6 files changed, 424 insertions, 90 deletions

diff --git a/src/encoder/encoder.c b/src/encoder/encoder.c
new file mode 100644
index 0000000..0f8eda5
--- /dev/null
+++ b/src/encoder/encoder.c
@@ -0,0 +1,155 @@
+#include "../../include/encoder/encoder.h"
+#include "../../include/utils.h"
+
+#include <string.h>
+#include <stdio.h>
+
+#include <libavcodec/avcodec.h>
+#include <libavformat/avformat.h>
+
+bool gsr_encoder_init(gsr_encoder *self, gsr_replay_storage replay_storage, size_t replay_buffer_num_packets, double replay_buffer_time, const char *replay_directory) {
+    memset(self, 0, sizeof(*self));
+    self->num_recording_destinations = 0;
+    self->recording_destination_id_counter = 0;
+
+    if(pthread_mutex_init(&self->file_write_mutex, NULL) != 0) {
+        fprintf(stderr, "gsr error: gsr_encoder_init: failed to create mutex\n");
+        return false;
+    }
+    self->mutex_created = true;
+
+    if(replay_buffer_num_packets > 0) {
+        self->replay_buffer = gsr_replay_buffer_create(replay_storage, replay_directory, replay_buffer_time, replay_buffer_num_packets);
+        if(!self->replay_buffer) {
+            fprintf(stderr, "gsr error: gsr_encoder_init: failed to create replay buffer\n");
+            gsr_encoder_deinit(self);
+            return false;
+        }
+    }
+
+    return true;
+}
+
+void gsr_encoder_deinit(gsr_encoder *self)  {
+    if(self->mutex_created) {
+        self->mutex_created = false;
+        pthread_mutex_destroy(&self->file_write_mutex);
+    }
+
+    if(self->replay_buffer) {
+        gsr_replay_buffer_destroy(self->replay_buffer);
+        self->replay_buffer = NULL;
+    }
+
+    self->num_recording_destinations = 0;
+    self->recording_destination_id_counter = 0;
+}
+
+void gsr_encoder_receive_packets(gsr_encoder *self, AVCodecContext *codec_context, int64_t pts, int stream_index) {
+    for(;;) {
+        AVPacket *av_packet = av_packet_alloc();
+        if(!av_packet)
+            break;
+
+        av_packet->data = NULL;
+        av_packet->size = 0;
+        int res = avcodec_receive_packet(codec_context, av_packet);
+        if(res == 0) { // we have a packet, send the packet to the muxer
+            av_packet->stream_index = stream_index;
+            av_packet->pts = pts;
+            av_packet->dts = pts;
+
+            if(self->replay_buffer) {
+                const double time_now = clock_get_monotonic_seconds();
+                if(!gsr_replay_buffer_append(self->replay_buffer, av_packet, time_now))
+                    fprintf(stderr, "gsr error: gsr_encoder_receive_packets: failed to add replay buffer data\n");
+            }
+
+            pthread_mutex_lock(&self->file_write_mutex);
+            const bool is_keyframe = av_packet->flags & AV_PKT_FLAG_KEY;
+            for(size_t i = 0; i < self->num_recording_destinations; ++i) {
+                gsr_encoder_recording_destination *recording_destination = &self->recording_destinations[i];
+                if(recording_destination->codec_context != codec_context)
+                    continue;
+
+                if(is_keyframe)
+                    recording_destination->has_received_keyframe = true;
+                else if(!recording_destination->has_received_keyframe)
+                    continue;
+
+                av_packet->pts = pts - recording_destination->start_pts;
+                av_packet->dts = pts - recording_destination->start_pts;
+
+                av_packet_rescale_ts(av_packet, codec_context->time_base, recording_destination->stream->time_base);
+                // TODO: Is av_interleaved_write_frame needed?. Answer: might be needed for mkv but dont use it! it causes frames to be inconsistent, skipping frames and duplicating frames.
+                // TODO: av_interleaved_write_frame might be needed for cfr, or always for flv
+                const int ret = av_write_frame(recording_destination->format_context, av_packet);
+                if(ret < 0) {
+                    char error_buffer[AV_ERROR_MAX_STRING_SIZE];
+                    if(av_strerror(ret, error_buffer, sizeof(error_buffer)) < 0)
+                        snprintf(error_buffer, sizeof(error_buffer), "Unknown error");
+                    fprintf(stderr, "gsr error: gsr_encoder_receive_packets: failed to write frame index %d to muxer, reason: %s (%d)\n", av_packet->stream_index, error_buffer, ret);
+                }
+            }
+            pthread_mutex_unlock(&self->file_write_mutex);
+
+            av_packet_free(&av_packet);
+        } else if (res == AVERROR(EAGAIN)) { // we have no packet
+                                             // fprintf(stderr, "No packet!\n");
+            av_packet_free(&av_packet);
+            break;
+        } else if (res == AVERROR_EOF) { // this is the end of the stream
+            av_packet_free(&av_packet);
+            fprintf(stderr, "End of stream!\n");
+            break;
+        } else {
+            av_packet_free(&av_packet);
+            fprintf(stderr, "Unexpected error: %d\n", res);
+            break;
+        }
+    }
+}
+
+size_t gsr_encoder_add_recording_destination(gsr_encoder *self, AVCodecContext *codec_context, AVFormatContext *format_context, AVStream *stream, int64_t start_pts) {
+    if(self->num_recording_destinations >= GSR_MAX_RECORDING_DESTINATIONS) {
+        fprintf(stderr, "gsr error: gsr_encoder_add_recording_destination: failed to add destination, reached the max amount of recording destinations (%d)\n", GSR_MAX_RECORDING_DESTINATIONS);
+        return (size_t)-1;
+    }
+
+    for(size_t i = 0; i < self->num_recording_destinations; ++i) {
+        if(self->recording_destinations[i].stream == stream) {
+            fprintf(stderr, "gsr error: gsr_encoder_add_recording_destination: failed to add destination, the stream %p already exists as an output\n", (void*)stream);
+            return (size_t)-1;
+        }
+    }
+
+    pthread_mutex_lock(&self->file_write_mutex);
+    gsr_encoder_recording_destination *recording_destination = &self->recording_destinations[self->num_recording_destinations];
+    recording_destination->id = self->recording_destination_id_counter;
+    recording_destination->codec_context = codec_context;
+    recording_destination->format_context = format_context;
+    recording_destination->stream = stream;
+    recording_destination->start_pts = start_pts;
+    recording_destination->has_received_keyframe = false;
+
+    ++self->recording_destination_id_counter;
+    ++self->num_recording_destinations;
+    pthread_mutex_unlock(&self->file_write_mutex);
+
+    return recording_destination->id;
+}
+
+bool gsr_encoder_remove_recording_destination(gsr_encoder *self, size_t id) {
+    bool found = false;
+    pthread_mutex_lock(&self->file_write_mutex);
+    for(size_t i = 0; i < self->num_recording_destinations; ++i) {
+        if(self->recording_destinations[i].id == id) {
+            self->recording_destinations[i] = self->recording_destinations[self->num_recording_destinations - 1];
+            --self->num_recording_destinations;
+            found = true;
+            break;
+        }
+    }
+    pthread_mutex_unlock(&self->file_write_mutex);
+    return found;
+}
diff --git a/src/encoder/video/cuda.c b/src/encoder/video/nvenc.c
index f3cc9a4..5f578c2 100644
--- a/src/encoder/video/cuda.c
+++ b/src/encoder/video/nvenc.c
@@ -1,6 +1,7 @@
-#include "../../../include/encoder/video/cuda.h"
+#include "../../../include/encoder/video/nvenc.h"
 #include "../../../include/egl.h"
 #include "../../../include/cuda.h"
+#include "../../../include/window/window.h"
 
 #include <libavcodec/avcodec.h>
 #include <libavutil/hwcontext_cuda.h>
@@ -8,7 +9,7 @@
 #include <stdlib.h>
 
 typedef struct {
-    gsr_video_encoder_cuda_params params;
+    gsr_video_encoder_nvenc_params params;
 
     unsigned int target_textures[2];
 
@@ -18,12 +19,12 @@ typedef struct {
     CUgraphicsResource cuda_graphics_resources[2];
     CUarray mapped_arrays[2];
     CUstream cuda_stream;
-} gsr_video_encoder_cuda;
+} gsr_video_encoder_nvenc;
 
-static bool gsr_video_encoder_cuda_setup_context(gsr_video_encoder_cuda *self, AVCodecContext *video_codec_context) {
+static bool gsr_video_encoder_nvenc_setup_context(gsr_video_encoder_nvenc *self, AVCodecContext *video_codec_context) {
     self->device_ctx = av_hwdevice_ctx_alloc(AV_HWDEVICE_TYPE_CUDA);
     if(!self->device_ctx) {
-        fprintf(stderr, "gsr error: gsr_video_encoder_cuda_setup_context failed: failed to create hardware device context\n");
+        fprintf(stderr, "gsr error: gsr_video_encoder_nvenc_setup_context failed: failed to create hardware device context\n");
         return false;
     }
 
@@ -31,14 +32,14 @@ static bool gsr_video_encoder_cuda_setup_context(gsr_video_encoder_cuda *self, A
     AVCUDADeviceContext *cuda_device_context = (AVCUDADeviceContext*)hw_device_context->hwctx;
     cuda_device_context->cuda_ctx = self->cuda.cu_ctx;
     if(av_hwdevice_ctx_init(self->device_ctx) < 0) {
-        fprintf(stderr, "gsr error: gsr_video_encoder_cuda_setup_context failed: failed to create hardware device context\n");
+        fprintf(stderr, "gsr error: gsr_video_encoder_nvenc_setup_context failed: failed to create hardware device context\n");
         av_buffer_unref(&self->device_ctx);
         return false;
     }
 
     AVBufferRef *frame_context = av_hwframe_ctx_alloc(self->device_ctx);
     if(!frame_context) {
-        fprintf(stderr, "gsr error: gsr_video_encoder_cuda_setup_context failed: failed to create hwframe context\n");
+        fprintf(stderr, "gsr error: gsr_video_encoder_nvenc_setup_context failed: failed to create hwframe context\n");
         av_buffer_unref(&self->device_ctx);
         return false;
     }
@@ -51,7 +52,7 @@ static bool gsr_video_encoder_cuda_setup_context(gsr_video_encoder_cuda *self, A
     hw_frame_context->device_ctx = (AVHWDeviceContext*)self->device_ctx->data;
 
     if (av_hwframe_ctx_init(frame_context) < 0) {
-        fprintf(stderr, "gsr error: gsr_video_encoder_cuda_setup_context failed: failed to initialize hardware frame context "
+        fprintf(stderr, "gsr error: gsr_video_encoder_nvenc_setup_context failed: failed to initialize hardware frame context "
                         "(note: ffmpeg version needs to be > 4.0)\n");
         av_buffer_unref(&self->device_ctx);
         //av_buffer_unref(&frame_context);
@@ -64,21 +65,6 @@ static bool gsr_video_encoder_cuda_setup_context(gsr_video_encoder_cuda *self, A
     return true;
 }
 
-static unsigned int gl_create_texture(gsr_egl *egl, int width, int height, int internal_format, unsigned int format) {
-    unsigned int texture_id = 0;
-    egl->glGenTextures(1, &texture_id);
-    egl->glBindTexture(GL_TEXTURE_2D, texture_id);
-    egl->glTexImage2D(GL_TEXTURE_2D, 0, internal_format, width, height, 0, format, GL_UNSIGNED_BYTE, NULL);
-
-    egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
-    egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
-    egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
-    egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
-
-    egl->glBindTexture(GL_TEXTURE_2D, 0);
-    return texture_id;
-}
-
 static bool cuda_register_opengl_texture(gsr_cuda *cuda, CUgraphicsResource *cuda_graphics_resource, CUarray *mapped_array, unsigned int texture_id) {
     CUresult res;
     res = cuda->cuGraphicsGLRegisterImage(cuda_graphics_resource, texture_id, GL_TEXTURE_2D, CU_GRAPHICS_REGISTER_FLAGS_NONE);
@@ -96,10 +82,10 @@ static bool cuda_register_opengl_texture(gsr_cuda *cuda, CUgraphicsResource *cud
     return true;
 }
 
-static bool gsr_video_encoder_cuda_setup_textures(gsr_video_encoder_cuda *self, AVCodecContext *video_codec_context, AVFrame *frame) {
+static bool gsr_video_encoder_nvenc_setup_textures(gsr_video_encoder_nvenc *self, AVCodecContext *video_codec_context, AVFrame *frame) {
     const int res = av_hwframe_get_buffer(video_codec_context->hw_frames_ctx, frame, 0);
     if(res < 0) {
-        fprintf(stderr, "gsr error: gsr_video_encoder_cuda_setup_textures: av_hwframe_get_buffer failed: %d\n", res);
+        fprintf(stderr, "gsr error: gsr_video_encoder_nvenc_setup_textures: av_hwframe_get_buffer failed: %d\n", res);
         return false;
     }
 
@@ -109,9 +95,9 @@ static bool gsr_video_encoder_cuda_setup_textures(gsr_video_encoder_cuda *self,
     const int div[2] = {1, 2}; // divide UV texture size by 2 because chroma is half size
 
     for(int i = 0; i < 2; ++i) {
-        self->target_textures[i] = gl_create_texture(self->params.egl, video_codec_context->width / div[i], video_codec_context->height / div[i], self->params.color_depth == GSR_COLOR_DEPTH_8_BITS ? internal_formats_nv12[i] : internal_formats_p010[i], formats[i]);
+        self->target_textures[i] = gl_create_texture(self->params.egl, video_codec_context->width / div[i], video_codec_context->height / div[i], self->params.color_depth == GSR_COLOR_DEPTH_8_BITS ? internal_formats_nv12[i] : internal_formats_p010[i], formats[i], GL_NEAREST);
         if(self->target_textures[i] == 0) {
-            fprintf(stderr, "gsr error: gsr_video_encoder_cuda_setup_textures: failed to create opengl texture\n");
+            fprintf(stderr, "gsr error: gsr_video_encoder_nvenc_setup_textures: failed to create opengl texture\n");
             return false;
         }
 
@@ -123,32 +109,46 @@ static bool gsr_video_encoder_cuda_setup_textures(gsr_video_encoder_cuda *self,
     return true;
 }
 
-static void gsr_video_encoder_cuda_stop(gsr_video_encoder_cuda *self, AVCodecContext *video_codec_context);
+static void gsr_video_encoder_nvenc_stop(gsr_video_encoder_nvenc *self, AVCodecContext *video_codec_context);
 
-static bool gsr_video_encoder_cuda_start(gsr_video_encoder *encoder, AVCodecContext *video_codec_context, AVFrame *frame) {
-    gsr_video_encoder_cuda *self = encoder->priv;
+static bool gsr_video_encoder_nvenc_start(gsr_video_encoder *encoder, AVCodecContext *video_codec_context, AVFrame *frame) {
+    gsr_video_encoder_nvenc *self = encoder->priv;
 
-    const bool overclock = gsr_egl_get_display_server(self->params.egl) == GSR_DISPLAY_SERVER_X11 ? self->params.overclock : false;
-    if(!gsr_cuda_load(&self->cuda, self->params.egl->x11.dpy, overclock)) {
-        fprintf(stderr, "gsr error: gsr_video_encoder_cuda_start: failed to load cuda\n");
-        gsr_video_encoder_cuda_stop(self, video_codec_context);
+    const bool is_x11 = gsr_window_get_display_server(self->params.egl->window) == GSR_DISPLAY_SERVER_X11;
+    const bool overclock = is_x11 ? self->params.overclock : false;
+    Display *display = is_x11 ? gsr_window_get_display(self->params.egl->window) : NULL;
+    if(!gsr_cuda_load(&self->cuda, display, overclock)) {
+        fprintf(stderr, "gsr error: gsr_video_encoder_nvenc_start: failed to load cuda\n");
+        gsr_video_encoder_nvenc_stop(self, video_codec_context);
         return false;
     }
 
-    if(!gsr_video_encoder_cuda_setup_context(self, video_codec_context)) {
-        gsr_video_encoder_cuda_stop(self, video_codec_context);
+    video_codec_context->width = FFALIGN(video_codec_context->width, 2);
+    video_codec_context->height = FFALIGN(video_codec_context->height, 2);
+
+    if(video_codec_context->width < 128)
+        video_codec_context->width = 128;
+
+    if(video_codec_context->height < 128)
+        video_codec_context->height = 128;
+
+    frame->width = video_codec_context->width;
+    frame->height = video_codec_context->height;
+
+    if(!gsr_video_encoder_nvenc_setup_context(self, video_codec_context)) {
+        gsr_video_encoder_nvenc_stop(self, video_codec_context);
         return false;
     }
 
-    if(!gsr_video_encoder_cuda_setup_textures(self, video_codec_context, frame)) {
-        gsr_video_encoder_cuda_stop(self, video_codec_context);
+    if(!gsr_video_encoder_nvenc_setup_textures(self, video_codec_context, frame)) {
+        gsr_video_encoder_nvenc_stop(self, video_codec_context);
         return false;
     }
 
     return true;
 }
 
-void gsr_video_encoder_cuda_stop(gsr_video_encoder_cuda *self, AVCodecContext *video_codec_context) {
+void gsr_video_encoder_nvenc_stop(gsr_video_encoder_nvenc *self, AVCodecContext *video_codec_context) {
     self->params.egl->glDeleteTextures(2, self->target_textures);
     self->target_textures[0] = 0;
     self->target_textures[1] = 0;
@@ -171,8 +171,8 @@ void gsr_video_encoder_cuda_stop(gsr_video_encoder_cuda *self, AVCodecContext *v
     gsr_cuda_unload(&self->cuda);
 }
 
-static void gsr_video_encoder_cuda_copy_textures_to_frame(gsr_video_encoder *encoder, AVFrame *frame) {
-    gsr_video_encoder_cuda *self = encoder->priv;
+static void gsr_video_encoder_nvenc_copy_textures_to_frame(gsr_video_encoder *encoder, AVFrame *frame, gsr_color_conversion *color_conversion) {
+    gsr_video_encoder_nvenc *self = encoder->priv;
     const int div[2] = {1, 2}; // divide UV texture size by 2 because chroma is half size
     for(int i = 0; i < 2; ++i) {
         CUDA_MEMCPY2D memcpy_struct;
@@ -198,26 +198,26 @@ static void gsr_video_encoder_cuda_copy_textures_to_frame(gsr_video_encoder *enc
     self->cuda.cuStreamSynchronize(self->cuda_stream);
 }
 
-static void gsr_video_encoder_cuda_get_textures(gsr_video_encoder *encoder, unsigned int *textures, int *num_textures, gsr_destination_color *destination_color) {
-    gsr_video_encoder_cuda *self = encoder->priv;
+static void gsr_video_encoder_nvenc_get_textures(gsr_video_encoder *encoder, unsigned int *textures, int *num_textures, gsr_destination_color *destination_color) {
+    gsr_video_encoder_nvenc *self = encoder->priv;
     textures[0] = self->target_textures[0];
     textures[1] = self->target_textures[1];
     *num_textures = 2;
     *destination_color = self->params.color_depth == GSR_COLOR_DEPTH_10_BITS ? GSR_DESTINATION_COLOR_P010 : GSR_DESTINATION_COLOR_NV12;
 }
 
-static void gsr_video_encoder_cuda_destroy(gsr_video_encoder *encoder, AVCodecContext *video_codec_context) {
-    gsr_video_encoder_cuda_stop(encoder->priv, video_codec_context);
+static void gsr_video_encoder_nvenc_destroy(gsr_video_encoder *encoder, AVCodecContext *video_codec_context) {
+    gsr_video_encoder_nvenc_stop(encoder->priv, video_codec_context);
     free(encoder->priv);
     free(encoder);
 }
 
-gsr_video_encoder* gsr_video_encoder_cuda_create(const gsr_video_encoder_cuda_params *params) {
+gsr_video_encoder* gsr_video_encoder_nvenc_create(const gsr_video_encoder_nvenc_params *params) {
     gsr_video_encoder *encoder = calloc(1, sizeof(gsr_video_encoder));
     if(!encoder)
         return NULL;
 
-    gsr_video_encoder_cuda *encoder_cuda = calloc(1, sizeof(gsr_video_encoder_cuda));
+    gsr_video_encoder_nvenc *encoder_cuda = calloc(1, sizeof(gsr_video_encoder_nvenc));
     if(!encoder_cuda) {
         free(encoder);
         return NULL;
@@ -226,10 +226,10 @@ gsr_video_encoder* gsr_video_encoder_cuda_create(const gsr_video_encoder_cuda_pa
     encoder_cuda->params = *params;
 
     *encoder = (gsr_video_encoder) {
-        .start = gsr_video_encoder_cuda_start,
-        .copy_textures_to_frame = gsr_video_encoder_cuda_copy_textures_to_frame,
-        .get_textures = gsr_video_encoder_cuda_get_textures,
-        .destroy = gsr_video_encoder_cuda_destroy,
+        .start = gsr_video_encoder_nvenc_start,
+        .copy_textures_to_frame = gsr_video_encoder_nvenc_copy_textures_to_frame,
+        .get_textures = gsr_video_encoder_nvenc_get_textures,
+        .destroy = gsr_video_encoder_nvenc_destroy,
         .priv = encoder_cuda
     };
 
diff --git a/src/encoder/video/software.c b/src/encoder/video/software.c
index cedcc1b..d8d9828 100644
--- a/src/encoder/video/software.c
+++ b/src/encoder/video/software.c
@@ -1,5 +1,6 @@
 #include "../../../include/encoder/video/software.h"
 #include "../../../include/egl.h"
+#include "../../../include/utils.h"
 
 #include <libavcodec/avcodec.h>
 #include <libavutil/frame.h>
@@ -14,21 +15,6 @@ typedef struct {
     unsigned int target_textures[2];
 } gsr_video_encoder_software;
 
-static unsigned int gl_create_texture(gsr_egl *egl, int width, int height, int internal_format, unsigned int format) {
-    unsigned int texture_id = 0;
-    egl->glGenTextures(1, &texture_id);
-    egl->glBindTexture(GL_TEXTURE_2D, texture_id);
-    egl->glTexImage2D(GL_TEXTURE_2D, 0, internal_format, width, height, 0, format, GL_UNSIGNED_BYTE, NULL);
-
-    egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
-    egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
-    egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
-    egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
-
-    egl->glBindTexture(GL_TEXTURE_2D, 0);
-    return texture_id;
-}
-
 static bool gsr_video_encoder_software_setup_textures(gsr_video_encoder_software *self, AVCodecContext *video_codec_context, AVFrame *frame) {
     int res = av_frame_get_buffer(frame, LINESIZE_ALIGNMENT);
     if(res < 0) {
@@ -48,7 +34,7 @@ static bool gsr_video_encoder_software_setup_textures(gsr_video_encoder_software
     const int div[2] = {1, 2}; // divide UV texture size by 2 because chroma is half size
 
     for(int i = 0; i < 2; ++i) {
-        self->target_textures[i] = gl_create_texture(self->params.egl, video_codec_context->width / div[i], video_codec_context->height / div[i], self->params.color_depth == GSR_COLOR_DEPTH_8_BITS ? internal_formats_nv12[i] : internal_formats_p010[i], formats[i]);
+        self->target_textures[i] = gl_create_texture(self->params.egl, video_codec_context->width / div[i], video_codec_context->height / div[i], self->params.color_depth == GSR_COLOR_DEPTH_8_BITS ? internal_formats_nv12[i] : internal_formats_p010[i], formats[i], GL_NEAREST);
         if(self->target_textures[i] == 0) {
             fprintf(stderr, "gsr error: gsr_capture_kms_setup_cuda_textures: failed to create opengl texture\n");
             return false;
@@ -84,21 +70,20 @@ void gsr_video_encoder_software_stop(gsr_video_encoder_software *self, AVCodecCo
     self->target_textures[1] = 0;
 }
 
-static void gsr_video_encoder_software_copy_textures_to_frame(gsr_video_encoder *encoder, AVFrame *frame) {
-    gsr_video_encoder_software *self = encoder->priv;
+static void gsr_video_encoder_software_copy_textures_to_frame(gsr_video_encoder *encoder, AVFrame *frame, gsr_color_conversion *color_conversion) {
+    (void)encoder;
+    //gsr_video_encoder_software *self = encoder->priv;
     // TODO: hdr support
     const unsigned int formats[2] = { GL_RED, GL_RG };
+    const int div[2] = {1, 2}; // divide UV texture size by 2 because chroma is half size
     for(int i = 0; i < 2; ++i) {
-        self->params.egl->glBindTexture(GL_TEXTURE_2D, self->target_textures[i]);
-        // We could use glGetTexSubImage and then we wouldn't have to use a specific linesize (LINESIZE_ALIGNMENT) that adds padding,
-        // but glGetTexSubImage is only available starting from opengl 4.5.
-        self->params.egl->glGetTexImage(GL_TEXTURE_2D, 0, formats[i], GL_UNSIGNED_BYTE, frame->data[i]);
+        // TODO: Use glPixelStore?
+        gsr_color_conversion_read_destination_texture(color_conversion, i, 0, 0, frame->width / div[i], frame->height / div[i], formats[i], GL_UNSIGNED_BYTE, frame->data[i]);
     }
-    self->params.egl->glBindTexture(GL_TEXTURE_2D, 0);
     // cap_kms->kms.base.egl->eglSwapBuffers(cap_kms->kms.base.egl->egl_display, cap_kms->kms.base.egl->egl_surface);
 
-    self->params.egl->glFlush();
-    self->params.egl->glFinish();
+    //self->params.egl->glFlush();
+    //self->params.egl->glFinish();
 }
 
 static void gsr_video_encoder_software_get_textures(gsr_video_encoder *encoder, unsigned int *textures, int *num_textures, gsr_destination_color *destination_color) {
diff --git a/src/encoder/video/vaapi.c b/src/encoder/video/vaapi.c
index b5ccce9..0daf4d8 100644
--- a/src/encoder/video/vaapi.c
+++ b/src/encoder/video/vaapi.c
@@ -4,6 +4,7 @@
 
 #include <libavcodec/avcodec.h>
 #include <libavutil/hwcontext_vaapi.h>
+#include <libavutil/intreadwrite.h>
 
 #include <va/va_drmcommon.h>
 
@@ -120,10 +121,8 @@ static bool gsr_video_encoder_vaapi_setup_textures(gsr_video_encoder_vaapi *self
             }
 
             self->params.egl->glBindTexture(GL_TEXTURE_2D, self->target_textures[i]);
-            self->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
-            self->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
-            self->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
-            self->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
+            self->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
+            self->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
 
             while(self->params.egl->glGetError()) {}
             while(self->params.egl->eglGetError() != EGL_SUCCESS){}
@@ -166,12 +165,21 @@ static bool gsr_video_encoder_vaapi_start(gsr_video_encoder *encoder, AVCodecCon
         } else {
             video_codec_context->height = FFALIGN(video_codec_context->height, 16);
         }
+    } else {
+        video_codec_context->width = FFALIGN(video_codec_context->width, 2);
+        video_codec_context->height = FFALIGN(video_codec_context->height, 2);
     }
 
-    if(video_codec_context->width != frame->width || video_codec_context->height != frame->height) {
+    if(FFALIGN(video_codec_context->width, 2) != FFALIGN(frame->width, 2) || FFALIGN(video_codec_context->height, 2) != FFALIGN(frame->height, 2)) {
         fprintf(stderr, "gsr warning: gsr_video_encoder_vaapi_start: black bars have been added to the video because of a bug in AMD drivers/hardware. Record with h264 codec instead (-k h264) to get around this issue\n");
     }
 
+    if(video_codec_context->width < 128)
+        video_codec_context->width = 128;
+
+    if(video_codec_context->height < 128)
+        video_codec_context->height = 128;
+
     frame->width = video_codec_context->width;
     frame->height = video_codec_context->height;
 
@@ -235,7 +243,6 @@ gsr_video_encoder* gsr_video_encoder_vaapi_create(const gsr_video_encoder_vaapi_
 
     *encoder = (gsr_video_encoder) {
         .start = gsr_video_encoder_vaapi_start,
-        .copy_textures_to_frame = NULL,
         .get_textures = gsr_video_encoder_vaapi_get_textures,
         .destroy = gsr_video_encoder_vaapi_destroy,
         .priv = encoder_vaapi
diff --git a/src/encoder/video/video.c b/src/encoder/video/video.c
index 9b0def0..ce3b61b 100644
--- a/src/encoder/video/video.c
+++ b/src/encoder/video/video.c
@@ -1,4 +1,5 @@
 #include "../../../include/encoder/video/video.h"
+
 #include <assert.h>
 
 bool gsr_video_encoder_start(gsr_video_encoder *encoder, AVCodecContext *video_codec_context, AVFrame *frame) {
@@ -9,18 +10,19 @@ bool gsr_video_encoder_start(gsr_video_encoder *encoder, AVCodecContext *video_c
     return res;
 }
 
-void gsr_video_encoder_copy_textures_to_frame(gsr_video_encoder *encoder, AVFrame *frame) {
+void gsr_video_encoder_destroy(gsr_video_encoder *encoder, AVCodecContext *video_codec_context) {
     assert(encoder->started);
-    if(encoder->copy_textures_to_frame)
-        encoder->copy_textures_to_frame(encoder, frame);
+    encoder->started = false;
+    encoder->destroy(encoder, video_codec_context);
 }
 
-void gsr_video_encoder_get_textures(gsr_video_encoder *encoder, unsigned int *textures, int *num_textures, gsr_destination_color *destination_color) {
+void gsr_video_encoder_copy_textures_to_frame(gsr_video_encoder *encoder, AVFrame *frame, gsr_color_conversion *color_conversion) {
     assert(encoder->started);
-    encoder->get_textures(encoder, textures, num_textures, destination_color);
+    if(encoder->copy_textures_to_frame)
+        encoder->copy_textures_to_frame(encoder, frame, color_conversion);
 }
 
-void gsr_video_encoder_destroy(gsr_video_encoder *encoder, AVCodecContext *video_codec_context) {
+void gsr_video_encoder_get_textures(gsr_video_encoder *encoder, unsigned int *textures, int *num_textures, gsr_destination_color *destination_color) {
     assert(encoder->started);
-    encoder->destroy(encoder, video_codec_context);
+    encoder->get_textures(encoder, textures, num_textures, destination_color);
 }
diff --git a/src/encoder/video/vulkan.c b/src/encoder/video/vulkan.c
index 4b0f1ea..802934d 100644
--- a/src/encoder/video/vulkan.c
+++ b/src/encoder/video/vulkan.c
@@ -3,8 +3,16 @@
 #include "../../../include/egl.h"
 
 #include <libavcodec/avcodec.h>
+#define VK_NO_PROTOTYPES
 #include <libavutil/hwcontext_vulkan.h>
 
+//#include <vulkan/vulkan_core.h>
+
+#define GL_HANDLE_TYPE_OPAQUE_FD_EXT      0x9586
+#define GL_TEXTURE_TILING_EXT             0x9580
+#define GL_OPTIMAL_TILING_EXT             0x9584
+#define GL_LINEAR_TILING_EXT              0x9585
+
 typedef struct {
     gsr_video_encoder_vulkan_params params;
     unsigned int target_textures[2];
@@ -12,8 +20,12 @@ typedef struct {
 } gsr_video_encoder_vulkan;
 
 static bool gsr_video_encoder_vulkan_setup_context(gsr_video_encoder_vulkan *self, AVCodecContext *video_codec_context) {
+    AVDictionary *options = NULL;
+    //av_dict_set(&options, "linear_images", "1", 0);
+    //av_dict_set(&options, "disable_multiplane", "1", 0);
+#if 0
     // TODO: Use correct device
-    if(av_hwdevice_ctx_create(&self->device_ctx, AV_HWDEVICE_TYPE_VULKAN, NULL, NULL, 0) < 0) {
+    if(av_hwdevice_ctx_create(&self->device_ctx, AV_HWDEVICE_TYPE_VULKAN, NULL, options, 0) < 0) {
         fprintf(stderr, "gsr error: gsr_video_encoder_vulkan_setup_context: failed to create hardware device context\n");
         return false;
     }
@@ -32,6 +44,7 @@ static bool gsr_video_encoder_vulkan_setup_context(gsr_video_encoder_vulkan *sel
     hw_frame_context->format = video_codec_context->pix_fmt;
     hw_frame_context->device_ctx = (AVHWDeviceContext*)self->device_ctx->data;
 
+    //AVVulkanFramesContext *vk_frame_ctx = (AVVulkanFramesContext*)hw_frame_context->hwctx;
     //hw_frame_context->initial_pool_size = 20;
 
     if (av_hwframe_ctx_init(frame_context) < 0) {
@@ -44,15 +57,175 @@ static bool gsr_video_encoder_vulkan_setup_context(gsr_video_encoder_vulkan *sel
 
     video_codec_context->hw_frames_ctx = av_buffer_ref(frame_context);
     av_buffer_unref(&frame_context);
+#endif
     return true;
 }
 
+static AVVulkanDeviceContext* video_codec_context_get_vulkan_data(AVCodecContext *video_codec_context) {
+    AVBufferRef *hw_frames_ctx = video_codec_context->hw_frames_ctx;
+    if(!hw_frames_ctx)
+        return NULL;
+
+    AVHWFramesContext *hw_frame_context = (AVHWFramesContext*)hw_frames_ctx->data;
+    AVHWDeviceContext *device_context = (AVHWDeviceContext*)hw_frame_context->device_ctx;
+    if(device_context->type != AV_HWDEVICE_TYPE_VULKAN)
+        return NULL;
+
+    return (AVVulkanDeviceContext*)device_context->hwctx;
+}
+
+static uint32_t get_memory_type_idx(VkPhysicalDevice pdev, const VkMemoryRequirements *mem_reqs, VkMemoryPropertyFlagBits prop_flags, PFN_vkGetPhysicalDeviceMemoryProperties vkGetPhysicalDeviceMemoryProperties) {
+    VkPhysicalDeviceMemoryProperties pdev_mem_props;
+    uint32_t i;
+
+    vkGetPhysicalDeviceMemoryProperties(pdev, &pdev_mem_props);
+
+    for (i = 0; i < pdev_mem_props.memoryTypeCount; i++) {
+        const VkMemoryType *type = &pdev_mem_props.memoryTypes[i];
+
+        if ((mem_reqs->memoryTypeBits & (1 << i)) &&
+            (type->propertyFlags & prop_flags) == prop_flags) {
+            return i;
+            break;
+        }
+    }
+    return UINT32_MAX;
+}
+
 static bool gsr_video_encoder_vulkan_setup_textures(gsr_video_encoder_vulkan *self, AVCodecContext *video_codec_context, AVFrame *frame) {
     const int res = av_hwframe_get_buffer(video_codec_context->hw_frames_ctx, frame, 0);
     if(res < 0) {
         fprintf(stderr, "gsr error: gsr_video_encoder_vulkan_setup_textures: av_hwframe_get_buffer failed: %d\n", res);
         return false;
     }
+
+    while(self->params.egl->glGetError()) {}
+#if 0
+    AVVkFrame *target_surface_id = (AVVkFrame*)frame->data[0];
+    AVVulkanDeviceContext* vv = video_codec_context_get_vulkan_data(video_codec_context);
+    const size_t luma_size = frame->width * frame->height;
+    if(vv) {
+        PFN_vkGetImageMemoryRequirements vkGetImageMemoryRequirements = (PFN_vkGetImageMemoryRequirements)vv->get_proc_addr(vv->inst, "vkGetImageMemoryRequirements");
+        PFN_vkAllocateMemory vkAllocateMemory = (PFN_vkAllocateMemory)vv->get_proc_addr(vv->inst, "vkAllocateMemory");
+        PFN_vkGetPhysicalDeviceMemoryProperties vkGetPhysicalDeviceMemoryProperties = (PFN_vkGetPhysicalDeviceMemoryProperties)vv->get_proc_addr(vv->inst, "vkGetPhysicalDeviceMemoryProperties");
+        PFN_vkGetMemoryFdKHR vkGetMemoryFdKHR = (PFN_vkGetMemoryFdKHR)vv->get_proc_addr(vv->inst, "vkGetMemoryFdKHR");
+
+        VkMemoryRequirements mem_reqs = {0};
+        vkGetImageMemoryRequirements(vv->act_dev, target_surface_id->img[0], &mem_reqs);
+
+        fprintf(stderr, "size: %lu, alignment: %lu, memory bits: 0x%08x\n", mem_reqs.size, mem_reqs.alignment, mem_reqs.memoryTypeBits);
+        VkDeviceMemory mem;
+        {
+            VkExportMemoryAllocateInfo exp_mem_info;
+            VkMemoryAllocateInfo mem_alloc_info;
+            VkMemoryDedicatedAllocateInfoKHR ded_info;
+
+            memset(&exp_mem_info, 0, sizeof(exp_mem_info));
+            exp_mem_info.sType = VK_STRUCTURE_TYPE_EXPORT_MEMORY_ALLOCATE_INFO;
+            exp_mem_info.handleTypes = VK_EXTERNAL_MEMORY_HANDLE_TYPE_OPAQUE_FD_BIT;
+            
+            memset(&ded_info, 0, sizeof(ded_info));
+            ded_info.sType = VK_STRUCTURE_TYPE_MEMORY_DEDICATED_ALLOCATE_INFO;
+            ded_info.image = target_surface_id->img[0];
+
+            exp_mem_info.pNext = &ded_info;
+
+            memset(&mem_alloc_info, 0, sizeof(mem_alloc_info));
+            mem_alloc_info.sType = VK_STRUCTURE_TYPE_MEMORY_ALLOCATE_INFO;
+            mem_alloc_info.pNext = &exp_mem_info;
+            mem_alloc_info.allocationSize = target_surface_id->size[0];
+            mem_alloc_info.memoryTypeIndex = get_memory_type_idx(vv->phys_dev, &mem_reqs, VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT, vkGetPhysicalDeviceMemoryProperties);
+
+            if (mem_alloc_info.memoryTypeIndex == UINT32_MAX) {
+                fprintf(stderr, "No suitable memory type index found.\n");
+                return VK_NULL_HANDLE;
+            }
+
+            if (vkAllocateMemory(vv->act_dev, &mem_alloc_info, 0, &mem) !=
+                VK_SUCCESS)
+                return VK_NULL_HANDLE;
+ 
+            fprintf(stderr, "memory: %p\n", (void*)mem);
+
+        }
+
+        fprintf(stderr, "target surface id: %p, %zu, %zu\n", (void*)target_surface_id->mem[0], target_surface_id->offset[0], target_surface_id->offset[1]);
+        fprintf(stderr, "vkGetMemoryFdKHR: %p\n", (void*)vkGetMemoryFdKHR);
+
+        int fd = 0;
+        VkMemoryGetFdInfoKHR fd_info;
+        memset(&fd_info, 0, sizeof(fd_info));
+        fd_info.sType = VK_STRUCTURE_TYPE_MEMORY_GET_FD_INFO_KHR;
+        fd_info.memory = target_surface_id->mem[0];
+        fd_info.handleType = VK_EXTERNAL_MEMORY_HANDLE_TYPE_OPAQUE_FD_BIT;
+        if(vkGetMemoryFdKHR(vv->act_dev, &fd_info, &fd) != VK_SUCCESS) {
+            fprintf(stderr, "failed!\n");
+        } else {
+            fprintf(stderr, "fd: %d\n", fd);
+        }
+
+        fprintf(stderr, "glImportMemoryFdEXT: %p, size: %zu\n", (void*)self->params.egl->glImportMemoryFdEXT, target_surface_id->size[0]);
+        const int tiling = target_surface_id->tiling == VK_IMAGE_TILING_LINEAR ? GL_LINEAR_TILING_EXT : GL_OPTIMAL_TILING_EXT;
+
+        if(tiling != GL_OPTIMAL_TILING_EXT) {
+            fprintf(stderr, "tiling %d is not supported, only GL_OPTIMAL_TILING_EXT (%d) is supported\n", tiling, GL_OPTIMAL_TILING_EXT);
+        }
+
+
+        unsigned int gl_memory_obj = 0;
+        self->params.egl->glCreateMemoryObjectsEXT(1, &gl_memory_obj);
+
+        //const int dedicated = GL_TRUE;
+        //self->params.egl->glMemoryObjectParameterivEXT(gl_memory_obj, GL_DEDICATED_MEMORY_OBJECT_EXT, &dedicated);
+
+        self->params.egl->glImportMemoryFdEXT(gl_memory_obj, target_surface_id->size[0], GL_HANDLE_TYPE_OPAQUE_FD_EXT, fd);
+        if(!self->params.egl->glIsMemoryObjectEXT(gl_memory_obj))
+            fprintf(stderr, "failed to create object!\n");
+
+        fprintf(stderr, "gl memory obj: %u, error: %d\n", gl_memory_obj, self->params.egl->glGetError());
+
+        // fprintf(stderr, "0 gl error: %d\n", self->params.egl->glGetError());
+        // unsigned int vertex_buffer = 0;
+        // self->params.egl->glGenBuffers(1, &vertex_buffer);
+        // self->params.egl->glBindBuffer(GL_ARRAY_BUFFER, vertex_buffer);
+        // self->params.egl->glBufferStorageMemEXT(GL_ARRAY_BUFFER, target_surface_id->size[0], gl_memory_obj, target_surface_id->offset[0]);
+        // fprintf(stderr, "1 gl error: %d\n", self->params.egl->glGetError());
+
+        // fprintf(stderr, "0 gl error: %d\n", self->params.egl->glGetError());
+        // unsigned int buffer = 0;
+        // self->params.egl->glCreateBuffers(1, &buffer);
+        // self->params.egl->glNamedBufferStorageMemEXT(buffer, target_surface_id->size[0], gl_memory_obj, target_surface_id->offset[0]);
+        // fprintf(stderr, "1 gl error: %d\n", self->params.egl->glGetError());
+
+        self->params.egl->glGenTextures(1, &self->target_textures[0]);
+        self->params.egl->glBindTexture(GL_TEXTURE_2D, self->target_textures[0]);
+
+        fprintf(stderr, "1 gl error: %d\n", self->params.egl->glGetError());
+        self->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_TILING_EXT, tiling);
+
+        fprintf(stderr, "tiling: %d\n", tiling);
+
+        fprintf(stderr, "2 gl error: %d\n", self->params.egl->glGetError());
+        self->params.egl->glTexStorageMem2DEXT(GL_TEXTURE_2D, 1, GL_R8, frame->width, frame->height, gl_memory_obj, target_surface_id->offset[0]);
+
+        fprintf(stderr, "3 gl error: %d\n", self->params.egl->glGetError());
+        self->params.egl->glBindTexture(GL_TEXTURE_2D, 0);
+
+        self->params.egl->glGenTextures(1, &self->target_textures[1]);
+        self->params.egl->glBindTexture(GL_TEXTURE_2D, self->target_textures[1]);
+
+        fprintf(stderr, "1 gl error: %d\n", self->params.egl->glGetError());
+        self->params.egl->glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_TILING_EXT, tiling);
+
+        fprintf(stderr, "tiling: %d\n", tiling);
+
+        fprintf(stderr, "2 gl error: %d\n", self->params.egl->glGetError());
+        self->params.egl->glTexStorageMem2DEXT(GL_TEXTURE_2D, 1, GL_RG8, frame->width/2, frame->height/2, gl_memory_obj, target_surface_id->offset[0] + luma_size);
+
+        fprintf(stderr, "3 gl error: %d\n", self->params.egl->glGetError());
+        self->params.egl->glBindTexture(GL_TEXTURE_2D, 0);
+     }
+#endif
     return true;
 }
 
@@ -61,6 +234,18 @@ static void gsr_video_encoder_vulkan_stop(gsr_video_encoder_vulkan *self, AVCode
 static bool gsr_video_encoder_vulkan_start(gsr_video_encoder *encoder, AVCodecContext *video_codec_context, AVFrame *frame) {
     gsr_video_encoder_vulkan *self = encoder->priv;
 
+    video_codec_context->width = FFALIGN(video_codec_context->width, 2);
+    video_codec_context->height = FFALIGN(video_codec_context->height, 2);
+
+    if(video_codec_context->width < 128)
+        video_codec_context->width = 128;
+
+    if(video_codec_context->height < 128)
+        video_codec_context->height = 128;
+
+    frame->width = video_codec_context->width;
+    frame->height = video_codec_context->height;
+
     if(!gsr_video_encoder_vulkan_setup_context(self, video_codec_context)) {
         gsr_video_encoder_vulkan_stop(self, video_codec_context);
         return false;