diff --git a/debian/patches/0021-add-fixes-for-qsv-vpp-filters.patch b/debian/patches/0021-add-fixes-for-qsv-vpp-filters.patch
index 2db4a730d87..422c2b2e2aa 100644
--- a/debian/patches/0021-add-fixes-for-qsv-vpp-filters.patch
+++ b/debian/patches/0021-add-fixes-for-qsv-vpp-filters.patch
@@ -25,11 +25,32 @@ Index: FFmpeg/libavfilter/qsvvpp.c
      else if (qsv_frame->frame->repeat_pict == 2)
          qsv_frame->surface.Info.PicStruct |= MFX_PICSTRUCT_FRAME_DOUBLING;
      else if (qsv_frame->frame->repeat_pict == 4)
+@@ -1024,6 +1028,20 @@ int ff_qsvvpp_filter_frame(QSVVPPContext
+         out_frame->frame->pts = av_rescale_q(out_frame->surface.Data.TimeStamp,
+                                              default_tb, outlink->time_base);
+ 
++        /* Copy the color side data */
++        if (in_frame->frame->color_primaries != -1)
++            out_frame->frame->color_primaries = in_frame->frame->color_primaries;
++        if (in_frame->frame->color_trc != -1)
++            out_frame->frame->color_trc = in_frame->frame->color_trc;
++        if (in_frame->frame->colorspace != -1)
++            out_frame->frame->colorspace = in_frame->frame->colorspace;
++        if (in_frame->frame->color_range != -1)
++            out_frame->frame->color_range = in_frame->frame->color_range;
++
++        ret = av_frame_copy_side_data(out_frame->frame, in_frame->frame, 0);
++        if (ret < 0)
++            return ret;
++
+         out_frame->queued++;
+         aframe = (QSVAsyncFrame){ sync, out_frame };
+         av_fifo_write(s->async_fifo, &aframe, 1);
 Index: FFmpeg/libavfilter/vf_overlay_qsv.c
 ===================================================================
 --- FFmpeg.orig/libavfilter/vf_overlay_qsv.c
 +++ FFmpeg/libavfilter/vf_overlay_qsv.c
-@@ -228,40 +228,47 @@ static int config_overlay_input(AVFilter
+@@ -228,40 +228,51 @@ static int config_overlay_input(AVFilter
  
  static int process_frame(FFFrameSync *fs)
  {
@@ -69,6 +90,10 @@ Index: FFmpeg/libavfilter/vf_overlay_qsv.c
 +    if (ret < 0 && ret != AVERROR(EAGAIN))
 +        return ret;
 +
++    /* remove all side data of the overlay frame*/
++    if (overlay)
++        av_frame_remove_all_side_data(overlay);
++
 +    /* composite overlay frame */
 +    /* or overwrite main frame again if the overlay frame isn't ready yet */
 +    return ff_qsvvpp_filter_frame(qsv, overlay ? in1 : in0, overlay ? overlay : main);
@@ -102,7 +127,7 @@ Index: FFmpeg/libavfilter/vf_overlay_qsv.c
  
      return ff_framesync_configure(&s->fs);
  }
-@@ -282,12 +289,6 @@ static int config_output(AVFilterLink *o
+@@ -282,12 +293,6 @@ static int config_output(AVFilterLink *o
          return AVERROR(EINVAL);
      } else if (in0->format == AV_PIX_FMT_QSV) {
          AVHWFramesContext *hw_frame0 = (AVHWFramesContext *)in0->hw_frames_ctx->data;
@@ -115,7 +140,7 @@ Index: FFmpeg/libavfilter/vf_overlay_qsv.c
          vpp->qsv_param.out_sw_format = hw_frame0->sw_format;
      }
  
-@@ -369,6 +370,7 @@ static int overlay_qsv_query_formats(AVF
+@@ -369,6 +374,7 @@ static int overlay_qsv_query_formats(AVF
      static const enum AVPixelFormat main_in_fmts[] = {
          AV_PIX_FMT_YUV420P,
          AV_PIX_FMT_NV12,
@@ -123,7 +148,7 @@ Index: FFmpeg/libavfilter/vf_overlay_qsv.c
          AV_PIX_FMT_YUYV422,
          AV_PIX_FMT_RGB32,
          AV_PIX_FMT_QSV,
-@@ -376,6 +378,7 @@ static int overlay_qsv_query_formats(AVF
+@@ -376,6 +382,7 @@ static int overlay_qsv_query_formats(AVF
      };
      static const enum AVPixelFormat out_pix_fmts[] = {
          AV_PIX_FMT_NV12,
@@ -166,7 +191,34 @@ Index: FFmpeg/libavfilter/vf_vpp_qsv.c
  static int vpp_set_frame_ext_params(AVFilterContext *ctx, const AVFrame *in, AVFrame *out,  QSVVPPFrameParam *fp)
  {
  #if QSV_ONEVPL
-@@ -494,9 +518,9 @@ static int vpp_set_frame_ext_params(AVFi
+@@ -461,14 +485,19 @@ static int vpp_set_frame_ext_params(AVFi
+ 
+     memset(&clli_conf, 0, sizeof(mfxExtContentLightLevelInfo));
+     sd = av_frame_get_side_data(in, AV_FRAME_DATA_CONTENT_LIGHT_LEVEL);
+-    if (vpp->tonemap && sd) {
+-        AVContentLightMetadata *clm = (AVContentLightMetadata *)sd->data;
++    if (vpp->tonemap) {
++        AVContentLightMetadata *clm = sd ? (AVContentLightMetadata *)sd->data : NULL;
+ 
+-        clli_conf.Header.BufferId         = MFX_EXTBUFF_CONTENT_LIGHT_LEVEL_INFO;
+-        clli_conf.Header.BufferSz         = sizeof(mfxExtContentLightLevelInfo);
+-        clli_conf.MaxContentLightLevel    = FFMIN(clm->MaxCLL,  65535);
+-        clli_conf.MaxPicAverageLightLevel = FFMIN(clm->MaxFALL, 65535);
+-        tm = 1;
++        // Dumped from VP HAL, VPL requires at least one type of the metadata to trigger tone-mapping
++        #define HAL_HDR_DEFAULT_MAXCLL 4000
++        #define HAL_HDR_DEFAULT_MAXFALL 400
++        if (clm || !tm) {
++            clli_conf.Header.BufferId         = MFX_EXTBUFF_CONTENT_LIGHT_LEVEL_INFO;
++            clli_conf.Header.BufferSz         = sizeof(mfxExtContentLightLevelInfo);
++            clli_conf.MaxContentLightLevel    = FFMIN(clm ? clm->MaxCLL  : HAL_HDR_DEFAULT_MAXCLL,  65535);
++            clli_conf.MaxPicAverageLightLevel = FFMIN(clm ? clm->MaxFALL : HAL_HDR_DEFAULT_MAXFALL, 65535);
++            tm = 1;
++        }
+     }
+ 
+     if (tm) {
+@@ -494,9 +523,9 @@ static int vpp_set_frame_ext_params(AVFi
      outvsi_conf.Header.BufferId          = MFX_EXTBUFF_VIDEO_SIGNAL_INFO_OUT;
      outvsi_conf.Header.BufferSz          = sizeof(mfxExtVideoSignalInfo);
      outvsi_conf.VideoFullRange           = (out->color_range == AVCOL_RANGE_JPEG);
@@ -179,7 +231,7 @@ Index: FFmpeg/libavfilter/vf_vpp_qsv.c
      outvsi_conf.ColourDescriptionPresent = 1;
  
      if (memcmp(&vpp->invsi_conf, &invsi_conf, sizeof(mfxExtVideoSignalInfo)) ||
-@@ -686,12 +710,24 @@ static int config_output(AVFilterLink *o
+@@ -686,12 +715,24 @@ static int config_output(AVFilterLink *o
  
      if (inlink->w != outlink->w || inlink->h != outlink->h || in_format != vpp->out_format) {
          if (QSV_RUNTIME_VERSION_ATLEAST(mfx_version, 1, 19)) {
@@ -208,7 +260,7 @@ Index: FFmpeg/libavfilter/vf_vpp_qsv.c
  
              INIT_MFX_EXTBUF(scale_conf, MFX_EXTBUFF_VPP_SCALING);
              SET_MFX_PARAM_FIELD(scale_conf, ScalingMode, mode);
-@@ -880,19 +916,13 @@ static const AVOption vpp_options[] = {
+@@ -880,19 +921,13 @@ static const AVOption vpp_options[] = {
      { "height", "Output video height(0=input video height, -1=keep input video aspect)", OFFSET(oh), AV_OPT_TYPE_STRING, { .str="w*ch/cw" }, 0, 255, .flags = FLAGS },
      { "format", "Output pixel format", OFFSET(output_format_str), AV_OPT_TYPE_STRING, { .str = "same" }, .flags = FLAGS },
      { "async_depth", "Internal parallelization depth, the higher the value the higher the latency.", OFFSET(qsv.async_depth), AV_OPT_TYPE_INT, { .i64 = 4 }, 0, INT_MAX, .flags = FLAGS },
@@ -229,7 +281,7 @@ Index: FFmpeg/libavfilter/vf_vpp_qsv.c
  
      { "rate", "Generate output at frame rate or field rate, available only for deinterlace mode",
        OFFSET(field_rate), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, FLAGS, .unit = "rate" },
-@@ -923,8 +953,9 @@ static const AVOption vpp_options[] = {
+@@ -923,8 +958,9 @@ static const AVOption vpp_options[] = {
      { "out_color_transfer", "Output color transfer characteristics",
        OFFSET(color_transfer_str),  AV_OPT_TYPE_STRING, { .str = NULL }, .flags = FLAGS },
  
@@ -240,7 +292,7 @@ Index: FFmpeg/libavfilter/vf_vpp_qsv.c
      { NULL }
  };
  
-@@ -978,19 +1009,14 @@ static const AVOption qsvscale_options[]
+@@ -978,19 +1014,14 @@ static const AVOption qsvscale_options[]
      { "h",      "Output video height(0=input video height, -1=keep input video aspect)", OFFSET(oh), AV_OPT_TYPE_STRING, { .str = "ih"   }, .flags = FLAGS },
      { "format", "Output pixel format", OFFSET(output_format_str), AV_OPT_TYPE_STRING, { .str = "same" }, .flags = FLAGS },
  
@@ -262,7 +314,7 @@ Index: FFmpeg/libavfilter/vf_vpp_qsv.c
      { NULL },
  };
  
-@@ -1015,6 +1041,7 @@ static const AVOption qsvdeint_options[]
+@@ -1015,6 +1046,7 @@ static const AVOption qsvdeint_options[]
      { "bob",   "bob algorithm",                  0, AV_OPT_TYPE_CONST,      {.i64 = MFX_DEINTERLACING_BOB}, MFX_DEINTERLACING_BOB, MFX_DEINTERLACING_ADVANCED, FLAGS, .unit = "mode"},
      { "advanced", "Motion adaptive algorithm",   0, AV_OPT_TYPE_CONST, {.i64 = MFX_DEINTERLACING_ADVANCED}, MFX_DEINTERLACING_BOB, MFX_DEINTERLACING_ADVANCED, FLAGS, .unit = "mode"},
  
@@ -270,3 +322,148 @@ Index: FFmpeg/libavfilter/vf_vpp_qsv.c
      { NULL },
  };
  
+Index: FFmpeg/libavutil/frame.c
+===================================================================
+--- FFmpeg.orig/libavutil/frame.c
++++ FFmpeg/libavutil/frame.c
+@@ -98,6 +98,18 @@ static void remove_side_data(AVFrameSide
+     }
+ }
+ 
++static void remove_all_side_data(AVFrameSideData ***sd, int *nb_side_data)
++{
++    for (int i = *nb_side_data - 1; i >= 0; i--) {
++        AVFrameSideData *entry = ((*sd)[i]);
++
++        free_side_data(&entry);
++
++        ((*sd)[i]) = ((*sd)[*nb_side_data - 1]);
++        (*nb_side_data)--;
++    }
++}
++
+ static void remove_side_data_by_entry(AVFrameSideData ***sd, int *nb_sd,
+                                       const AVFrameSideData *target)
+ {
+@@ -269,6 +281,36 @@ int av_frame_get_buffer(AVFrame *frame,
+     return AVERROR(EINVAL);
+ }
+ 
++int av_frame_copy_side_data(AVFrame* dst, const AVFrame* src, int flags)
++{
++    for (unsigned i = 0; i < src->nb_side_data; i++) {
++        const AVFrameSideData *sd_src = src->side_data[i];
++        AVFrameSideData *sd_dst;
++        if (   sd_src->type == AV_FRAME_DATA_PANSCAN
++            && (src->width != dst->width || src->height != dst->height))
++            continue;
++        if (flags & AV_FRAME_COPY_PROPS_FORCECOPY) {
++            sd_dst = av_frame_new_side_data(dst, sd_src->type,
++                                            sd_src->size);
++            if (!sd_dst) {
++                frame_side_data_wipe(dst);
++                return AVERROR(ENOMEM);
++            }
++            memcpy(sd_dst->data, sd_src->data, sd_src->size);
++        } else {
++            AVBufferRef *ref = av_buffer_ref(sd_src->buf);
++            sd_dst = av_frame_new_side_data_from_buf(dst, sd_src->type, ref);
++            if (!sd_dst) {
++                av_buffer_unref(&ref);
++                frame_side_data_wipe(dst);
++                return AVERROR(ENOMEM);
++            }
++        }
++        av_dict_copy(&sd_dst->metadata, sd_src->metadata, 0);
++    }
++    return 0;
++}
++
+ static int frame_copy_props(AVFrame *dst, const AVFrame *src, int force_copy)
+ {
+     int ret;
+@@ -320,31 +362,9 @@ FF_ENABLE_DEPRECATION_WARNINGS
+ 
+     av_dict_copy(&dst->metadata, src->metadata, 0);
+ 
+-    for (int i = 0; i < src->nb_side_data; i++) {
+-        const AVFrameSideData *sd_src = src->side_data[i];
+-        AVFrameSideData *sd_dst;
+-        if (   sd_src->type == AV_FRAME_DATA_PANSCAN
+-            && (src->width != dst->width || src->height != dst->height))
+-            continue;
+-        if (force_copy) {
+-            sd_dst = av_frame_new_side_data(dst, sd_src->type,
+-                                            sd_src->size);
+-            if (!sd_dst) {
+-                frame_side_data_wipe(dst);
+-                return AVERROR(ENOMEM);
+-            }
+-            memcpy(sd_dst->data, sd_src->data, sd_src->size);
+-        } else {
+-            AVBufferRef *ref = av_buffer_ref(sd_src->buf);
+-            sd_dst = av_frame_new_side_data_from_buf(dst, sd_src->type, ref);
+-            if (!sd_dst) {
+-                av_buffer_unref(&ref);
+-                frame_side_data_wipe(dst);
+-                return AVERROR(ENOMEM);
+-            }
+-        }
+-        av_dict_copy(&sd_dst->metadata, sd_src->metadata, 0);
+-    }
++    if (ret = av_frame_copy_side_data(dst, src,
++        force_copy ? AV_FRAME_COPY_PROPS_FORCECOPY : 0) < 0)
++        return ret;
+ 
+     ret = av_buffer_replace(&dst->opaque_ref, src->opaque_ref);
+     ret |= av_buffer_replace(&dst->private_ref, src->private_ref);
+@@ -906,6 +926,11 @@ void av_frame_remove_side_data(AVFrame *
+     remove_side_data(&frame->side_data, &frame->nb_side_data, type);
+ }
+ 
++void av_frame_remove_all_side_data(AVFrame *frame)
++{
++    remove_all_side_data(&frame->side_data, &frame->nb_side_data);
++}
++
+ const char *av_frame_side_data_name(enum AVFrameSideDataType type)
+ {
+     switch(type) {
+Index: FFmpeg/libavutil/frame.h
+===================================================================
+--- FFmpeg.orig/libavutil/frame.h
++++ FFmpeg/libavutil/frame.h
+@@ -901,6 +901,21 @@ int av_frame_copy(AVFrame *dst, const AV
+  */
+ int av_frame_copy_props(AVFrame *dst, const AVFrame *src);
+ 
++/** Copy actual data buffers instead of references.
++ */
++#define AV_FRAME_COPY_PROPS_FORCECOPY  1
++
++/**
++ * Copy only side-data from src to dst.
++ *
++ * @param dst a frame to which the side data should be copied.
++ * @param src a frame from which to copy the side data.
++ * @param flags flags of type AV_FRAME_COPY_PROPS_*, controlling copy behavior.
++ *
++ * @return >= 0 on success, a negative AVERROR on error.
++ */
++int av_frame_copy_side_data(AVFrame* dst, const AVFrame* src, int flags);
++
+ /**
+  * Get the buffer reference a given data plane is stored in.
+  *
+@@ -953,6 +968,11 @@ AVFrameSideData *av_frame_get_side_data(
+  */
+ void av_frame_remove_side_data(AVFrame *frame, enum AVFrameSideDataType type);
+ 
++/**
++ * Remove and free all side data in this frame.
++ */
++void av_frame_remove_all_side_data(AVFrame *frame);
++
+ 
+ /**
+  * Flags for frame cropping.