diff --git a/DOCS/man/options.rst b/DOCS/man/options.rst
index 3c2a404189..83ef901eed 100644
--- a/DOCS/man/options.rst
+++ b/DOCS/man/options.rst
@@ -779,6 +779,8 @@ Video
     the given hardware. ``nv12``, the default, works better on modern hardware,
     while ``uyvy422`` appears to be better for old hardware. ``yuv420p`` also
     works.
+    Since mpv 0.25.0, ``no`` is an accepted value, which lets the decoder pick
+    the format on newer FFmpeg versions (will use ``nv12`` on older versions).
 
 ``--panscan=<0.0-1.0>``
     Enables pan-and-scan functionality (cropping the sides of e.g. a 16:9
diff --git a/options/options.c b/options/options.c
index 7723ddec50..012707cb2c 100644
--- a/options/options.c
+++ b/options/options.c
@@ -448,7 +448,7 @@ const m_option_t mp_opts[] = {
     OPT_CHOICE_C("hwdec", hwdec_api, 0, mp_hwdec_names),
     OPT_STRING("hwdec-codecs", hwdec_codecs, 0),
 #if HAVE_VIDEOTOOLBOX_HWACCEL
-    OPT_IMAGEFORMAT("videotoolbox-format", videotoolbox_format, 0),
+    OPT_IMAGEFORMAT("videotoolbox-format", videotoolbox_format, 0, .min = -1),
 #endif
 
     // -1 means auto aspect (prefer container size until aspect change)
diff --git a/video/decode/hw_videotoolbox.c b/video/decode/hw_videotoolbox.c
index 8d8c165a84..b343b1de0e 100644
--- a/video/decode/hw_videotoolbox.c
+++ b/video/decode/hw_videotoolbox.c
@@ -105,8 +105,14 @@ static int init_decoder(struct lavc_ctx *ctx, int w, int h)
     if (!vtctx)
         return -1;
 
-    vtctx->cv_pix_fmt_type =
-        mp_imgfmt_to_cvpixelformat(ctx->opts->videotoolbox_format);
+    int imgfmt = ctx->opts->videotoolbox_format;
+#if LIBAVCODEC_VERSION_INT < AV_VERSION_INT(57, 81, 103)
+    if (!imgfmt)
+        imgfmt = IMGFMT_NV12;
+#endif
+    vtctx->cv_pix_fmt_type = mp_imgfmt_to_cvpixelformat(imgfmt);
+    MP_VERBOSE(ctx, "Requesting cv_pix_fmt_type=0x%x\n",
+               (unsigned)vtctx->cv_pix_fmt_type);
 
     int err = av_videotoolbox_default_init2(ctx->avctx, vtctx);
     if (err < 0) {