diff --git a/video/out/aspect.c b/video/out/aspect.c
index 1439f5a5db..e92c5e1b5d 100644
--- a/video/out/aspect.c
+++ b/video/out/aspect.c
@@ -27,18 +27,18 @@
 #include "sub/osd.h"
 
 static void aspect_calc_panscan(struct mp_log *log, struct mp_vo_opts *opts,
-                                struct mp_image_params *video,
+                                int w, int h, int d_w, int d_h,
                                 int window_w, int window_h, double monitor_par,
                                 int *out_w, int *out_h)
 {
     mp_dbg(log, "aspect(0) fitin: %dx%d monitor_par: %.2f\n",
            window_w, window_h, monitor_par);
     int fwidth = window_w;
-    int fheight = (float)window_w / video->d_w * video->d_h / monitor_par;
+    int fheight = (float)window_w / d_w * d_h / monitor_par;
     mp_dbg(log, "aspect(1) wh: %dx%d (org: %dx%d)\n",
-           fwidth, fheight, video->d_w, video->d_h);
-    if (fheight > window_h || fheight < video->h) {
-        int tmpw = (float)window_h / video->d_h * video->d_w * monitor_par;
+           fwidth, fheight, d_w, d_h);
+    if (fheight > window_h || fheight < h) {
+        int tmpw = (float)window_h / d_h * d_w * monitor_par;
         if (tmpw <= window_w) {
             fheight = window_h;
             fwidth = tmpw;
@@ -47,7 +47,7 @@ static void aspect_calc_panscan(struct mp_log *log, struct mp_vo_opts *opts,
         }
     }
     mp_dbg(log, "aspect(2) wh: %dx%d (org: %dx%d)\n",
-           fwidth, fheight, video->d_w, video->d_h);
+           fwidth, fheight, d_w, d_h);
 
     int vo_panscan_area = window_h - fheight;
     if (!vo_panscan_area)
@@ -127,7 +127,7 @@ static void src_dst_split_scaling(int src_size, int dst_size,
 }
 
 void mp_get_src_dst_rects(struct mp_log *log, struct mp_vo_opts *opts,
-                          struct mp_image_params *video,
+                          int vo_caps, struct mp_image_params *video,
                           int window_w, int window_h, double monitor_par,
                           struct mp_rect *out_src,
                           struct mp_rect *out_dst,
@@ -135,6 +135,12 @@ void mp_get_src_dst_rects(struct mp_log *log, struct mp_vo_opts *opts,
 {
     int src_w = video->w;
     int src_h = video->h;
+    int src_dw = video->d_w;
+    int src_dh = video->d_h;
+    if (video->rotate % 180 == 90 && (vo_caps & VO_CAP_ROTATE90)) {
+        MPSWAP(int, src_w, src_h);
+        MPSWAP(int, src_dw, src_dh);
+    }
     window_w = MPMAX(1, window_w);
     window_h = MPMAX(1, window_h);
     struct mp_rect dst = {0, 0, window_w, window_h};
@@ -146,7 +152,8 @@ void mp_get_src_dst_rects(struct mp_log *log, struct mp_vo_opts *opts,
     };
     if (opts->keepaspect) {
         int scaled_width, scaled_height;
-        aspect_calc_panscan(log, opts, video, window_w, window_h, monitor_par,
+        aspect_calc_panscan(log, opts, src_w, src_h, src_dw, src_dh,
+                            window_w, window_h, monitor_par,
                             &scaled_width, &scaled_height);
         src_dst_split_scaling(src_w, window_w, scaled_width, opts->unscaled,
                               opts->zoom, opts->align_x, opts->pan_x,
diff --git a/video/out/aspect.h b/video/out/aspect.h
index db8aa37040..04d314971e 100644
--- a/video/out/aspect.h
+++ b/video/out/aspect.h
@@ -25,7 +25,7 @@ struct mp_image_params;
 struct mp_rect;
 struct mp_osd_res;
 void mp_get_src_dst_rects(struct mp_log *log, struct mp_vo_opts *opts,
-                          struct mp_image_params *video,
+                          int vo_caps, struct mp_image_params *video,
                           int window_w, int window_h, double monitor_par,
                           struct mp_rect *out_src,
                           struct mp_rect *out_dst,
diff --git a/video/out/vo.c b/video/out/vo.c
index 340817633d..9af1d59c6d 100644
--- a/video/out/vo.c
+++ b/video/out/vo.c
@@ -424,7 +424,11 @@ int vo_reconfig(struct vo *vo, struct mp_image_params *params, int flags)
     int d_height = params->d_h;
 
     if (vo_control(vo, VOCTRL_UPDATE_SCREENINFO, NULL) == VO_TRUE) {
-        determine_window_geometry(vo, params->d_w, params->d_h);
+        int w = params->d_w, h = params->d_h;
+        if ((vo->driver->caps & VO_CAP_ROTATE90) &&
+            params->rotate % 180 == 90)
+            MPSWAP(int, w, h);
+        determine_window_geometry(vo, w, h);
         d_width = vo->dwidth;
         d_height = vo->dheight;
     }
@@ -472,8 +476,9 @@ int lookup_keymap_table(const struct mp_keymap *map, int key) {
 void vo_get_src_dst_rects(struct vo *vo, struct mp_rect *out_src,
                           struct mp_rect *out_dst, struct mp_osd_res *out_osd)
 {
-    mp_get_src_dst_rects(vo->log, vo->opts, vo->params, vo->dwidth, vo->dheight,
-                         vo->monitor_par, out_src, out_dst, out_osd);
+    mp_get_src_dst_rects(vo->log, vo->opts, vo->driver->caps, vo->params,
+                         vo->dwidth, vo->dheight, vo->monitor_par,
+                         out_src, out_dst, out_osd);
 }
 
 // Return the window title the VO should set. Always returns a null terminated
diff --git a/video/out/vo.h b/video/out/vo.h
index 28cb3a00ed..b9887d2119 100644
--- a/video/out/vo.h
+++ b/video/out/vo.h
@@ -130,6 +130,9 @@ struct voctrl_screenshot_args {
 #define VOFLAG_GL_DEBUG         0x40  // Hint to request debug OpenGL context
 #define VOFLAG_ALPHA            0x80  // Hint to request alpha framebuffer
 
+// VO does handle mp_image_params.rotate in 90 degree steps
+#define VO_CAP_ROTATE90 1
+
 struct vo;
 struct osd_state;
 struct mp_image;
@@ -143,6 +146,9 @@ struct vo_driver {
     // Encoding functionality, which can be invoked via --o only.
     bool encode;
 
+    // VO_CAP_* bits
+    int caps;
+
     const char *name;
     const char *description;