api.rs 21.1 KB
Newer Older
rzumer's avatar
rzumer committed
1 2 3 4 5 6 7 8 9
// Copyright (c) 2018, The rav1e contributors. All rights reserved
//
// This source code is subject to the terms of the BSD 2 Clause License and
// the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
// was not distributed with this source code in the LICENSE file, you can
// obtain it at www.aomedia.org/license/software. If the Alliance for Open
// Media Patent License 1.0 was not distributed with this source code in the
// PATENTS file, you can obtain it at www.aomedia.org/license/patent.

Raphaël Zumer's avatar
Raphaël Zumer committed
10
use bitstream_io::*;
11 12 13
use crate::encoder::*;
use crate::metrics::calculate_frame_psnr;
use crate::partition::*;
14
use crate::rate::RCState;
15
use crate::scenechange::SceneChangeDetector;
Raphaël Zumer's avatar
Raphaël Zumer committed
16
use self::EncoderStatus::*;
Luca Barbato's avatar
Luca Barbato committed
17

Raphaël Zumer's avatar
Raphaël Zumer committed
18
use std::{cmp, fmt, io};
fbossen's avatar
fbossen committed
19
use std::collections::BTreeMap;
Luca Barbato's avatar
Luca Barbato committed
20
use std::sync::Arc;
Josh Holmer's avatar
Josh Holmer committed
21
use std::collections::BTreeSet;
22
use std::path::PathBuf;
Josh Holmer's avatar
Josh Holmer committed
23 24

const LOOKAHEAD_FRAMES: u64 = 10;
Luca Barbato's avatar
Luca Barbato committed
25 26 27

// TODO: use the num crate?
#[derive(Clone, Copy, Debug)]
28
#[repr(C)]
Luca Barbato's avatar
Luca Barbato committed
29 30 31
pub struct Rational {
  pub num: u64,
  pub den: u64
Luca Barbato's avatar
Luca Barbato committed
32 33
}

Luca Barbato's avatar
Luca Barbato committed
34 35 36
impl Rational {
  pub fn new(num: u64, den: u64) -> Self {
    Rational { num, den }
Luca Barbato's avatar
Luca Barbato committed
37 38 39
  }
}

40 41 42 43 44 45 46
#[derive(Clone, Copy, Debug)]
#[repr(C)]
pub struct Point {
  pub x: u16,
  pub y: u16
}

47
#[derive(Clone, Debug)]
48
pub struct EncoderConfig {
49
  // output size
50 51
  pub width: usize,
  pub height: usize,
52 53

  // data format and ancillary color information
54 55 56
  pub bit_depth: usize,
  pub chroma_sampling: ChromaSampling,
  pub chroma_sample_position: ChromaSamplePosition,
57 58 59 60 61 62
  pub pixel_range: PixelRange,
  pub color_description: Option<ColorDescription>,
  pub mastering_display: Option<MasteringDisplay>,
  pub content_light: Option<ContentLight>,

  // encoder configuration
63
  pub time_base: Rational,
Josh Holmer's avatar
Josh Holmer committed
64 65 66 67
  /// The *minimum* interval between two keyframes
  pub min_key_frame_interval: u64,
  /// The *maximum* interval between two keyframes
  pub max_key_frame_interval: u64,
68
  pub low_latency: bool,
69
  pub quantizer: usize,
70 71
  pub tune: Tune,
  pub speed_settings: SpeedSettings,
72 73
  /// `None` for one-pass encode. `Some(1)` or `Some(2)` for two-pass encoding.
  pub pass: Option<u8>,
74
  pub show_psnr: bool,
75
  pub stats_file: Option<PathBuf>,
76 77 78 79
}

impl Default for EncoderConfig {
  fn default() -> Self {
80 81
    const DEFAULT_SPEED: usize = 3;
    Self::with_speed_preset(DEFAULT_SPEED)
82 83 84
  }
}

85 86 87
impl EncoderConfig {
  pub fn with_speed_preset(speed: usize) -> Self {
    EncoderConfig {
88 89
      width: 640,
      height: 480,
90

91 92 93
      bit_depth: 8,
      chroma_sampling: ChromaSampling::Cs420,
      chroma_sample_position: ChromaSamplePosition::Unknown,
94 95 96 97 98
      pixel_range: PixelRange::Unspecified,
      color_description: None,
      mastering_display: None,
      content_light: None,

99
      time_base: Rational { num: 30, den: 1 },
Josh Holmer's avatar
Josh Holmer committed
100 101
      min_key_frame_interval: 12,
      max_key_frame_interval: 240,
102
      low_latency: false,
103 104
      quantizer: 100,
      tune: Tune::Psnr,
105
      speed_settings: SpeedSettings::from_preset(speed),
106
      pass: None,
107
      show_psnr: false,
108
      stats_file: None,
109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167
    }
  }
}

#[derive(Clone, Copy, Debug)]
pub struct SpeedSettings {
  pub min_block_size: BlockSize,
  pub multiref: bool,
  pub fast_deblock: bool,
  pub reduced_tx_set: bool,
  pub tx_domain_distortion: bool,
  pub encode_bottomup: bool,
  pub rdo_tx_decision: bool,
  pub prediction_modes: PredictionModesSetting,
  pub include_near_mvs: bool,
}

impl SpeedSettings {
  pub fn from_preset(speed: usize) -> Self {
    SpeedSettings {
      min_block_size: Self::min_block_size_preset(speed),
      multiref: Self::multiref_preset(speed),
      fast_deblock: Self::fast_deblock_preset(speed),
      reduced_tx_set: Self::reduced_tx_set_preset(speed),
      tx_domain_distortion: Self::tx_domain_distortion_preset(speed),
      encode_bottomup: Self::encode_bottomup_preset(speed),
      rdo_tx_decision: Self::rdo_tx_decision_preset(speed),
      prediction_modes: Self::prediction_modes_preset(speed),
      include_near_mvs: Self::include_near_mvs_preset(speed),
    }
  }

  fn min_block_size_preset(speed: usize) -> BlockSize {
    if speed <= 1 {
      BlockSize::BLOCK_4X4
    } else if speed <= 2 {
      BlockSize::BLOCK_8X8
    } else if speed <= 3 {
      BlockSize::BLOCK_16X16
    } else if speed <= 4 {
      BlockSize::BLOCK_32X32
    } else {
      BlockSize::BLOCK_64X64
    }
  }

  fn multiref_preset(speed: usize) -> bool {
    speed <= 2
  }

  fn fast_deblock_preset(speed: usize) -> bool {
    speed >= 4
  }

  fn reduced_tx_set_preset(speed: usize) -> bool {
    speed >= 2
  }

  fn tx_domain_distortion_preset(speed: usize) -> bool {
168
    speed >= 4
169 170 171 172 173 174 175 176 177 178 179
  }

  fn encode_bottomup_preset(speed: usize) -> bool {
    speed == 0
  }

  fn rdo_tx_decision_preset(speed: usize) -> bool {
    speed <= 3
  }

  fn prediction_modes_preset(speed: usize) -> PredictionModesSetting {
180
    if speed <= 1 {
181
      PredictionModesSetting::ComplexAll
182
    } else if speed <= 3 {
183 184 185 186 187 188 189 190 191 192 193
      PredictionModesSetting::ComplexKeyframes
    } else {
      PredictionModesSetting::Simple
    }
  }

  fn include_near_mvs_preset(speed: usize) -> bool {
    speed <= 2
  }
}

194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215
#[allow(dead_code, non_camel_case_types)]
#[derive(Debug, PartialEq, Clone, Copy, Serialize, Deserialize)]
#[repr(C)]
pub enum FrameType {
  KEY,
  INTER,
  INTRA_ONLY,
  SWITCH
}

impl fmt::Display for FrameType {
  fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
    use self::FrameType::*;
    match self {
      KEY => write!(f, "Key frame"),
      INTER => write!(f, "Inter frame"),
      INTRA_ONLY => write!(f, "Intra only frame"),
      SWITCH => write!(f, "Switching frame"),
    }
  }
}

216 217 218 219 220 221 222
#[derive(Clone, Copy, Debug, PartialOrd, PartialEq)]
pub enum PredictionModesSetting {
  Simple,
  ComplexKeyframes,
  ComplexAll,
}

223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264
#[derive(Copy, Clone, Debug, PartialEq)]
#[repr(C)]
pub enum ChromaSampling {
  Cs420,
  Cs422,
  Cs444,
  Cs400,
}

impl Default for ChromaSampling {
  fn default() -> Self {
    ChromaSampling::Cs420
  }
}

impl ChromaSampling {
  // Provides the sampling period in the horizontal and vertical axes.
  pub fn sampling_period(self) -> (usize, usize) {
    use self::ChromaSampling::*;
    match self {
      Cs420 => (2, 2),
      Cs422 => (2, 1),
      Cs444 => (1, 1),
      Cs400 => (2, 2),
    }
  }
}

#[derive(Copy, Clone, Debug, PartialEq)]
#[repr(C)]
pub enum ChromaSamplePosition {
  Unknown,
  Vertical,
  Colocated
}

impl Default for ChromaSamplePosition {
  fn default() -> Self {
    ChromaSamplePosition::Unknown
  }
}

265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280
arg_enum!{
  #[derive(Debug, Clone, Copy, PartialEq)]
  #[repr(C)]
  pub enum PixelRange {
      Unspecified = 0,
      Limited,
      Full,
  }
}

impl Default for PixelRange {
    fn default() -> Self {
        PixelRange::Unspecified
    }
}

281 282
arg_enum!{
  #[derive(Debug, Clone, Copy, PartialEq)]
283
  #[repr(C)]
284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299
  pub enum MatrixCoefficients {
      Identity = 0,
      BT709,
      Unspecified,
      BT470M = 4,
      BT470BG,
      ST170M,
      ST240M,
      YCgCo,
      BT2020NonConstantLuminance,
      BT2020ConstantLuminance,
      ST2085,
      ChromaticityDerivedNonConstantLuminance,
      ChromaticityDerivedConstantLuminance,
      ICtCp,
  }
300 301 302 303 304 305 306 307
}

impl Default for MatrixCoefficients {
    fn default() -> Self {
        MatrixCoefficients::Unspecified
    }
}

308
arg_enum!{
309
  #[derive(Debug, Clone, Copy, PartialEq)]
310
  #[repr(C)]
311 312 313 314 315 316 317 318 319 320 321 322 323 324
  pub enum ColorPrimaries {
      BT709 = 1,
      Unspecified,
      BT470M = 4,
      BT470BG,
      ST170M,
      ST240M,
      Film,
      BT2020,
      ST428,
      P3DCI,
      P3Display,
      Tech3213 = 22,
  }
325 326 327 328 329 330 331 332
}

impl Default for ColorPrimaries {
    fn default() -> Self {
        ColorPrimaries::Unspecified
    }
}

333
arg_enum!{
334
  #[derive(Debug, Clone, Copy, PartialEq)]
335
  #[repr(C)]
336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354
  pub enum TransferCharacteristics {
      BT1886 = 1,
      Unspecified,
      BT470M = 4,
      BT470BG,
      ST170M,
      ST240M,
      Linear,
      Logarithmic100,
      Logarithmic316,
      XVYCC,
      BT1361E,
      SRGB,
      BT2020Ten,
      BT2020Twelve,
      PerceptualQuantizer,
      ST428,
      HybridLogGamma,
  }
355 356 357 358 359 360 361 362
}

impl Default for TransferCharacteristics {
    fn default() -> Self {
        TransferCharacteristics::Unspecified
    }
}

363 364 365 366 367 368 369
#[derive(Copy, Clone, Debug)]
pub struct ColorDescription {
    pub color_primaries: ColorPrimaries,
    pub transfer_characteristics: TransferCharacteristics,
    pub matrix_coefficients: MatrixCoefficients
}

370 371 372 373 374 375 376 377 378 379 380 381 382 383
#[derive(Copy, Clone, Debug)]
pub struct MasteringDisplay {
    pub primaries: [Point; 3],
    pub white_point: Point,
    pub max_luminance: u32,
    pub min_luminance: u32,
}

#[derive(Copy, Clone, Debug)]
pub struct ContentLight {
    pub max_content_light_level: u16,
    pub max_frame_average_light_level: u16,
}

384
/// Contain all the encoder configuration
385
#[derive(Clone, Debug)]
386
pub struct Config {
Luca Barbato's avatar
Luca Barbato committed
387 388 389 390
  pub enc: EncoderConfig
}

impl Config {
Luca Barbato's avatar
Luca Barbato committed
391 392
  pub fn parse(&mut self, key: &str, value: &str) -> Result<(), EncoderStatus> {
    match key {
393
      "low_latency" => self.enc.low_latency = value.parse().map_err(|_e| ParseError)?,
Josh Holmer's avatar
Josh Holmer committed
394 395
      "min_key_frame_interval" => self.enc.min_key_frame_interval = value.parse().map_err(|_e| ParseError)?,
      "key_frame_interval" => self.enc.max_key_frame_interval = value.parse().map_err(|_e| ParseError)?,
396
      "quantizer" => self.enc.quantizer = value.parse().map_err(|_e| ParseError)?,
397
      "speed" => self.enc.speed_settings = SpeedSettings::from_preset(value.parse().map_err(|_e| ParseError)?),
398 399
      "tune" => self.enc.tune = value.parse().map_err(|_e| ParseError)?,
      _ => return Err(InvalidKey)
Luca Barbato's avatar
Luca Barbato committed
400 401 402 403 404
    }

    Ok(())
  }

Luca Barbato's avatar
Luca Barbato committed
405
  pub fn new_context(&self) -> Context {
406
    #[cfg(feature = "aom")]
Luca Barbato's avatar
Luca Barbato committed
407
    unsafe {
Luca Barbato's avatar
Luca Barbato committed
408 409
      av1_rtcd();
      aom_dsp_rtcd();
Luca Barbato's avatar
Luca Barbato committed
410 411
    }

412 413 414
    // initialize with temporal delimiter
    let packet_data = TEMPORAL_DELIMITER.to_vec();

Josh Holmer's avatar
Josh Holmer committed
415 416
    Context {
      frame_count: 0,
417
      limit: 0,
Josh Holmer's avatar
Josh Holmer committed
418
      idx: 0,
Josh Holmer's avatar
Josh Holmer committed
419
      frames_processed: 0,
Josh Holmer's avatar
Josh Holmer committed
420
      frame_q: BTreeMap::new(),
Josh Holmer's avatar
Josh Holmer committed
421 422
      frame_data: BTreeMap::new(),
      keyframes: BTreeSet::new(),
423
      packet_data,
Josh Holmer's avatar
Josh Holmer committed
424 425
      segment_start_idx: 0,
      segment_start_frame: 0,
426
      keyframe_detector: SceneChangeDetector::new(self.enc.bit_depth),
427 428 429 430 431
      config: self.clone(),
      rc_state: RCState::new(),
      first_pass_data: FirstPassData {
        frames: Vec::new(),
      },
Josh Holmer's avatar
Josh Holmer committed
432
    }
Luca Barbato's avatar
Luca Barbato committed
433 434 435 436
  }
}

pub struct Context {
Luca Barbato's avatar
Luca Barbato committed
437
  //    timebase: Rational,
fbossen's avatar
fbossen committed
438
  frame_count: u64,
439
  limit: u64,
fbossen's avatar
fbossen committed
440
  idx: u64,
Josh Holmer's avatar
Josh Holmer committed
441 442
  frames_processed: u64,
  /// Maps frame *number* to frames
443
  frame_q: BTreeMap<u64, Option<Arc<Frame>>>, //    packet_q: VecDeque<Packet>
Josh Holmer's avatar
Josh Holmer committed
444 445 446 447 448
  /// Maps frame *idx* to frame data
  frame_data: BTreeMap<u64, FrameInvariants>,
  /// A list of keyframe *numbers* in this encode. Needed so that we don't
  /// need to keep all of the frame_data in memory for the whole life of the encode.
  keyframes: BTreeSet<u64>,
449
  /// A storage space for reordered frames.
Josh Holmer's avatar
Josh Holmer committed
450 451 452 453
  packet_data: Vec<u8>,
  segment_start_idx: u64,
  segment_start_frame: u64,
  keyframe_detector: SceneChangeDetector,
454
  pub config: Config,
455 456
  rc_state: RCState,
  pub first_pass_data: FirstPassData,
Luca Barbato's avatar
Luca Barbato committed
457 458 459 460 461 462 463 464 465
}

#[derive(Clone, Copy, Debug)]
pub enum EncoderStatus {
  /// The encoder needs more Frames to produce an output Packet
  NeedMoreData,
  /// There are enough Frames queue
  EnoughData,
  ///
Luca Barbato's avatar
Luca Barbato committed
466 467 468
  Failure,
  InvalidKey,
  ParseError
Luca Barbato's avatar
Luca Barbato committed
469 470 471 472
}

pub struct Packet {
  pub data: Vec<u8>,
fbossen's avatar
fbossen committed
473 474
  pub rec: Option<Frame>,
  pub number: u64,
475 476 477
  pub frame_type: FrameType,
  /// PSNR for Y, U, and V planes
  pub psnr: Option<(f64, f64, f64)>,
478 479 480 481
}

impl fmt::Display for Packet {
  fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
Luca Barbato's avatar
Luca Barbato committed
482 483 484 485 486 487 488
    write!(
      f,
      "Frame {} - {} - {} bytes",
      self.number,
      self.frame_type,
      self.data.len()
    )
489
  }
Luca Barbato's avatar
Luca Barbato committed
490 491 492 493
}

impl Context {
  pub fn new_frame(&self) -> Arc<Frame> {
Josh Holmer's avatar
Josh Holmer committed
494
    Arc::new(Frame::new(
495 496
      self.config.enc.width,
      self.config.enc.height,
497
      self.config.enc.chroma_sampling
Josh Holmer's avatar
Josh Holmer committed
498
    ))
Luca Barbato's avatar
Luca Barbato committed
499 500 501 502 503 504
  }

  pub fn send_frame<F>(&mut self, frame: F) -> Result<(), EncoderStatus>
  where
    F: Into<Option<Arc<Frame>>>
  {
Josh Holmer's avatar
Josh Holmer committed
505 506
    let idx = self.frame_count;
    self.frame_q.insert(idx, frame.into());
507
    self.frame_count += 1;
Luca Barbato's avatar
Luca Barbato committed
508 509 510
    Ok(())
  }

511 512 513 514
  pub fn get_frame_count(&self) -> u64 {
    self.frame_count
  }

515 516
  pub fn set_limit(&mut self, limit: u64) {
    self.limit = limit;
517 518
  }

Josh Holmer's avatar
Josh Holmer committed
519 520 521 522
  pub fn needs_more_lookahead(&self) -> bool {
    self.needs_more_frames(self.frame_count) && self.frames_processed + LOOKAHEAD_FRAMES > self.frame_q.keys().last().cloned().unwrap_or(0)
  }

523
  pub fn needs_more_frames(&self, frame_count: u64) -> bool {
524
    self.limit == 0 || frame_count < self.limit
525 526
  }

527 528 529
  pub fn container_sequence_header(&mut self) -> Vec<u8> {
    fn sequence_header_inner(seq: &Sequence) -> io::Result<Vec<u8>> {
      let mut buf = Vec::new();
530

531
      {
532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548
        let mut bw = BitWriter::endian(&mut buf, BigEndian);
        bw.write_bit(true)?; // marker
        bw.write(7, 1)?; // version
        bw.write(3, seq.profile)?;
        bw.write(5, 32)?; // level
        bw.write_bit(false)?; // tier
        bw.write_bit(seq.bit_depth > 8)?; // high_bitdepth
        bw.write_bit(seq.bit_depth == 12)?; // twelve_bit
        bw.write_bit(seq.bit_depth == 1)?; // monochrome
        bw.write_bit(seq.bit_depth == 12)?; // twelve_bit
        bw.write_bit(seq.chroma_sampling != ChromaSampling::Cs444)?; // chroma_subsampling_x
        bw.write_bit(seq.chroma_sampling == ChromaSampling::Cs420)?; // chroma_subsampling_y
        bw.write(2, 0)?; // sample_position
        bw.write(3, 0)?; // reserved
        bw.write_bit(false)?; // initial_presentation_delay_present

        bw.write(4, 0)?; // reserved
549
      }
550

551 552 553
      Ok(buf)
    }

Josh Holmer's avatar
Josh Holmer committed
554
    sequence_header_inner(&self.frame_data[&0].sequence).unwrap()
555 556
  }

Josh Holmer's avatar
Josh Holmer committed
557
  fn next_keyframe(&self) -> u64 {
Josh Holmer's avatar
Josh Holmer committed
558 559 560 561
    let next_detected = self.frame_data.values()
      .find(|fi| fi.frame_type == FrameType::KEY && fi.number > self.segment_start_frame)
      .map(|fi| fi.number);
    let next_limit = self.segment_start_frame + self.config.enc.max_key_frame_interval;
Josh Holmer's avatar
Josh Holmer committed
562 563 564 565 566
    if next_detected.is_none() {
      return next_limit;
    }
    cmp::min(next_detected.unwrap(), next_limit)
  }
fbossen's avatar
fbossen committed
567

568
  fn set_frame_properties(&mut self, idx: u64) -> bool {
569
    let (fi, end_of_subgop) = self.build_frame_properties(idx);
570 571 572
    self.frame_data.insert(idx, fi);

    end_of_subgop
Josh Holmer's avatar
Josh Holmer committed
573 574
  }

575
  fn build_frame_properties(&mut self, idx: u64) -> (FrameInvariants, bool) {
Josh Holmer's avatar
Josh Holmer committed
576
    if idx == 0 {
577
      let seq = Sequence::new(&self.config.enc);
578

Josh Holmer's avatar
Josh Holmer committed
579
      // The first frame will always be a key frame
Josh Holmer's avatar
Josh Holmer committed
580 581
      let fi = FrameInvariants::new_key_frame(
        &FrameInvariants::new(
582
          self.config.enc.clone(),
583
          seq
Josh Holmer's avatar
Josh Holmer committed
584 585 586
        ),
        0
      );
587
      return (fi, true);
Josh Holmer's avatar
Josh Holmer committed
588 589
    }

Josh Holmer's avatar
Josh Holmer committed
590 591
    let mut fi = self.frame_data[&(idx - 1)].clone();

592 593 594 595 596
    // FIXME: inter unsupported with 4:2:2 and 4:4:4 chroma sampling
    let chroma_sampling = self.config.enc.chroma_sampling;
    let keyframe_only = chroma_sampling == ChromaSampling::Cs444 ||
      chroma_sampling == ChromaSampling::Cs422;

Josh Holmer's avatar
Josh Holmer committed
597 598 599 600 601 602
    // Initially set up the frame as an inter frame.
    // We need to determine what the frame number is before we can
    // look up the frame type. If reordering is enabled, the idx
    // may not match the frame number.
    let idx_in_segment = idx - self.segment_start_idx;
    if idx_in_segment > 0 {
603
      let next_keyframe = if keyframe_only { self.segment_start_frame + 1 } else { self.next_keyframe() };
604
      let (fi_temp, end_of_subgop) = FrameInvariants::new_inter_frame(
Josh Holmer's avatar
Josh Holmer committed
605
        &fi,
606 607 608 609
        self.segment_start_frame,
        idx_in_segment,
        next_keyframe
      );
Josh Holmer's avatar
Josh Holmer committed
610
      fi = fi_temp;
611
      if !end_of_subgop {
Josh Holmer's avatar
Josh Holmer committed
612 613 614
        if !fi.inter_cfg.unwrap().reorder
          || ((idx_in_segment - 1) % fi.inter_cfg.unwrap().group_len == 0
          && fi.number == (next_keyframe - 1))
615
        {
Josh Holmer's avatar
Josh Holmer committed
616
          self.segment_start_idx = idx;
Josh Holmer's avatar
Josh Holmer committed
617
          self.segment_start_frame = next_keyframe;
Josh Holmer's avatar
Josh Holmer committed
618
          fi.number = next_keyframe;
Josh Holmer's avatar
Josh Holmer committed
619
        } else {
620
          return (fi, false);
Josh Holmer's avatar
Josh Holmer committed
621 622
        }
      }
fbossen's avatar
fbossen committed
623
    }
Luca Barbato's avatar
Luca Barbato committed
624

625 626
    match self.frame_q.get(&fi.number) {
      Some(Some(_)) => {},
Josh Holmer's avatar
Josh Holmer committed
627
      _ => { return (fi, false); }
628 629
    }

Josh Holmer's avatar
Josh Holmer committed
630
    // Now that we know the frame number, look up the correct frame type
Josh Holmer's avatar
Josh Holmer committed
631 632 633 634 635 636 637
    let frame_type = self.determine_frame_type(fi.number);
    if frame_type == FrameType::KEY {
      self.segment_start_idx = idx;
      self.segment_start_frame = fi.number;
      self.keyframes.insert(fi.number);
    }
    fi.frame_type = frame_type;
Josh Holmer's avatar
Josh Holmer committed
638

Josh Holmer's avatar
Josh Holmer committed
639 640 641 642 643
    let idx_in_segment = idx - self.segment_start_idx;
    if idx_in_segment == 0 {
      fi = FrameInvariants::new_key_frame(&fi, self.segment_start_frame);
    } else {
      let next_keyframe = self.next_keyframe();
644
      let (fi_temp, end_of_subgop) = FrameInvariants::new_inter_frame(
Josh Holmer's avatar
Josh Holmer committed
645 646 647 648 649 650
        &fi,
        self.segment_start_frame,
        idx_in_segment,
        next_keyframe
      );
      fi = fi_temp;
651 652
      if !end_of_subgop {
        return (fi, false);
Josh Holmer's avatar
Josh Holmer committed
653 654
      }
    }
655
    (fi, true)
fbossen's avatar
fbossen committed
656
  }
Luca Barbato's avatar
Luca Barbato committed
657

fbossen's avatar
fbossen committed
658
  pub fn receive_packet(&mut self) -> Result<Packet, EncoderStatus> {
659 660 661 662 663 664
    let idx = {
      let mut idx = self.idx;
      while !self.set_frame_properties(idx) {
        self.idx += 1;
        idx = self.idx;
      }
Luca Barbato's avatar
Luca Barbato committed
665

666 667 668 669 670 671
      if !self.needs_more_frames(self.frame_data.get(&idx).unwrap().number) {
        self.idx += 1;
        return Err(EncoderStatus::EnoughData);
      }
      idx
    };
672

673
    let ret = {
Luca Barbato's avatar
Luca Barbato committed
674 675 676
      let fi = self.frame_data.get_mut(&idx).unwrap();
      if fi.show_existing_frame {
        self.idx += 1;
Luca Barbato's avatar
Luca Barbato committed
677

Luca Barbato's avatar
Luca Barbato committed
678
        let mut fs = FrameState::new(fi);
fbossen's avatar
fbossen committed
679

680 681
        let sef_data = encode_frame(fi, &mut fs);
        self.packet_data.extend(sef_data);
Luca Barbato's avatar
Luca Barbato committed
682

Luca Barbato's avatar
Luca Barbato committed
683
        let rec = if fi.show_frame { Some(fs.rec) } else { None };
684 685
        let fi = fi.clone();
        self.finalize_packet(&*fs.input, rec, &fi)
Luca Barbato's avatar
Luca Barbato committed
686 687 688
      } else {
        if let Some(f) = self.frame_q.get(&fi.number) {
          self.idx += 1;
fbossen's avatar
fbossen committed
689

690
          if let Some(frame) = f.clone() {
691 692 693
            let fti = fi.get_frame_subtype();
            let qps = self.rc_state.select_qi(fi, fti);
            fi.set_quantizers(&qps);
Luca Barbato's avatar
Luca Barbato committed
694
            let mut fs = FrameState::new_with_frame(fi, frame.clone());
fbossen's avatar
fbossen committed
695

Luca Barbato's avatar
Luca Barbato committed
696 697
            let data = encode_frame(fi, &mut fs);
            self.packet_data.extend(data);
fbossen's avatar
fbossen committed
698

Luca Barbato's avatar
Luca Barbato committed
699
            fs.rec.pad(fi.width, fi.height);
fbossen's avatar
fbossen committed
700

Luca Barbato's avatar
Luca Barbato committed
701 702
            // TODO avoid the clone by having rec Arc.
            let rec = if fi.show_frame { Some(fs.rec.clone()) } else { None };
fbossen's avatar
fbossen committed
703

Luca Barbato's avatar
Luca Barbato committed
704
            update_rec_buffer(fi, fs);
fbossen's avatar
fbossen committed
705

Luca Barbato's avatar
Luca Barbato committed
706
            if fi.show_frame {
707 708
              let fi = fi.clone();
              self.finalize_packet(&*frame, rec, &fi)
Luca Barbato's avatar
Luca Barbato committed
709 710 711
            } else {
              Err(EncoderStatus::NeedMoreData)
            }
712 713 714
          } else {
            Err(EncoderStatus::NeedMoreData)
          }
fbossen's avatar
fbossen committed
715 716 717 718
        } else {
          Err(EncoderStatus::NeedMoreData)
        }
      }
719 720 721
    };

    if let Ok(ref pkt) = ret {
Luca Barbato's avatar
Luca Barbato committed
722
      self.garbage_collect(pkt.number);
723 724 725
    }

    ret
Luca Barbato's avatar
Luca Barbato committed
726 727
  }

728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745
  fn finalize_packet(&mut self, original_frame: &Frame, rec: Option<Frame>, fi: &FrameInvariants) -> Result<Packet, EncoderStatus> {
    let data = self.packet_data.clone();
    self.packet_data.clear();
    if write_temporal_delimiter(&mut self.packet_data).is_err() {
      return Err(EncoderStatus::Failure);
    }

    let mut psnr = None;
    if self.config.enc.show_psnr {
      if let Some(ref rec) = rec {
        psnr = Some(calculate_frame_psnr(
          &*original_frame,
          rec,
          fi.sequence.bit_depth
        ));
      }
    }

746 747 748 749
    if self.config.enc.pass == Some(1) {
      self.first_pass_data.frames.push(FirstPassFrame::from(fi));
    }

750 751 752 753 754 755 756 757 758 759
    self.frames_processed += 1;
    Ok(Packet {
      data,
      rec,
      number: fi.number,
      frame_type: fi.frame_type,
      psnr
    })
  }

760
  fn garbage_collect(&mut self, cur_frame: u64) {
Josh Holmer's avatar
Josh Holmer committed
761 762 763 764 765 766 767 768 769 770 771 772 773 774
    if cur_frame == 0 {
      return;
    }
    for i in 0..cur_frame {
      self.frame_q.remove(&i);
    }
    if self.idx < 2 {
      return;
    }
    for i in 0..(self.idx - 1) {
      self.frame_data.remove(&i);
    }
  }

Luca Barbato's avatar
Luca Barbato committed
775
  pub fn flush(&mut self) {
fbossen's avatar
fbossen committed
776
    self.frame_q.insert(self.frame_count, None);
777
    self.frame_count += 1;
Luca Barbato's avatar
Luca Barbato committed
778
  }
Josh Holmer's avatar
Josh Holmer committed
779

Josh Holmer's avatar
Josh Holmer committed
780 781
  fn determine_frame_type(&mut self, frame_number: u64) -> FrameType {
    if frame_number == 0 {
Josh Holmer's avatar
Josh Holmer committed
782 783 784
      return FrameType::KEY;
    }

Josh Holmer's avatar
Josh Holmer committed
785 786 787 788 789 790 791 792
    let prev_keyframe = self.keyframes.iter()
      .rfind(|&&keyframe| keyframe < frame_number)
      .cloned()
      .unwrap_or(0);
    let frame = match self.frame_q.get(&frame_number).cloned() {
      Some(frame) => frame,
      None => { return FrameType::KEY; }
    };
Josh Holmer's avatar
Josh Holmer committed
793
    if let Some(frame) = frame {
Josh Holmer's avatar
Josh Holmer committed
794 795 796
      let distance = frame_number - prev_keyframe;
      if distance < self.config.enc.min_key_frame_interval {
        if distance + 1 == self.config.enc.min_key_frame_interval {
Josh Holmer's avatar
Josh Holmer committed
797 798
          // Run the detector for the current frame, so that it will contain this frame's information
          // to compare against the next frame. We can ignore the results for this frame.
Josh Holmer's avatar
Josh Holmer committed
799
          self.keyframe_detector.detect_scene_change(frame, frame_number as usize);
Josh Holmer's avatar
Josh Holmer committed
800 801 802
        }
        return FrameType::INTER;
      }
Josh Holmer's avatar
Josh Holmer committed
803
      if distance >= self.config.enc.max_key_frame_interval {
Josh Holmer's avatar
Josh Holmer committed
804 805
        return FrameType::KEY;
      }
Josh Holmer's avatar
Josh Holmer committed
806
      if self.keyframe_detector.detect_scene_change(frame, frame_number as usize) {
Josh Holmer's avatar
Josh Holmer committed
807 808 809 810 811
        return FrameType::KEY;
      }
    }
    FrameType::INTER
  }
Luca Barbato's avatar
Luca Barbato committed
812
}
813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831

#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct FirstPassData {
  frames: Vec<FirstPassFrame>,
}

#[derive(Debug, Clone, Copy, Serialize, Deserialize)]
pub struct FirstPassFrame {
  number: u64,
  frame_type: FrameType,
}

impl From<&FrameInvariants> for FirstPassFrame {
  fn from(fi: &FrameInvariants) -> FirstPassFrame {
    FirstPassFrame {
      number: fi.number,
      frame_type: fi.frame_type,
    }
  }
832
}