Correctly update encoder settings, clear csOptions after avifEncoderAddImage()

Author: tongyuantongyu

include/avif/avif.h

@@ -1130,6 +1130,7 @@ AVIF_API avifResult avifEncoderFinish(avifEncoder * encoder, avifRWData * output
 
 // Codec-specific, optional "advanced" tuning settings, in the form of string key/value pairs.
 // key must be non-NULL, but passing a NULL value will delete that key, if it exists.
+// Settings will be sent to codec once at the next avifEncoderAddImage() call.
 // Setting an incorrect or unknown option for the current codec will cause errors of type
 // AVIF_RESULT_INVALID_CODEC_SPECIFIC_OPTION from avifEncoderWrite() or avifEncoderAddImage().
 AVIF_API void avifEncoderSetCodecSpecificOption(avifEncoder * encoder, const char * key, const char * value);

src/codec_aom.c

@@ -518,6 +518,7 @@ static avifBool avifProcessAOMOptionsPostInit(avifCodec * codec, avifBool alpha)
         }
 #endif // defined(HAVE_AOM_CODEC_SET_OPTION)
     }
+    codec->csOptions->count = 0;
     return AVIF_TRUE;
 }
 
@@ -606,26 +607,30 @@ static avifResult aomCodecEncodeImage(avifCodec * codec,
                 avifDiagnosticsPrintf(codec->diag, "aom_codec_enc_config_default() failed: %s", aom_codec_err_to_string(err));
                 return AVIF_RESULT_UNKNOWN_ERROR;
             }
+        } else {
+            cfg->g_usage = aomUsage;
         }
 
-        // Set our own default cfg->rc_end_usage value, which may differ from libaom's default.
-        switch (aomUsage) {
-            case AOM_USAGE_GOOD_QUALITY:
-                // libaom's default is AOM_VBR. Change the default to AOM_Q since we don't need to
-                // hit a certain target bit rate. It's easier to control the worst quality in Q
-                // mode.
-                cfg->rc_end_usage = AOM_Q;
-                break;
-            case AOM_USAGE_REALTIME:
-                // For real-time mode we need to use CBR rate control mode. AOM_Q doesn't fit the
-                // rate control requirements for real-time mode. CBR does.
-                cfg->rc_end_usage = AOM_CBR;
-                break;
+        if (!codec->internal->endUsageSet) {
+            // Set our own default cfg->rc_end_usage value, which may differ from libaom's default.
+            switch (aomUsage) {
+                case AOM_USAGE_GOOD_QUALITY:
+                    // libaom's default is AOM_VBR. Change the default to AOM_Q since we don't need to
+                    // hit a certain target bit rate. It's easier to control the worst quality in Q
+                    // mode.
+                    cfg->rc_end_usage = AOM_Q;
+                    break;
+                case AOM_USAGE_REALTIME:
+                    // For real-time mode we need to use CBR rate control mode. AOM_Q doesn't fit the
+                    // rate control requirements for real-time mode. CBR does.
+                    cfg->rc_end_usage = AOM_CBR;
+                    break;
 #if defined(AOM_USAGE_ALL_INTRA)
-            case AOM_USAGE_ALL_INTRA:
-                cfg->rc_end_usage = AOM_Q;
-                break;
+                case AOM_USAGE_ALL_INTRA:
+                    cfg->rc_end_usage = AOM_Q;
+                    break;
 #endif
+            }
         }
 
         // Profile 0.  8-bit and 10-bit 4:2:0 and 4:0:0 only.
@@ -744,20 +749,12 @@ static avifResult aomCodecEncodeImage(avifCodec * codec,
             }
         }
 
-        if (lossless) {
-            aom_codec_control(&codec->internal->encoder, AV1E_SET_LOSSLESS, 1);
-        }
-        if (encoder->maxThreads > 1) {
-            aom_codec_control(&codec->internal->encoder, AV1E_SET_ROW_MT, 1);
-        }
-        if (encoder->tileRowsLog2 != 0) {
-            int tileRowsLog2 = AVIF_CLAMP(encoder->tileRowsLog2, 0, 6);
-            aom_codec_control(&codec->internal->encoder, AV1E_SET_TILE_ROWS, tileRowsLog2);
-        }
-        if (encoder->tileColsLog2 != 0) {
-            int tileColsLog2 = AVIF_CLAMP(encoder->tileColsLog2, 0, 6);
-            aom_codec_control(&codec->internal->encoder, AV1E_SET_TILE_COLUMNS, tileColsLog2);
-        }
+        aom_codec_control(&codec->internal->encoder, AV1E_SET_LOSSLESS, lossless);
+        aom_codec_control(&codec->internal->encoder, AV1E_SET_ROW_MT, encoder->maxThreads > 1);
+        int tileRowsLog2 = AVIF_CLAMP(encoder->tileRowsLog2, 0, 6);
+        aom_codec_control(&codec->internal->encoder, AV1E_SET_TILE_ROWS, tileRowsLog2);
+        int tileColsLog2 = AVIF_CLAMP(encoder->tileColsLog2, 0, 6);
+        aom_codec_control(&codec->internal->encoder, AV1E_SET_TILE_COLUMNS, tileColsLog2);
         if (aomCpuUsed != -1) {
             if (aom_codec_control(&codec->internal->encoder, AOME_SET_CPUUSED, aomCpuUsed) != AOM_CODEC_OK) {
                 return AVIF_RESULT_UNKNOWN_ERROR;

src/write.c

@@ -128,7 +128,6 @@ typedef struct avifEncoderData
     uint16_t primaryItemID;
     avifBool singleImage; // if true, the AVIF_ADD_IMAGE_FLAG_SINGLE flag was set on the first call to avifEncoderAddImage()
     avifBool alphaPresent;
-    avifBool csOptionsUpdated;
 } avifEncoderData;
 
 static void avifEncoderDataDestroy(avifEncoderData * data);
@@ -319,7 +318,6 @@ void avifEncoderDestroy(avifEncoder * encoder)
 void avifEncoderSetCodecSpecificOption(avifEncoder * encoder, const char * key, const char * value)
 {
     avifCodecSpecificOptionsSet(encoder->csOptions, key, value);
-    encoder->data->csOptionsUpdated = AVIF_TRUE; // False positive is possible but not important.
 }
 
 static void avifBackupSettings(avifEncoder * encoder)
@@ -339,7 +337,6 @@ static void avifBackupSettings(avifEncoder * encoder)
     lastEncoder->tileRowsLog2 = encoder->tileRowsLog2;
     lastEncoder->tileColsLog2 = encoder->tileColsLog2;
     lastEncoder->speed = encoder->speed;
-    encoder->data->csOptionsUpdated = AVIF_FALSE;
 }
 
 // This function detect changes made on avifEncoder.
@@ -362,7 +359,7 @@ static avifBool avifEncoderSettingsChanged(const avifEncoder * encoder, avifBool
         (lastEncoder->maxQuantizer != encoder->maxQuantizer) || (lastEncoder->minQuantizerAlpha != encoder->minQuantizerAlpha) ||
         (lastEncoder->maxQuantizerAlpha != encoder->maxQuantizerAlpha) || (lastEncoder->tileRowsLog2 != encoder->tileRowsLog2) ||
         (lastEncoder->tileColsLog2 != encoder->tileColsLog2) || (lastEncoder->speed != encoder->speed) ||
-        (encoder->data->csOptionsUpdated)) {
+        (encoder->csOptions->count > 0)) {
         *needUpdate = AVIF_TRUE;
     }
 
@@ -863,19 +860,11 @@ static avifResult avifEncoderAddImageInternal(avifEncoder * encoder,
         // Another frame in an image sequence
 
         const avifImage * imageMetadata = encoder->data->imageMetadata;
-        // HEIF (ISO 23008-12:2017), Section 6.6.2.3.1:
-        //   All input images shall have exactly the same width and height; call those tile_width and tile_height.
-        // MIAF (ISO 23000-22:2019), Section 7.3.11.4.1:
-        //   All input images of a grid image item shall use the same coding format, chroma sampling format, and the
-        //   same decoder configuration (see 7.3.6.2).
-        // If the first image in the sequence had an alpha plane (even if fully opaque), all
-        // subsequence images must have alpha as well.
-        if ((imageMetadata->width != firstCell->width) || (imageMetadata->height != firstCell->height) ||
-            (imageMetadata->depth != firstCell->depth) || (imageMetadata->yuvFormat != firstCell->yuvFormat) ||
+        // If the first image had an alpha plane (even if fully opaque), all subsequent images must have alpha as well.
+        if ((imageMetadata->depth != firstCell->depth) || (imageMetadata->yuvFormat != firstCell->yuvFormat) ||
             (imageMetadata->yuvRange != firstCell->yuvRange) || (imageMetadata->colorPrimaries != firstCell->colorPrimaries) ||
             (imageMetadata->transferCharacteristics != firstCell->transferCharacteristics) ||
             (imageMetadata->matrixCoefficients != firstCell->matrixCoefficients) ||
-            (!!imageMetadata->alphaPlane != !!firstCell->alphaPlane) ||
             (imageMetadata->alphaPremultiplied != firstCell->alphaPremultiplied) ||
             (encoder->data->alphaPresent && !firstCell->alphaPlane)) {
             return AVIF_RESULT_INCOMPATIBLE_IMAGE;