[VPX] vpx dynamic encoding (#647)

* Add vp8 decoder and dynamic vp8 decoding

* Add QPController

* change parameters into const

* move decoder into another PR

* use explicit parameter name

pkg/codec/codec.go

@@ -179,6 +179,12 @@ type BitRateController interface {
 	SetBitRate(int) error
 }
 
+type QPController interface {
+	EncoderController
+	// DynamicQPControl adjusts the QP of the encoder based on the current and target bitrate
+	DynamicQPControl(currentBitrate int, targetBitrate int) error
+}
+
 // BaseParams represents an codec's encoding properties
 type BaseParams struct {
 	// Target bitrate in bps.

pkg/codec/vpx/vpx.go

@@ -54,6 +54,7 @@ import (
 	"fmt"
 	"image"
 	"io"
+	"math"
 	"sync"
 	"time"
 	"unsafe"
@@ -81,6 +82,12 @@ type encoder struct {
 	closed bool
 }
 
+const (
+	kRateControlThreshold = 0.15
+	kMinQuantizer         = 20
+	kMaxQuantizer         = 63
+)
+
 // VP8Params is codec specific paramaters
 type VP8Params struct {
 	Params
@@ -254,6 +261,10 @@ func (e *encoder) Read() ([]byte, func(), error) {
 		e.raw.d_w, e.raw.d_h = C.uint(width), C.uint(height)
 	}
 
+	if ec := C.vpx_codec_enc_config_set(e.codec, e.cfg); ec != 0 {
+		return nil, func() {}, fmt.Errorf("vpx_codec_enc_config_set failed (%d)", ec)
+	}
+
 	duration := t.Sub(e.tLastFrame).Microseconds()
 	// VPX doesn't allow 0 duration. If 0 is given, vpx_codec_encode will fail with VPX_CODEC_INVALID_PARAM.
 	// 0 duration is possible because mediadevices first gets the frame meta data by reading from the source,
@@ -322,6 +333,24 @@ func (e *encoder) SetBitRate(bitrate int) error {
 	return nil
 }
 
+func (e *encoder) DynamicQPControl(currentBitrate int, targetBitrate int) error {
+	e.mu.Lock()
+	defer e.mu.Unlock()
+	bitrateDiff := math.Abs(float64(currentBitrate - targetBitrate))
+	if bitrateDiff <= float64(currentBitrate)*kRateControlThreshold {
+		return nil
+	}
+	currentMax := e.cfg.rc_max_quantizer
+
+	if targetBitrate < currentBitrate {
+		e.cfg.rc_max_quantizer = min(currentMax+1, kMaxQuantizer)
+	} else {
+		e.cfg.rc_max_quantizer = max(currentMax-1, kMinQuantizer)
+	}
+	e.cfg.rc_min_quantizer = e.cfg.rc_max_quantizer
+	return nil
+}
+
 func (e *encoder) Controller() codec.EncoderController {
 	return e
 }

pkg/codec/vpx/vpx_test.go

@@ -4,6 +4,8 @@ import (
 	"context"
 	"image"
 	"io"
+	"math"
+	"math/rand"
 	"sync/atomic"
 	"testing"
 	"time"
@@ -13,6 +15,7 @@ import (
 	"github.com/pion/mediadevices/pkg/frame"
 	"github.com/pion/mediadevices/pkg/io/video"
 	"github.com/pion/mediadevices/pkg/prop"
+	"github.com/stretchr/testify/assert"
 )
 
 func TestEncoder(t *testing.T) {
@@ -360,3 +363,65 @@ func TestEncoderFrameMonotonic(t *testing.T) {
 		}
 	}
 }
+
+func TestVP8DynamicQPControl(t *testing.T) {
+	t.Run("VP8", func(t *testing.T) {
+		p, err := NewVP8Params()
+		if err != nil {
+			t.Fatal(err)
+		}
+		p.LagInFrames = 0 // Disable frame lag buffering for real-time encoding
+		p.RateControlEndUsage = RateControlCBR
+		totalFrames := 100
+		frameRate := 10
+		initialWidth, initialHeight := 800, 600
+		var cnt uint32
+
+		r, err := p.BuildVideoEncoder(
+			video.ReaderFunc(func() (image.Image, func(), error) {
+				i := atomic.AddUint32(&cnt, 1)
+				if i == uint32(totalFrames+1) {
+					return nil, nil, io.EOF
+				}
+				img := image.NewYCbCr(image.Rect(0, 0, initialWidth, initialHeight), image.YCbCrSubsampleRatio420)
+				r := rand.New(rand.NewSource(time.Now().UnixNano()))
+				for i := range img.Y {
+					img.Y[i] = uint8(r.Intn(256))
+				}
+				for i := range img.Cb {
+					img.Cb[i] = uint8(r.Intn(256))
+				}
+				for i := range img.Cr {
+					img.Cr[i] = uint8(r.Intn(256))
+				}
+				return img, func() {}, nil
+			}),
+			prop.Media{
+				Video: prop.Video{
+					Width:       initialWidth,
+					Height:      initialHeight,
+					FrameRate:   float32(frameRate),
+					FrameFormat: frame.FormatI420,
+				},
+			},
+		)
+		if err != nil {
+			t.Fatal(err)
+		}
+		initialBitrate := 100
+		currentBitrate := initialBitrate
+		targetBitrate := 300
+		for i := 0; i < totalFrames; i++ {
+			r.Controller().(codec.KeyFrameController).ForceKeyFrame()
+			r.Controller().(codec.QPController).DynamicQPControl(currentBitrate, targetBitrate)
+			data, rel, err := r.Read()
+			if err != nil {
+				t.Fatal(err)
+			}
+			rel()
+			encodedSize := len(data)
+			currentBitrate = encodedSize * 8 / 1000 / frameRate
+		}
+		assert.Less(t, math.Abs(float64(targetBitrate-currentBitrate)), math.Abs(float64(initialBitrate-currentBitrate)))
+	})
+}