Compare commits

...

3 Commits

Author SHA1 Message Date
renovate[bot]
1158ae5212 fix(deps): update module golang.org/x/image to v0.31.0
Generated by Renovate Bot
2025-09-08 17:46:48 +00:00
Leo (Lei) Kang
6047a32ea0 [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
2025-09-04 14:33:07 -07:00
Leo (Lei) Kang
60bf158757 [CODEC] Add encoder bitrate tracker (#646)
add encoder bitrate tracker
2025-09-03 15:55:37 -07:00
7 changed files with 171 additions and 4 deletions

4
go.mod
View File

@@ -1,6 +1,6 @@
module github.com/pion/mediadevices module github.com/pion/mediadevices
go 1.21 go 1.24.0
require ( require (
github.com/blackjack/webcam v0.6.1 github.com/blackjack/webcam v0.6.1
@@ -13,7 +13,7 @@ require (
github.com/pion/rtp v1.8.19 github.com/pion/rtp v1.8.19
github.com/pion/webrtc/v4 v4.1.2 github.com/pion/webrtc/v4 v4.1.2
github.com/stretchr/testify v1.10.0 github.com/stretchr/testify v1.10.0
golang.org/x/image v0.23.0 golang.org/x/image v0.31.0
) )
require ( require (

4
go.sum
View File

@@ -62,8 +62,8 @@ github.com/wlynxg/anet v0.0.5 h1:J3VJGi1gvo0JwZ/P1/Yc/8p63SoW98B5dHkYDmpgvvU=
github.com/wlynxg/anet v0.0.5/go.mod h1:eay5PRQr7fIVAMbTbchTnO9gG65Hg/uYGdc7mguHxoA= github.com/wlynxg/anet v0.0.5/go.mod h1:eay5PRQr7fIVAMbTbchTnO9gG65Hg/uYGdc7mguHxoA=
golang.org/x/crypto v0.33.0 h1:IOBPskki6Lysi0lo9qQvbxiQ+FvsCC/YWOecCHAixus= golang.org/x/crypto v0.33.0 h1:IOBPskki6Lysi0lo9qQvbxiQ+FvsCC/YWOecCHAixus=
golang.org/x/crypto v0.33.0/go.mod h1:bVdXmD7IV/4GdElGPozy6U7lWdRXA4qyRVGJV57uQ5M= golang.org/x/crypto v0.33.0/go.mod h1:bVdXmD7IV/4GdElGPozy6U7lWdRXA4qyRVGJV57uQ5M=
golang.org/x/image v0.23.0 h1:HseQ7c2OpPKTPVzNjG5fwJsOTCiiwS4QdsYi5XU6H68= golang.org/x/image v0.31.0 h1:mLChjE2MV6g1S7oqbXC0/UcKijjm5fnJLUYKIYrLESA=
golang.org/x/image v0.23.0/go.mod h1:wJJBTdLfCCf3tiHa1fNxpZmUI4mmoZvwMCPP0ddoNKY= golang.org/x/image v0.31.0/go.mod h1:R9ec5Lcp96v9FTF+ajwaH3uGxPH4fKfHHAVbUILxghA=
golang.org/x/net v0.35.0 h1:T5GQRQb2y08kTAByq9L4/bz8cipCdA8FbRTXewonqY8= golang.org/x/net v0.35.0 h1:T5GQRQb2y08kTAByq9L4/bz8cipCdA8FbRTXewonqY8=
golang.org/x/net v0.35.0/go.mod h1:EglIi67kWsHKlRzzVMUD93VMSWGFOMSZgxFjparz1Qk= golang.org/x/net v0.35.0/go.mod h1:EglIi67kWsHKlRzzVMUD93VMSWGFOMSZgxFjparz1Qk=
golang.org/x/sys v0.0.0-20201018230417-eeed37f84f13/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs= golang.org/x/sys v0.0.0-20201018230417-eeed37f84f13/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=

View File

@@ -0,0 +1,48 @@
package codec
import (
"time"
)
type BitrateTracker struct {
windowSize time.Duration
buffer []int
times []time.Time
}
func NewBitrateTracker(windowSize time.Duration) *BitrateTracker {
return &BitrateTracker{
windowSize: windowSize,
}
}
func (bt *BitrateTracker) AddFrame(sizeBytes int, timestamp time.Time) {
bt.buffer = append(bt.buffer, sizeBytes)
bt.times = append(bt.times, timestamp)
// Remove old entries outside the window
cutoff := timestamp.Add(-bt.windowSize)
i := 0
for ; i < len(bt.times); i++ {
if bt.times[i].After(cutoff) {
break
}
}
bt.buffer = bt.buffer[i:]
bt.times = bt.times[i:]
}
func (bt *BitrateTracker) GetBitrate() float64 {
if len(bt.times) < 2 {
return 0
}
totalBytes := 0
for _, b := range bt.buffer {
totalBytes += b
}
duration := bt.times[len(bt.times)-1].Sub(bt.times[0]).Seconds()
if duration <= 0 {
return 0
}
return float64(totalBytes*8) / duration // bits per second
}

View File

@@ -0,0 +1,19 @@
package codec
import (
"math"
"testing"
"time"
)
func TestBitrateTracker(t *testing.T) {
packetSize := 1000
bt := NewBitrateTracker(time.Second)
bt.AddFrame(packetSize, time.Now())
bt.AddFrame(packetSize, time.Now().Add(time.Millisecond*100))
bt.AddFrame(packetSize, time.Now().Add(time.Millisecond*999))
eps := float64(packetSize*8) / 10
if got, want := bt.GetBitrate(), float64(packetSize*8)*3; math.Abs(got-want) > eps {
t.Fatalf("GetBitrate() = %v, want %v (|diff| <= %v)", got, want, eps)
}
}

View File

@@ -179,6 +179,12 @@ type BitRateController interface {
SetBitRate(int) error 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 // BaseParams represents an codec's encoding properties
type BaseParams struct { type BaseParams struct {
// Target bitrate in bps. // Target bitrate in bps.

View File

@@ -54,6 +54,7 @@ import (
"fmt" "fmt"
"image" "image"
"io" "io"
"math"
"sync" "sync"
"time" "time"
"unsafe" "unsafe"
@@ -81,6 +82,12 @@ type encoder struct {
closed bool closed bool
} }
const (
kRateControlThreshold = 0.15
kMinQuantizer = 20
kMaxQuantizer = 63
)
// VP8Params is codec specific paramaters // VP8Params is codec specific paramaters
type VP8Params struct { type VP8Params struct {
Params 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) 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() 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. // 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, // 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 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 { func (e *encoder) Controller() codec.EncoderController {
return e return e
} }

View File

@@ -4,6 +4,8 @@ import (
"context" "context"
"image" "image"
"io" "io"
"math"
"math/rand"
"sync/atomic" "sync/atomic"
"testing" "testing"
"time" "time"
@@ -13,6 +15,7 @@ import (
"github.com/pion/mediadevices/pkg/frame" "github.com/pion/mediadevices/pkg/frame"
"github.com/pion/mediadevices/pkg/io/video" "github.com/pion/mediadevices/pkg/io/video"
"github.com/pion/mediadevices/pkg/prop" "github.com/pion/mediadevices/pkg/prop"
"github.com/stretchr/testify/assert"
) )
func TestEncoder(t *testing.T) { 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)))
})
}