psxavenc: fixes

.github/scripts/build.sh

@@ -1,7 +1,7 @@
 #!/bin/bash
 
 ROOT_DIR="$(pwd)"
-FFMPEG_VERSION="6.0"
+FFMPEG_VERSION="7.1"
 NUM_JOBS="4"
 
 if [ $# -eq 1 ]; then

.github/workflows/build.yml

@@ -23,7 +23,7 @@ jobs:
         psxavenc/.github/scripts/build.sh psxavenc-windows x86_64-w64-mingw32 psxavenc/.github/scripts/mingw-cross.txt
 
     - name: Upload Windows build artifacts
-      uses: actions/upload-artifact@v3
+      uses: actions/upload-artifact@v4
       with:
         name: psxavenc-windows
         path: psxavenc-windows.zip
@@ -33,7 +33,7 @@ jobs:
         psxavenc/.github/scripts/build.sh psxavenc-linux
 
     - name: Upload Linux build artifacts
-      uses: actions/upload-artifact@v3
+      uses: actions/upload-artifact@v4
       with:
         name: psxavenc-linux
         path: psxavenc-linux.zip

libpsxav/adpcm.c

@@ -232,23 +232,30 @@ uint32_t psx_audio_xa_get_sector_interleave(psx_audio_xa_settings_t settings) {
 	return interleave;
 }
 
-static void psx_audio_xa_encode_init_sector(uint8_t *buffer, psx_audio_xa_settings_t settings) {
+static inline void psx_audio_xa_sync_subheader_copy(psx_cdrom_sector_mode2_t *buffer) {
+	memcpy(buffer->subheader + 1, buffer->subheader, sizeof(psx_cdrom_sector_xa_subheader_t));
+}
+
+static void psx_audio_xa_encode_init_sector(psx_cdrom_sector_mode2_t *buffer, psx_audio_xa_settings_t settings) {
 	if (settings.format == PSX_AUDIO_XA_FORMAT_XACD) {
-		memset(buffer, 0, 2352);
+		memset(buffer, 0, PSX_CDROM_SECTOR_SIZE);
-		memset(buffer+0x001, 0xFF, 10);
+		memset(buffer->sync + 1, 0xFF, 10);
-		buffer[0x00F] = 0x02;
+		buffer->header.mode = 0x02;
 	} else {
-		memset(buffer + 0x10, 0, 2336);
+		memset(buffer->subheader, 0, PSX_CDROM_SECTOR_SIZE - 16);
 	}
 
-	buffer[0x010] = settings.file_number;
+	buffer->subheader[0].file = settings.file_number;
-	buffer[0x011] = settings.channel_number & 0x1F;
+	buffer->subheader[0].channel = settings.channel_number & PSX_CDROM_SECTOR_XA_CHANNEL_MASK;
-	buffer[0x012] = 0x24 | 0x40;
+	buffer->subheader[0].submode =
-	buffer[0x013] =
+		PSX_CDROM_SECTOR_XA_SUBMODE_AUDIO
-		(settings.stereo ? 1 : 0)
+		| PSX_CDROM_SECTOR_XA_SUBMODE_FORM2
-		| (settings.frequency >= PSX_AUDIO_XA_FREQ_DOUBLE ? 0 : 4)
+		| PSX_CDROM_SECTOR_XA_SUBMODE_RT;
-		| (settings.bits_per_sample >= 8 ? 16 : 0);
+	buffer->subheader[0].coding =
-	memcpy(buffer + 0x014, buffer + 0x010, 4);
+		(settings.stereo ? PSX_CDROM_SECTOR_XA_CODING_STEREO : PSX_CDROM_SECTOR_XA_CODING_MONO)
+		| (settings.frequency >= PSX_AUDIO_XA_FREQ_DOUBLE ? PSX_CDROM_SECTOR_XA_CODING_FREQ_DOUBLE : PSX_CDROM_SECTOR_XA_CODING_FREQ_SINGLE)
+		| (settings.bits_per_sample >= 8 ? PSX_CDROM_SECTOR_XA_CODING_BITS_8 : PSX_CDROM_SECTOR_XA_CODING_BITS_4);
+	psx_audio_xa_sync_subheader_copy(buffer);
 }
 
 int psx_audio_xa_encode(psx_audio_xa_settings_t settings, psx_audio_encoder_state_t *state, int16_t* samples, int sample_count, uint8_t *output) {
@@ -261,8 +268,8 @@ int psx_audio_xa_encode(psx_audio_xa_settings_t settings, psx_audio_encoder_stat
 	if (settings.stereo) { sample_count <<= 1; }
 	
 	for (i = 0, j = 0; i < sample_count || ((j % 18) != 0); i += sample_jump, j++) {
-		uint8_t *sector_data = output + ((j/18) * xa_sector_size) - xa_offset;
+		psx_cdrom_sector_mode2_t *sector_data = (psx_cdrom_sector_mode2_t*) (output + ((j/18) * xa_sector_size) - xa_offset);
-		uint8_t *block_data = sector_data + 0x18 + ((j%18) * 0x80);
+		uint8_t *block_data = sector_data->data + ((j%18) * 0x80);
 
 		if (init_sector) {
 			psx_audio_xa_encode_init_sector(sector_data, settings);
@@ -275,7 +282,7 @@ int psx_audio_xa_encode(psx_audio_xa_settings_t settings, psx_audio_encoder_stat
 		memcpy(block_data + 12, block_data + 8, 4);
 
 		if ((j+1)%18 == 0) {
-			psx_cdrom_calculate_checksums(sector_data, PSX_CDROM_SECTOR_TYPE_MODE2_FORM2);
+			psx_cdrom_calculate_checksums((uint8_t*) sector_data, PSX_CDROM_SECTOR_TYPE_MODE2_FORM2);
 			init_sector = 1;
 		}
 	}
@@ -285,8 +292,9 @@ int psx_audio_xa_encode(psx_audio_xa_settings_t settings, psx_audio_encoder_stat
 
 void psx_audio_xa_encode_finalize(psx_audio_xa_settings_t settings, uint8_t *output, int output_length) {
 	if (output_length >= 2336) {
-		output[output_length - 2352 + 0x12] |= 0x80;
+		psx_cdrom_sector_mode2_t *sector = (psx_cdrom_sector_mode2_t*) &output[output_length - 2352];
-		output[output_length - 2352 + 0x18] |= 0x80;
+		sector->subheader[0].submode |= PSX_CDROM_SECTOR_XA_SUBMODE_EOF;
+		psx_audio_xa_sync_subheader_copy(sector);
 	}
 }
 

libpsxav/libpsxav.h

@@ -79,6 +79,58 @@ void psx_audio_spu_set_flag_at_sample(uint8_t* spu_data, int sample_pos, int fla
 
 #define PSX_CDROM_SECTOR_SIZE 2352
 
+typedef struct {
+	uint8_t minute;
+	uint8_t second;
+	uint8_t sector;
+	uint8_t mode;
+} psx_cdrom_sector_header_t;
+
+typedef struct {
+	uint8_t file;
+	uint8_t channel;
+	uint8_t submode;
+	uint8_t coding;
+} psx_cdrom_sector_xa_subheader_t;
+
+typedef struct {
+	uint8_t sync[12];
+	psx_cdrom_sector_header_t header;
+	uint8_t data[0x920];
+} psx_cdrom_sector_mode1_t;
+
+typedef struct {
+	uint8_t sync[12];
+	psx_cdrom_sector_header_t header;
+	psx_cdrom_sector_xa_subheader_t subheader[2];
+	uint8_t data[0x918];
+} psx_cdrom_sector_mode2_t;
+
+_Static_assert(sizeof(psx_cdrom_sector_mode1_t) == PSX_CDROM_SECTOR_SIZE, "Invalid Mode1 sector size");
+_Static_assert(sizeof(psx_cdrom_sector_mode2_t) == PSX_CDROM_SECTOR_SIZE, "Invalid Mode2 sector size");
+
+#define PSX_CDROM_SECTOR_XA_CHANNEL_MASK 0x1F
+
+#define PSX_CDROM_SECTOR_XA_SUBMODE_EOR     0x01
+#define PSX_CDROM_SECTOR_XA_SUBMODE_VIDEO   0x02
+#define PSX_CDROM_SECTOR_XA_SUBMODE_AUDIO   0x04
+#define PSX_CDROM_SECTOR_XA_SUBMODE_DATA    0x08
+#define PSX_CDROM_SECTOR_XA_SUBMODE_TRIGGER 0x10
+#define PSX_CDROM_SECTOR_XA_SUBMODE_FORM2   0x20
+#define PSX_CDROM_SECTOR_XA_SUBMODE_RT      0x40
+#define PSX_CDROM_SECTOR_XA_SUBMODE_EOF     0x80
+
+#define PSX_CDROM_SECTOR_XA_CODING_MONO         0x00
+#define PSX_CDROM_SECTOR_XA_CODING_STEREO       0x01
+#define PSX_CDROM_SECTOR_XA_CODING_CHANNEL_MASK 0x03
+#define PSX_CDROM_SECTOR_XA_CODING_FREQ_DOUBLE  0x00
+#define PSX_CDROM_SECTOR_XA_CODING_FREQ_SINGLE  0x04
+#define PSX_CDROM_SECTOR_XA_CODING_FREQ_MASK    0x0C
+#define PSX_CDROM_SECTOR_XA_CODING_BITS_4       0x00
+#define PSX_CDROM_SECTOR_XA_CODING_BITS_8       0x10
+#define PSX_CDROM_SECTOR_XA_CODING_BITS_MASK    0x30
+#define PSX_CDROM_SECTOR_XA_CODING_EMPHASIS     0x40
+
 typedef enum {
 	PSX_CDROM_SECTOR_TYPE_MODE1,
 	PSX_CDROM_SECTOR_TYPE_MODE2_FORM1,
@@ -87,4 +139,4 @@ typedef enum {
 
 void psx_cdrom_calculate_checksums(uint8_t *sector, psx_cdrom_sector_type_t type);
 
-#endif /* __LIBPSXAV_H__ */
+#endif /* __LIBPSXAV_H__ */

psxavenc/cdrom.c

@@ -23,23 +23,22 @@ freely, subject to the following restrictions:
 
 #include "common.h"
 
-void init_sector_buffer_video(uint8_t *buffer, settings_t *settings) {
+void init_sector_buffer_video(psx_cdrom_sector_mode2_t *buffer, settings_t *settings) {
-	int offset;
 	if (settings->format == FORMAT_STR2CD) {
-		memset(buffer, 0, 2352);
+		memset(buffer, 0, PSX_CDROM_SECTOR_SIZE);
-		memset(buffer+0x001, 0xFF, 10);
+		memset(buffer->sync + 1, 0xFF, 10);
-		buffer[0x00F] = 0x02;
+		buffer->header.mode = 0x02;
-		offset = 0x10;
 	} else {
-		memset(buffer, 0, 2336);
+		memset(buffer->subheader, 0, PSX_CDROM_SECTOR_SIZE - 16);
-		offset = 0;
 	}
 
-	buffer[offset+0] = settings->file_number;
+	buffer->subheader[0].file = settings->file_number;
-	buffer[offset+1] = settings->channel_number & 0x1F;
+	buffer->subheader[0].channel = settings->channel_number & PSX_CDROM_SECTOR_XA_CHANNEL_MASK;
-	buffer[offset+2] = 0x08 | 0x40;
+	buffer->subheader[0].submode =
-	buffer[offset+3] = 0x00;
+		PSX_CDROM_SECTOR_XA_SUBMODE_DATA
-	memcpy(buffer + offset + 4, buffer + offset, 4);
+		| PSX_CDROM_SECTOR_XA_SUBMODE_RT;
+	buffer->subheader[0].coding = 0;
+	memcpy(buffer->subheader + 1, buffer->subheader, sizeof(psx_cdrom_sector_xa_subheader_t));
 }
 
 void calculate_edc_data(uint8_t *buffer)

psxavenc/common.h

@@ -124,7 +124,7 @@ typedef struct {
 } settings_t;
 
 // cdrom.c
-void init_sector_buffer_video(uint8_t *buffer, settings_t *settings);
+void init_sector_buffer_video(psx_cdrom_sector_mode2_t *buffer, settings_t *settings);
 void calculate_edc_data(uint8_t *buffer);
 
 // decoding.c

psxavenc/decoding.c

@@ -204,22 +204,18 @@ bool open_av_data(const char *filename, settings_t *settings, bool use_audio, bo
 		if (av->scaler == NULL) {
 			return false;
 		}
-#if 0
-		// FIXME: if this is uncommented libswscale may produce completely black
-		// frames for whatever reason...
 		if (sws_setColorspaceDetails(
 			av->scaler,
 			sws_getCoefficients(av->video_codec_context->colorspace),
-			(av->video_codec_context->color_range == AVCOL_RANGE_JPEG),
+			av->video_codec_context->color_range == AVCOL_RANGE_JPEG,
 			sws_getCoefficients(SWS_CS_ITU601),
 			true,
 			0,
-			0,
+			1 << 16,
-			0
+			1 << 16
 		) < 0) {
 			return false;
 		}
-#endif
 		if (settings->swscale_options) {
 			if (av_opt_set_from_string(av->scaler, settings->swscale_options, NULL, "=", ":,") < 0) {
 				return false;

psxavenc/filefmt.c

@@ -306,7 +306,7 @@ void encode_file_str(settings_t *settings, FILE *output) {
 
 		if ((j%interleave) < video_sectors_per_block) {
 			// Video sector
-			init_sector_buffer_video(buffer, settings);
+			init_sector_buffer_video((psx_cdrom_sector_mode2_t*) buffer, settings);
 			encode_sector_str(settings->video_frames, buffer, settings);
 		} else {
 			// Audio sector
@@ -333,16 +333,13 @@ void encode_file_str(settings_t *settings, FILE *output) {
 			buffer[0x00C] = ((t/75/60)%10)|(((t/75/60)/10)<<4);
 			buffer[0x00D] = (((t/75)%60)%10)|((((t/75)%60)/10)<<4);
 			buffer[0x00E] = ((t%75)%10)|(((t%75)/10)<<4);
-
-			// FIXME: EDC is not calculated in 2336-byte sector mode (shouldn't
-			// matter anyway, any CD image builder will have to recalculate it
-			// due to the sector's MSF changing)
-			if((j%interleave) < video_sectors_per_block) {
-				calculate_edc_data(buffer);
-			}
 		}
 
-		fwrite(buffer, sector_size, 1, output);
+		if((j%interleave) < video_sectors_per_block) {
+			calculate_edc_data(buffer);
+		}
+
+		fwrite(buffer + 2352 - sector_size, sector_size, 1, output);
 
 		time_t t = get_elapsed_time(settings);
 		if (t) {

psxavenc/mdec.c

@@ -614,13 +614,8 @@ void encode_sector_str(uint8_t *video_frames, uint8_t *output, settings_t *setti
 	header[0x00E] = (uint8_t)(settings->state_vid.bytes_used>>16);
 	header[0x00F] = (uint8_t)(settings->state_vid.bytes_used>>24);
 
-	if (settings->format == FORMAT_STR2CD) {
+	memcpy(output + 0x018, header, sizeof(header));
-		memcpy(output + 0x018, header, sizeof(header));
+	memcpy(output + 0x018 + 0x020, settings->state_vid.frame_output + settings->state_vid.frame_data_offset, 2016);
-		memcpy(output + 0x018 + 0x020, settings->state_vid.frame_output + settings->state_vid.frame_data_offset, 2016);
-	} else {
-		memcpy(output + 0x008, header, sizeof(header));
-		memcpy(output + 0x008 + 0x020, settings->state_vid.frame_output + settings->state_vid.frame_data_offset, 2016);
-	}
 
 	settings->state_vid.frame_data_offset += 2016;
 }

psxavenc/psxavenc.c

@@ -189,7 +189,7 @@ int parse_args(settings_t* settings, int argc, char** argv) {
 					return -1;
 				}
 				i = settings->video_fps_num / settings->video_fps_den;
-				if (i < 1 || i > 30) {
+				if (i < 1 || i > 60) {
 					fprintf(stderr, "Invalid frame rate: %d/%d\n", settings->video_fps_num, settings->video_fps_den);
 					return -1;
 				}