amf_hq_class,
+
+    FILTER_INPUTS(amf_scale_inputs),
+    FILTER_OUTPUTS(amf_scale_outputs),
+
+    FILTER_SINGLE_PIXFMT(AV_PIX_FMT_AMF),
+
+    .flags_internal = FF_FILTER_FLAG_HWFRAME_AWARE,
+    .flags          = AVFILTER_FLAG_HWDEVICE,
+};
\ No newline at end of file