More correctly calculate bitmap subtitle scaling (#2670).2670-again

This was partially fixed before in
6ac468554c7fea0dfaefde85fb6cdd0fceaf5cad

The last try accounted for cropping, but not for cases where the
source video (after crop) does not precisely fit the DCP container.
In those cases the x scale for the subtitles could be different
to the y scale, squashing or stretching them.

6 files changed, 38 insertions, 30 deletions

diff --git a/src/lib/bitmap_text.h b/src/lib/bitmap_text.h
index 04cb61e3e..50f3f507a 100644
--- a/src/lib/bitmap_text.h
+++ b/src/lib/bitmap_text.h
@@ -23,6 +23,7 @@
 
 
 #include "rect.h"
+#include <dcp/types.h>
 #include <memory>
 
 
@@ -31,11 +32,14 @@ class Image;
 class BitmapText
 {
 public:
-	BitmapText(std::shared_ptr<const Image> image_, dcpomatic::Rect<double> rectangle_)
-		: image(image_)
+	BitmapText(dcp::Size parent_size_, std::shared_ptr<const Image> image_, dcpomatic::Rect<double> rectangle_)
+		: parent_size(parent_size_)
+		, image(image_)
 		, rectangle(rectangle_)
 	{}
 
+	/** Size of the image that this subtitle has come from, with any crop already applied */
+	dcp::Size parent_size;
 	std::shared_ptr<const Image> image;
 	/** Area that the subtitle covers on its corresponding video, expressed in
 	 *  proportions of the image size; e.g. rectangle.x = 0.5 would mean that
diff --git a/src/lib/content_text.h b/src/lib/content_text.h
index 8a7f007ef..d04383c8b 100644
--- a/src/lib/content_text.h
+++ b/src/lib/content_text.h
@@ -57,9 +57,9 @@ public:
 		: ContentText(from)
 	{}
 
-	ContentBitmapText(dcpomatic::ContentTime from, std::shared_ptr<const Image> image, dcpomatic::Rect<double> rect)
+	ContentBitmapText(dcpomatic::ContentTime from, dcp::Size parent_size, std::shared_ptr<const Image> image, dcpomatic::Rect<double> rect)
 		: ContentText(from)
-		, subs{ {image, rect} }
+		, subs{ { parent_size, image, rect } }
 	{}
 
 	/* Our texts, with their rectangles unmodified by any offsets or scales that the content specifies */
diff --git a/src/lib/ffmpeg_decoder.cc b/src/lib/ffmpeg_decoder.cc
index 7f7a07863..ab3f0b5be 100644
--- a/src/lib/ffmpeg_decoder.cc
+++ b/src/lib/ffmpeg_decoder.cc
@@ -770,39 +770,41 @@ FFmpegDecoder::process_bitmap_subtitle (AVSubtitleRect const * rect)
 		out_p += image->stride()[0];
 	}
 
-	int target_width = subtitle_codec_context()->width;
-	if (target_width == 0 && video_codec_context()) {
-		/* subtitle_codec_context()->width == 0 has been seen in the wild but I don't
-		   know if it's supposed to mean something from FFmpeg's point of view.
-		*/
-		target_width = video_codec_context()->width;
+	optional<dcp::Size> video_size;
+	if (_ffmpeg_content->video) {
+		video_size = _ffmpeg_content->video->size();
 	}
-	int target_height = subtitle_codec_context()->height;
-	if (target_height == 0 && video_codec_context()) {
-		target_height = video_codec_context()->height;
+
+	dcp::Size target_size = { subtitle_codec_context()->width, subtitle_codec_context()->height };
+	if (target_size.width == 0 || target_size.height == 0 || (video_size && *video_size == target_size)) {
+		/* Either the subtitle codec has no specified size, or it's the same as the video.
+		 * In either case we'll use the target size once it has been cropped etc. as we
+		 * assume that whatever happens to the video should also be done to the subtitles.
+		 */
+		if (auto s = ffmpeg_content()->video->scaled_size(film()->frame_size())) {
+			target_size = *s;
+		}
 	}
 
 	int x_offset = 0;
 	int y_offset = 0;
 	if (_ffmpeg_content->video && _ffmpeg_content->video->use()) {
 		auto const crop = _ffmpeg_content->video->actual_crop();
-		target_width -= crop.left + crop.right;
-		target_height -= crop.top + crop.bottom;
 		x_offset = -crop.left;
 		y_offset = -crop.top;
 	}
 
-	DCPOMATIC_ASSERT(target_width > 0);
-	DCPOMATIC_ASSERT(target_height > 0);
+	DCPOMATIC_ASSERT(target_size.width > 0);
+	DCPOMATIC_ASSERT(target_size.height > 0);
 
 	dcpomatic::Rect<double> const scaled_rect (
-		static_cast<double>(rect->x + x_offset) / target_width,
-		static_cast<double>(rect->y + y_offset) / target_height,
-		static_cast<double>(rect->w) / target_width,
-		static_cast<double>(rect->h) / target_height
+		static_cast<double>(rect->x + x_offset) / target_size.width,
+		static_cast<double>(rect->y + y_offset) / target_size.height,
+		static_cast<double>(rect->w) / target_size.width,
+		static_cast<double>(rect->h) / target_size.height
 		);
 
-	return { image, scaled_rect };
+	return { target_size, image, scaled_rect };
 }
 
 
diff --git a/src/lib/player.cc b/src/lib/player.cc
index c03cb97a5..1b5eca65c 100644
--- a/src/lib/player.cc
+++ b/src/lib/player.cc
@@ -1233,7 +1233,8 @@ Player::bitmap_text_start (weak_ptr<Piece> weak_piece, weak_ptr<const TextConten
 
 	auto piece = weak_piece.lock ();
 	auto content = weak_content.lock ();
-	if (!piece || !content) {
+	auto film = _film.lock();
+	if (!piece || !content || !film) {
 		return;
 	}
 
@@ -1254,15 +1255,16 @@ Player::bitmap_text_start (weak_ptr<Piece> weak_piece, weak_ptr<const TextConten
 
 		auto image = sub.image;
 
-		/* We will scale the subtitle up to fit _video_container_size */
-		int const width = sub.rectangle.width * _video_container_size.load().width;
-		int const height = sub.rectangle.height * _video_container_size.load().height;
+		auto const inter_size = scale_for_display(sub.parent_size, _video_container_size.load(), film->frame_size(), {});
+
+		int const width = sub.rectangle.width * inter_size.width;
+		int const height = sub.rectangle.height * inter_size.height;
 		if (width == 0 || height == 0) {
 			return;
 		}
 
 		dcp::Size scaled_size (width, height);
-		ps.bitmap.push_back (BitmapText(image->scale(scaled_size, dcp::YUVToRGB::REC601, image->pixel_format(), Image::Alignment::PADDED, _fast), sub.rectangle));
+		ps.bitmap.push_back(BitmapText(sub.parent_size, image->scale(scaled_size, dcp::YUVToRGB::REC601, image->pixel_format(), Image::Alignment::PADDED, _fast), sub.rectangle));
 	}
 
 	DCPTime from(content_time_to_dcp(piece, subtitle.from()));
diff --git a/src/lib/text_decoder.cc b/src/lib/text_decoder.cc
index 75fa33605..894145c35 100644
--- a/src/lib/text_decoder.cc
+++ b/src/lib/text_decoder.cc
@@ -352,9 +352,9 @@ TextDecoder::emit_plain (ContentTimePeriod period, sub::Subtitle const& subtitle
  *  of the video frame)
  */
 void
-TextDecoder::emit_bitmap (ContentTimePeriod period, shared_ptr<const Image> image, dcpomatic::Rect<double> rect)
+TextDecoder::emit_bitmap(ContentTimePeriod period, dcp::Size parent_size, shared_ptr<const Image> image, dcpomatic::Rect<double> rect)
 {
-	emit_bitmap_start ({ period.from, image, rect });
+	emit_bitmap_start ({ period.from, parent_size, image, rect });
 	emit_stop (period.to);
 }
 
diff --git a/src/lib/text_decoder.h b/src/lib/text_decoder.h
index 1a7632fd8..47c1be399 100644
--- a/src/lib/text_decoder.h
+++ b/src/lib/text_decoder.h
@@ -51,7 +51,7 @@ public:
 	}
 
 	void emit_bitmap_start (ContentBitmapText const& bitmap);
-	void emit_bitmap (dcpomatic::ContentTimePeriod period, std::shared_ptr<const Image> image, dcpomatic::Rect<double> rect);
+	void emit_bitmap(dcpomatic::ContentTimePeriod period, dcp::Size parent_size, std::shared_ptr<const Image> image, dcpomatic::Rect<double> rect);
 	void emit_plain_start(dcpomatic::ContentTime from, std::vector<dcp::SubtitleString> s, dcp::SubtitleStandard valign_standard);
 	void emit_plain_start (dcpomatic::ContentTime from, sub::Subtitle const & subtitle);
 	void emit_plain(dcpomatic::ContentTimePeriod period, std::vector<dcp::SubtitleString> s, dcp::SubtitleStandard valign_standard);