ffmpeg/libavcodec/proresenc_kostya.c

/*
 * Apple ProRes encoder
 *
 * Copyright (c) 2012 Konstantin Shishkov
 *
 * This encoder appears to be based on Anatoliy Wassermans considering
 * similarities in the bugs.
 *
 * This file is part of Libav.
 *
 * Libav is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * Libav is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with Libav; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

#include "libavutil/opt.h"
#include "avcodec.h"
#include "put_bits.h"
#include "bytestream.h"
#include "internal.h"
#include "proresdsp.h"
#include "proresdata.h"

#define CFACTOR_Y422 2
#define CFACTOR_Y444 3

#define MAX_MBS_PER_SLICE 8

#define MAX_PLANES 3 // should be increased to 4 when there's PIX_FMT_YUV444AP10

enum {
    PRORES_PROFILE_PROXY = 0,
    PRORES_PROFILE_LT,
    PRORES_PROFILE_STANDARD,
    PRORES_PROFILE_HQ,
};

enum {
    QUANT_MAT_PROXY = 0,
    QUANT_MAT_LT,
    QUANT_MAT_STANDARD,
    QUANT_MAT_HQ,
    QUANT_MAT_DEFAULT,
};

static const uint8_t prores_quant_matrices[][64] = {
    { // proxy
         4,  7,  9, 11, 13, 14, 15, 63,
         7,  7, 11, 12, 14, 15, 63, 63,
         9, 11, 13, 14, 15, 63, 63, 63,
        11, 11, 13, 14, 63, 63, 63, 63,
        11, 13, 14, 63, 63, 63, 63, 63,
        13, 14, 63, 63, 63, 63, 63, 63,
        13, 63, 63, 63, 63, 63, 63, 63,
        63, 63, 63, 63, 63, 63, 63, 63,
    },
    { // LT
         4,  5,  6,  7,  9, 11, 13, 15,
         5,  5,  7,  8, 11, 13, 15, 17,
         6,  7,  9, 11, 13, 15, 15, 17,
         7,  7,  9, 11, 13, 15, 17, 19,
         7,  9, 11, 13, 14, 16, 19, 23,
         9, 11, 13, 14, 16, 19, 23, 29,
         9, 11, 13, 15, 17, 21, 28, 35,
        11, 13, 16, 17, 21, 28, 35, 41,
    },
    { // standard
         4,  4,  5,  5,  6,  7,  7,  9,
         4,  4,  5,  6,  7,  7,  9,  9,
         5,  5,  6,  7,  7,  9,  9, 10,
         5,  5,  6,  7,  7,  9,  9, 10,
         5,  6,  7,  7,  8,  9, 10, 12,
         6,  7,  7,  8,  9, 10, 12, 15,
         6,  7,  7,  9, 10, 11, 14, 17,
         7,  7,  9, 10, 11, 14, 17, 21,
    },
    { // high quality
         4,  4,  4,  4,  4,  4,  4,  4,
         4,  4,  4,  4,  4,  4,  4,  4,
         4,  4,  4,  4,  4,  4,  4,  4,
         4,  4,  4,  4,  4,  4,  4,  5,
         4,  4,  4,  4,  4,  4,  5,  5,
         4,  4,  4,  4,  4,  5,  5,  6,
         4,  4,  4,  4,  5,  5,  6,  7,
         4,  4,  4,  4,  5,  6,  7,  7,
    },
    { // codec default
         4,  4,  4,  4,  4,  4,  4,  4,
         4,  4,  4,  4,  4,  4,  4,  4,
         4,  4,  4,  4,  4,  4,  4,  4,
         4,  4,  4,  4,  4,  4,  4,  4,
         4,  4,  4,  4,  4,  4,  4,  4,
         4,  4,  4,  4,  4,  4,  4,  4,
         4,  4,  4,  4,  4,  4,  4,  4,
         4,  4,  4,  4,  4,  4,  4,  4,
    },
};

#define NUM_MB_LIMITS 4
static const int prores_mb_limits[NUM_MB_LIMITS] = {
    1620, // up to 720x576
    2700, // up to 960x720
    6075, // up to 1440x1080
    9216, // up to 2048x1152
};

static const struct prores_profile {
    const char *full_name;
    uint32_t    tag;
    int         min_quant;
    int         max_quant;
    int         br_tab[NUM_MB_LIMITS];
    int         quant;
} prores_profile_info[4] = {
    {
        .full_name = "proxy",
        .tag       = MKTAG('a', 'p', 'c', 'o'),
        .min_quant = 4,
        .max_quant = 8,
        .br_tab    = { 300, 242, 220, 194 },
        .quant     = QUANT_MAT_PROXY,
    },
    {
        .full_name = "LT",
        .tag       = MKTAG('a', 'p', 'c', 's'),
        .min_quant = 1,
        .max_quant = 9,
        .br_tab    = { 720, 560, 490, 440 },
        .quant     = QUANT_MAT_LT,
    },
    {
        .full_name = "standard",
        .tag       = MKTAG('a', 'p', 'c', 'n'),
        .min_quant = 1,
        .max_quant = 6,
        .br_tab    = { 1050, 808, 710, 632 },
        .quant     = QUANT_MAT_STANDARD,
    },
    {
        .full_name = "high quality",
        .tag       = MKTAG('a', 'p', 'c', 'h'),
        .min_quant = 1,
        .max_quant = 6,
        .br_tab    = { 1566, 1216, 1070, 950 },
        .quant     = QUANT_MAT_HQ,
    }
// for 4444 profile bitrate numbers are { 2350, 1828, 1600, 1425 }
};

#define TRELLIS_WIDTH 16
#define SCORE_LIMIT   INT_MAX / 2

struct TrellisNode {
    int prev_node;
    int quant;
    int bits;
    int score;
};

#define MAX_STORED_Q 16

typedef struct ProresThreadData {
    DECLARE_ALIGNED(16, DCTELEM, blocks)[MAX_PLANES][64 * 4 * MAX_MBS_PER_SLICE];
    DECLARE_ALIGNED(16, uint16_t, emu_buf)[16 * 16];
    int16_t custom_q[64];
    struct TrellisNode *nodes;
} ProresThreadData;

typedef struct ProresContext {
    AVClass *class;
    DECLARE_ALIGNED(16, DCTELEM, blocks)[MAX_PLANES][64 * 4 * MAX_MBS_PER_SLICE];
    DECLARE_ALIGNED(16, uint16_t, emu_buf)[16*16];
    int16_t quants[MAX_STORED_Q][64];
    int16_t custom_q[64];
    const uint8_t *quant_mat;

    ProresDSPContext dsp;
    ScanTable  scantable;

    int mb_width, mb_height;
    int mbs_per_slice;
    int num_chroma_blocks, chroma_factor;
    int slices_width;
    int slices_per_picture;
    int pictures_per_frame; // 1 for progressive, 2 for interlaced
    int cur_picture_idx;
    int num_planes;
    int bits_per_mb;
    int force_quant;

    char *vendor;
    int quant_sel;

    int frame_size_upper_bound;

    int profile;
    const struct prores_profile *profile_info;

    int *slice_q;

    ProresThreadData *tdata;
} ProresContext;

static void get_slice_data(ProresContext *ctx, const uint16_t *src,
                           int linesize, int x, int y, int w, int h,
                           DCTELEM *blocks, uint16_t *emu_buf,
                           int mbs_per_slice, int blocks_per_mb, int is_chroma)
{
    const uint16_t *esrc;
    const int mb_width = 4 * blocks_per_mb;
    int elinesize;
    int i, j, k;

    for (i = 0; i < mbs_per_slice; i++, src += mb_width) {
        if (x >= w) {
            memset(blocks, 0, 64 * (mbs_per_slice - i) * blocks_per_mb
                              * sizeof(*blocks));
            return;
        }
        if (x + mb_width <= w && y + 16 <= h) {
            esrc      = src;
            elinesize = linesize;
        } else {
            int bw, bh, pix;

            esrc      = emu_buf;
            elinesize = 16 * sizeof(*emu_buf);

            bw = FFMIN(w - x, mb_width);
            bh = FFMIN(h - y, 16);

            for (j = 0; j < bh; j++) {
                memcpy(emu_buf + j * 16,
                       (const uint8_t*)src + j * linesize,
                       bw * sizeof(*src));
                pix = emu_buf[j * 16 + bw - 1];
                for (k = bw; k < mb_width; k++)
                    emu_buf[j * 16 + k] = pix;
            }
            for (; j < 16; j++)
                memcpy(emu_buf + j * 16,
                       emu_buf + (bh - 1) * 16,
                       mb_width * sizeof(*emu_buf));
        }
        if (!is_chroma) {
            ctx->dsp.fdct(esrc, elinesize, blocks);
            blocks += 64;
            if (blocks_per_mb > 2) {
                ctx->dsp.fdct(esrc + 8, elinesize, blocks);
                blocks += 64;
            }
            ctx->dsp.fdct(esrc + elinesize * 4, elinesize, blocks);
            blocks += 64;
            if (blocks_per_mb > 2) {
                ctx->dsp.fdct(esrc + elinesize * 4 + 8, elinesize, blocks);
                blocks += 64;
            }
        } else {
            ctx->dsp.fdct(esrc, elinesize, blocks);
            blocks += 64;
            ctx->dsp.fdct(esrc + elinesize * 4, elinesize, blocks);
            blocks += 64;
            if (blocks_per_mb > 2) {
                ctx->dsp.fdct(esrc + 8, elinesize, blocks);
                blocks += 64;
                ctx->dsp.fdct(esrc + elinesize * 4 + 8, elinesize, blocks);
                blocks += 64;
            }
        }

        x += mb_width;
    }
}

/**
 * Write an unsigned rice/exp golomb codeword.
 */
static inline void encode_vlc_codeword(PutBitContext *pb, unsigned codebook, int val)
{
    unsigned int rice_order, exp_order, switch_bits, switch_val;
    int exponent;

    /* number of prefix bits to switch between Rice and expGolomb */
    switch_bits = (codebook & 3) + 1;
    rice_order  =  codebook >> 5;       /* rice code order */
    exp_order   = (codebook >> 2) & 7;  /* exp golomb code order */

    switch_val  = switch_bits << rice_order;

    if (val >= switch_val) {
        val -= switch_val - (1 << exp_order);
        exponent = av_log2(val);

        put_bits(pb, exponent - exp_order + switch_bits, 0);
        put_bits(pb, exponent + 1, val);
    } else {
        exponent = val >> rice_order;

        if (exponent)
            put_bits(pb, exponent, 0);
        put_bits(pb, 1, 1);
        if (rice_order)
            put_sbits(pb, rice_order, val);
    }
}

#define GET_SIGN(x)  ((x) >> 31)
#define MAKE_CODE(x) (((x) << 1) ^ GET_SIGN(x))

static void encode_dcs(PutBitContext *pb, DCTELEM *blocks,
                       int blocks_per_slice, int scale)
{
    int i;
    int codebook = 3, code, dc, prev_dc, delta, sign, new_sign;

    prev_dc = (blocks[0] - 0x4000) / scale;
    encode_vlc_codeword(pb, FIRST_DC_CB, MAKE_CODE(prev_dc));
    sign     = 0;
    codebook = 3;
    blocks  += 64;

    for (i = 1; i < blocks_per_slice; i++, blocks += 64) {
        dc       = (blocks[0] - 0x4000) / scale;
        delta    = dc - prev_dc;
        new_sign = GET_SIGN(delta);
        delta    = (delta ^ sign) - sign;
        code     = MAKE_CODE(delta);
        encode_vlc_codeword(pb, ff_prores_dc_codebook[codebook], code);
        codebook = (code + (code & 1)) >> 1;
        codebook = FFMIN(codebook, 3);
        sign     = new_sign;
        prev_dc  = dc;
    }
}

static void encode_acs(PutBitContext *pb, DCTELEM *blocks,
                       int blocks_per_slice,
                       int plane_size_factor,
                       const uint8_t *scan, const int16_t *qmat)
{
    int idx, i;
    int run, level, run_cb, lev_cb;
    int max_coeffs, abs_level;

    max_coeffs = blocks_per_slice << 6;
    run_cb     = ff_prores_run_to_cb_index[4];
    lev_cb     = ff_prores_lev_to_cb_index[2];
    run        = 0;

    for (i = 1; i < 64; i++) {
        for (idx = scan[i]; idx < max_coeffs; idx += 64) {
            level = blocks[idx] / qmat[scan[i]];
            if (level) {
                abs_level = FFABS(level);
                encode_vlc_codeword(pb, ff_prores_ac_codebook[run_cb], run);
                encode_vlc_codeword(pb, ff_prores_ac_codebook[lev_cb],
                                    abs_level - 1);
                put_sbits(pb, 1, GET_SIGN(level));

                run_cb = ff_prores_run_to_cb_index[FFMIN(run, 15)];
                lev_cb = ff_prores_lev_to_cb_index[FFMIN(abs_level, 9)];
                run    = 0;
            } else {
                run++;
            }
        }
    }
}

static int encode_slice_plane(ProresContext *ctx, PutBitContext *pb,
                              const uint16_t *src, int linesize,
                              int mbs_per_slice, DCTELEM *blocks,
                              int blocks_per_mb, int plane_size_factor,
                              const int16_t *qmat)
{
    int blocks_per_slice, saved_pos;

    saved_pos = put_bits_count(pb);
    blocks_per_slice = mbs_per_slice * blocks_per_mb;

    encode_dcs(pb, blocks, blocks_per_slice, qmat[0]);
    encode_acs(pb, blocks, blocks_per_slice, plane_size_factor,
               ctx->scantable.permutated, qmat);
    flush_put_bits(pb);

    return (put_bits_count(pb) - saved_pos) >> 3;
}

static int encode_slice(AVCodecContext *avctx, const AVFrame *pic,
                        PutBitContext *pb,
                        int sizes[4], int x, int y, int quant,
                        int mbs_per_slice)
{
    ProresContext *ctx = avctx->priv_data;
    int i, xp, yp;
    int total_size = 0;
    const uint16_t *src;
    int slice_width_factor = av_log2(mbs_per_slice);
    int num_cblocks, pwidth, linesize, line_add;
    int plane_factor, is_chroma;
    uint16_t *qmat;

    if (ctx->pictures_per_frame == 1)
        line_add = 0;
    else
        line_add = ctx->cur_picture_idx ^ !pic->top_field_first;

    if (ctx->force_quant) {
        qmat = ctx->quants[0];
    } else if (quant < MAX_STORED_Q) {
        qmat = ctx->quants[quant];
    } else {
        qmat = ctx->custom_q;
        for (i = 0; i < 64; i++)
            qmat[i] = ctx->quant_mat[i] * quant;
    }

    for (i = 0; i < ctx->num_planes; i++) {
        is_chroma    = (i == 1 || i == 2);
        plane_factor = slice_width_factor + 2;
        if (is_chroma)
            plane_factor += ctx->chroma_factor - 3;
        if (!is_chroma || ctx->chroma_factor == CFACTOR_Y444) {
            xp          = x << 4;
            yp          = y << 4;
            num_cblocks = 4;
            pwidth      = avctx->width;
        } else {
            xp          = x << 3;
            yp          = y << 4;
            num_cblocks = 2;
            pwidth      = avctx->width >> 1;
        }

        linesize = pic->linesize[i] * ctx->pictures_per_frame;
        src = (const uint16_t*)(pic->data[i] + yp * linesize +
                                line_add * pic->linesize[i]) + xp;

        get_slice_data(ctx, src, linesize, xp, yp,
                       pwidth, avctx->height / ctx->pictures_per_frame,
                       ctx->blocks[0], ctx->emu_buf,
                       mbs_per_slice, num_cblocks, is_chroma);
        sizes[i] = encode_slice_plane(ctx, pb, src, linesize,
                                      mbs_per_slice, ctx->blocks[0],
                                      num_cblocks, plane_factor,
                                      qmat);
        total_size += sizes[i];
    }
    return total_size;
}

static inline int estimate_vlc(unsigned codebook, int val)
{
    unsigned int rice_order, exp_order, switch_bits, switch_val;
    int exponent;

    /* number of prefix bits to switch between Rice and expGolomb */
    switch_bits = (codebook & 3) + 1;
    rice_order  =  codebook >> 5;       /* rice code order */
    exp_order   = (codebook >> 2) & 7;  /* exp golomb code order */

    switch_val  = switch_bits << rice_order;

    if (val >= switch_val) {
        val -= switch_val - (1 << exp_order);
        exponent = av_log2(val);

        return exponent * 2 - exp_order + switch_bits + 1;
    } else {
        return (val >> rice_order) + rice_order + 1;
    }
}

static int estimate_dcs(int *error, DCTELEM *blocks, int blocks_per_slice,
                        int scale)
{
    int i;
    int codebook = 3, code, dc, prev_dc, delta, sign, new_sign;
    int bits;

    prev_dc  = (blocks[0] - 0x4000) / scale;
    bits     = estimate_vlc(FIRST_DC_CB, MAKE_CODE(prev_dc));
    sign     = 0;
    codebook = 3;
    blocks  += 64;
    *error  += FFABS(blocks[0] - 0x4000) % scale;

    for (i = 1; i < blocks_per_slice; i++, blocks += 64) {
        dc       = (blocks[0] - 0x4000) / scale;
        *error  += FFABS(blocks[0] - 0x4000) % scale;
        delta    = dc - prev_dc;
        new_sign = GET_SIGN(delta);
        delta    = (delta ^ sign) - sign;
        code     = MAKE_CODE(delta);
        bits    += estimate_vlc(ff_prores_dc_codebook[codebook], code);
        codebook = (code + (code & 1)) >> 1;
        codebook = FFMIN(codebook, 3);
        sign     = new_sign;
        prev_dc  = dc;
    }

    return bits;
}

static int estimate_acs(int *error, DCTELEM *blocks, int blocks_per_slice,
                        int plane_size_factor,
                        const uint8_t *scan, const int16_t *qmat)
{
    int idx, i;
    int run, level, run_cb, lev_cb;
    int max_coeffs, abs_level;
    int bits = 0;

    max_coeffs = blocks_per_slice << 6;
    run_cb     = ff_prores_run_to_cb_index[4];
    lev_cb     = ff_prores_lev_to_cb_index[2];
    run        = 0;

    for (i = 1; i < 64; i++) {
        for (idx = scan[i]; idx < max_coeffs; idx += 64) {
            level   = blocks[idx] / qmat[scan[i]];
            *error += FFABS(blocks[idx]) % qmat[scan[i]];
            if (level) {
                abs_level = FFABS(level);
                bits += estimate_vlc(ff_prores_ac_codebook[run_cb], run);
                bits += estimate_vlc(ff_prores_ac_codebook[lev_cb],
                                     abs_level - 1) + 1;

                run_cb = ff_prores_run_to_cb_index[FFMIN(run, 15)];
                lev_cb = ff_prores_lev_to_cb_index[FFMIN(abs_level, 9)];
                run    = 0;
            } else {
                run++;
            }
        }
    }

    return bits;
}

static int estimate_slice_plane(ProresContext *ctx, int *error, int plane,
                                const uint16_t *src, int linesize,
                                int mbs_per_slice,
                                int blocks_per_mb, int plane_size_factor,
                                const int16_t *qmat, ProresThreadData *td)
{
    int blocks_per_slice;
    int bits;

    blocks_per_slice = mbs_per_slice * blocks_per_mb;

    bits  = estimate_dcs(error, td->blocks[plane], blocks_per_slice, qmat[0]);
    bits += estimate_acs(error, td->blocks[plane], blocks_per_slice,
                         plane_size_factor, ctx->scantable.permutated, qmat);

    return FFALIGN(bits, 8);
}

static int find_slice_quant(AVCodecContext *avctx, const AVFrame *pic,
                            int trellis_node, int x, int y, int mbs_per_slice,
                            ProresThreadData *td)
{
    ProresContext *ctx = avctx->priv_data;
    int i, q, pq, xp, yp;
    const uint16_t *src;
    int slice_width_factor = av_log2(mbs_per_slice);
    int num_cblocks[MAX_PLANES], pwidth;
    int plane_factor[MAX_PLANES], is_chroma[MAX_PLANES];
    const int min_quant = ctx->profile_info->min_quant;
    const int max_quant = ctx->profile_info->max_quant;
    int error, bits, bits_limit;
    int mbs, prev, cur, new_score;
    int slice_bits[TRELLIS_WIDTH], slice_score[TRELLIS_WIDTH];
    int overquant;
    uint16_t *qmat;
    int linesize[4], line_add;

    if (ctx->pictures_per_frame == 1)
        line_add = 0;
    else
        line_add = ctx->cur_picture_idx ^ !pic->top_field_first;
    mbs = x + mbs_per_slice;

    for (i = 0; i < ctx->num_planes; i++) {
        is_chroma[i]    = (i == 1 || i == 2);
        plane_factor[i] = slice_width_factor + 2;
        if (is_chroma[i])
            plane_factor[i] += ctx->chroma_factor - 3;
        if (!is_chroma[i] || ctx->chroma_factor == CFACTOR_Y444) {
            xp             = x << 4;
            yp             = y << 4;
            num_cblocks[i] = 4;
            pwidth         = avctx->width;
        } else {
            xp             = x << 3;
            yp             = y << 4;
            num_cblocks[i] = 2;
            pwidth         = avctx->width >> 1;
        }

        linesize[i] = pic->linesize[i] * ctx->pictures_per_frame;
        src = (const uint16_t*)(pic->data[i] + yp * linesize[i] +
                                line_add * pic->linesize[i]) + xp;

        get_slice_data(ctx, src, linesize[i], xp, yp,
                       pwidth, avctx->height / ctx->pictures_per_frame,
                       td->blocks[i], td->emu_buf,
                       mbs_per_slice, num_cblocks[i], is_chroma[i]);
    }

    for (q = min_quant; q < max_quant + 2; q++) {
        td->nodes[trellis_node + q].prev_node = -1;
        td->nodes[trellis_node + q].quant     = q;
    }

    // todo: maybe perform coarser quantising to fit into frame size when needed
    for (q = min_quant; q <= max_quant; q++) {
        bits  = 0;
        error = 0;
        for (i = 0; i < ctx->num_planes; i++) {
            bits += estimate_slice_plane(ctx, &error, i,
                                         src, linesize[i],
                                         mbs_per_slice,
                                         num_cblocks[i], plane_factor[i],
                                         ctx->quants[q], td);
        }
        if (bits > 65000 * 8) {
            error = SCORE_LIMIT;
            break;
        }
        slice_bits[q]  = bits;
        slice_score[q] = error;
    }
    if (slice_bits[max_quant] <= ctx->bits_per_mb * mbs_per_slice) {
        slice_bits[max_quant + 1]  = slice_bits[max_quant];
        slice_score[max_quant + 1] = slice_score[max_quant] + 1;
        overquant = max_quant;
    } else {
        for (q = max_quant + 1; q < 128; q++) {
            bits  = 0;
            error = 0;
            if (q < MAX_STORED_Q) {
                qmat = ctx->quants[q];
            } else {
                qmat = td->custom_q;
                for (i = 0; i < 64; i++)
                    qmat[i] = ctx->quant_mat[i] * q;
            }
            for (i = 0; i < ctx->num_planes; i++) {
                bits += estimate_slice_plane(ctx, &error, i,
                                             src, linesize[i],
                                             mbs_per_slice,
                                             num_cblocks[i], plane_factor[i],
                                             qmat, td);
            }
            if (bits <= ctx->bits_per_mb * mbs_per_slice)
                break;
        }

        slice_bits[max_quant + 1]  = bits;
        slice_score[max_quant + 1] = error;
        overquant = q;
    }
    td->nodes[trellis_node + max_quant + 1].quant = overquant;

    bits_limit = mbs * ctx->bits_per_mb;
    for (pq = min_quant; pq < max_quant + 2; pq++) {
        prev = trellis_node - TRELLIS_WIDTH + pq;

        for (q = min_quant; q < max_quant + 2; q++) {
            cur = trellis_node + q;

            bits  = td->nodes[prev].bits + slice_bits[q];
            error = slice_score[q];
            if (bits > bits_limit)
                error = SCORE_LIMIT;

            if (td->nodes[prev].score < SCORE_LIMIT && error < SCORE_LIMIT)
                new_score = td->nodes[prev].score + error;
            else
                new_score = SCORE_LIMIT;
            if (td->nodes[cur].prev_node == -1 ||
                td->nodes[cur].score >= new_score) {

                td->nodes[cur].bits      = bits;
                td->nodes[cur].score     = new_score;
                td->nodes[cur].prev_node = prev;
            }
        }
    }

    error = td->nodes[trellis_node + min_quant].score;
    pq    = trellis_node + min_quant;
    for (q = min_quant + 1; q < max_quant + 2; q++) {
        if (td->nodes[trellis_node + q].score <= error) {
            error = td->nodes[trellis_node + q].score;
            pq    = trellis_node + q;
        }
    }

    return pq;
}

static int find_quant_thread(AVCodecContext *avctx, void *arg,
                             int jobnr, int threadnr)
{
    ProresContext *ctx = avctx->priv_data;
    ProresThreadData *td = ctx->tdata + threadnr;
    int mbs_per_slice = ctx->mbs_per_slice;
    int x, y = jobnr, mb, q = 0;

    for (x = mb = 0; x < ctx->mb_width; x += mbs_per_slice, mb++) {
        while (ctx->mb_width - x < mbs_per_slice)
            mbs_per_slice >>= 1;
        q = find_slice_quant(avctx, avctx->coded_frame,
                             (mb + 1) * TRELLIS_WIDTH, x, y,
                             mbs_per_slice, td);
    }

    for (x = ctx->slices_width - 1; x >= 0; x--) {
        ctx->slice_q[x + y * ctx->slices_width] = td->nodes[q].quant;
        q = td->nodes[q].prev_node;
    }

    return 0;
}

static int encode_frame(AVCodecContext *avctx, AVPacket *pkt,
                        const AVFrame *pic, int *got_packet)
{
    ProresContext *ctx = avctx->priv_data;
    uint8_t *orig_buf, *buf, *slice_hdr, *slice_sizes, *tmp;
    uint8_t *picture_size_pos;
    PutBitContext pb;
    int x, y, i, mb, q = 0;
    int sizes[4] = { 0 };
    int slice_hdr_size = 2 + 2 * (ctx->num_planes - 1);
    int frame_size, picture_size, slice_size;
    int pkt_size, ret;
    uint8_t frame_flags;

    *avctx->coded_frame           = *pic;
    avctx->coded_frame->pict_type = AV_PICTURE_TYPE_I;
    avctx->coded_frame->key_frame = 1;

    pkt_size = ctx->frame_size_upper_bound + FF_MIN_BUFFER_SIZE;

    if ((ret = ff_alloc_packet2(avctx, pkt, pkt_size)) < 0)
        return ret;

    orig_buf = pkt->data;

    // frame atom
    orig_buf += 4;                              // frame size
    bytestream_put_be32  (&orig_buf, FRAME_ID); // frame container ID
    buf = orig_buf;

    // frame header
    tmp = buf;
    buf += 2;                                   // frame header size will be stored here
    bytestream_put_be16  (&buf, 0);             // version 1
    bytestream_put_buffer(&buf, ctx->vendor, 4);
    bytestream_put_be16  (&buf, avctx->width);
    bytestream_put_be16  (&buf, avctx->height);

    frame_flags = ctx->chroma_factor << 6;
    if (avctx->flags & CODEC_FLAG_INTERLACED_DCT)
        frame_flags |= pic->top_field_first ? 0x04 : 0x08;
    bytestream_put_byte  (&buf, frame_flags);

    bytestream_put_byte  (&buf, 0);             // reserved
    bytestream_put_byte  (&buf, avctx->color_primaries);
    bytestream_put_byte  (&buf, avctx->color_trc);
    bytestream_put_byte  (&buf, avctx->colorspace);
    bytestream_put_byte  (&buf, 0x40);          // source format and alpha information
    bytestream_put_byte  (&buf, 0);             // reserved
    if (ctx->quant_sel != QUANT_MAT_DEFAULT) {
        bytestream_put_byte  (&buf, 0x03);      // matrix flags - both matrices are present
        // luma quantisation matrix
        for (i = 0; i < 64; i++)
            bytestream_put_byte(&buf, ctx->quant_mat[i]);
        // chroma quantisation matrix
        for (i = 0; i < 64; i++)
            bytestream_put_byte(&buf, ctx->quant_mat[i]);
    } else {
        bytestream_put_byte  (&buf, 0x00);      // matrix flags - default matrices are used
    }
    bytestream_put_be16  (&tmp, buf - orig_buf); // write back frame header size

    for (ctx->cur_picture_idx = 0;
         ctx->cur_picture_idx < ctx->pictures_per_frame;
         ctx->cur_picture_idx++) {
        // picture header
        picture_size_pos = buf + 1;
        bytestream_put_byte  (&buf, 0x40);          // picture header size (in bits)
        buf += 4;                                   // picture data size will be stored here
        bytestream_put_be16  (&buf, ctx->slices_per_picture);
        bytestream_put_byte  (&buf, av_log2(ctx->mbs_per_slice) << 4); // slice width and height in MBs

        // seek table - will be filled during slice encoding
        slice_sizes = buf;
        buf += ctx->slices_per_picture * 2;

        // slices
        if (!ctx->force_quant) {
            ret = avctx->execute2(avctx, find_quant_thread, NULL, NULL,
                                  ctx->mb_height);
            if (ret)
                return ret;
        }

        for (y = 0; y < ctx->mb_height; y++) {
            int mbs_per_slice = ctx->mbs_per_slice;
            for (x = mb = 0; x < ctx->mb_width; x += mbs_per_slice, mb++) {
                q = ctx->force_quant ? ctx->force_quant
                                     : ctx->slice_q[mb + y * ctx->slices_width];

                while (ctx->mb_width - x < mbs_per_slice)
                    mbs_per_slice >>= 1;

                bytestream_put_byte(&buf, slice_hdr_size << 3);
                slice_hdr = buf;
                buf += slice_hdr_size - 1;
                init_put_bits(&pb, buf, (pkt_size - (buf - orig_buf)) * 8);
                encode_slice(avctx, pic, &pb, sizes, x, y, q, mbs_per_slice);

                bytestream_put_byte(&slice_hdr, q);
                slice_size = slice_hdr_size + sizes[ctx->num_planes - 1];
                for (i = 0; i < ctx->num_planes - 1; i++) {
                    bytestream_put_be16(&slice_hdr, sizes[i]);
                    slice_size += sizes[i];
                }
                bytestream_put_be16(&slice_sizes, slice_size);
                buf += slice_size - slice_hdr_size;
            }
        }

        picture_size = buf - (picture_size_pos - 1);
        bytestream_put_be32(&picture_size_pos, picture_size);
    }

    orig_buf -= 8;
    frame_size = buf - orig_buf;
    bytestream_put_be32(&orig_buf, frame_size);

    pkt->size   = frame_size;
    pkt->flags |= AV_PKT_FLAG_KEY;
    *got_packet = 1;

    return 0;
}

static av_cold int encode_close(AVCodecContext *avctx)
{
    ProresContext *ctx = avctx->priv_data;
    int i;

    av_freep(&avctx->coded_frame);

    if (ctx->tdata) {
        for (i = 0; i < avctx->thread_count; i++)
            av_free(ctx->tdata[i].nodes);
    }
    av_freep(&ctx->tdata);
    av_freep(&ctx->slice_q);

    return 0;
}

static av_cold int encode_init(AVCodecContext *avctx)
{
    ProresContext *ctx = avctx->priv_data;
    int mps;
    int i, j;
    int min_quant, max_quant;
    int interlaced = !!(avctx->flags & CODEC_FLAG_INTERLACED_DCT);

    avctx->bits_per_raw_sample = 10;
    avctx->coded_frame = avcodec_alloc_frame();
    if (!avctx->coded_frame)
        return AVERROR(ENOMEM);

    ff_proresdsp_init(&ctx->dsp, avctx);
    ff_init_scantable(ctx->dsp.dct_permutation, &ctx->scantable,
                      interlaced ? ff_prores_interlaced_scan
                                 : ff_prores_progressive_scan);

    mps = ctx->mbs_per_slice;
    if (mps & (mps - 1)) {
        av_log(avctx, AV_LOG_ERROR,
               "there should be an integer power of two MBs per slice\n");
        return AVERROR(EINVAL);
    }

    ctx->chroma_factor = avctx->pix_fmt == PIX_FMT_YUV422P10
                         ? CFACTOR_Y422
                         : CFACTOR_Y444;
    ctx->profile_info  = prores_profile_info + ctx->profile;
    ctx->num_planes    = 3;

    ctx->mb_width      = FFALIGN(avctx->width,  16) >> 4;

    if (interlaced)
        ctx->mb_height = FFALIGN(avctx->height, 32) >> 5;
    else
        ctx->mb_height = FFALIGN(avctx->height, 16) >> 4;

    ctx->slices_width  = ctx->mb_width / mps;
    ctx->slices_width += av_popcount(ctx->mb_width - ctx->slices_width * mps);
    ctx->slices_per_picture = ctx->mb_height * ctx->slices_width;
    ctx->pictures_per_frame = 1 + interlaced;

    if (ctx->quant_sel == -1)
        ctx->quant_mat = prores_quant_matrices[ctx->profile_info->quant];
    else
        ctx->quant_mat = prores_quant_matrices[ctx->quant_sel];

    if (strlen(ctx->vendor) != 4) {
        av_log(avctx, AV_LOG_ERROR, "vendor ID should be 4 bytes\n");
        return AVERROR_INVALIDDATA;
    }

    ctx->force_quant = avctx->global_quality / FF_QP2LAMBDA;
    if (!ctx->force_quant) {
        if (!ctx->bits_per_mb) {
            for (i = 0; i < NUM_MB_LIMITS - 1; i++)
                if (prores_mb_limits[i] >= ctx->mb_width * ctx->mb_height *
                                           ctx->pictures_per_frame)
                    break;
            ctx->bits_per_mb   = ctx->profile_info->br_tab[i];
        } else if (ctx->bits_per_mb < 128) {
            av_log(avctx, AV_LOG_ERROR, "too few bits per MB, please set at least 128\n");
            return AVERROR_INVALIDDATA;
        }

        min_quant = ctx->profile_info->min_quant;
        max_quant = ctx->profile_info->max_quant;
        for (i = min_quant; i < MAX_STORED_Q; i++) {
            for (j = 0; j < 64; j++)
                ctx->quants[i][j] = ctx->quant_mat[j] * i;
        }

        ctx->slice_q = av_malloc(ctx->slices_per_picture * sizeof(*ctx->slice_q));
        if (!ctx->slice_q) {
            encode_close(avctx);
            return AVERROR(ENOMEM);
        }

        ctx->tdata = av_mallocz(avctx->thread_count * sizeof(*ctx->tdata));
        if (!ctx->tdata) {
            encode_close(avctx);
            return AVERROR(ENOMEM);
        }

        for (j = 0; j < avctx->thread_count; j++) {
            ctx->tdata[j].nodes = av_malloc((ctx->slices_width + 1)
                                            * TRELLIS_WIDTH
                                            * sizeof(*ctx->tdata->nodes));
            if (!ctx->tdata[j].nodes) {
                encode_close(avctx);
                return AVERROR(ENOMEM);
            }
            for (i = min_quant; i < max_quant + 2; i++) {
                ctx->tdata[j].nodes[i].prev_node = -1;
                ctx->tdata[j].nodes[i].bits      = 0;
                ctx->tdata[j].nodes[i].score     = 0;
            }
        }
    } else {
        int ls = 0;

        if (ctx->force_quant > 64) {
            av_log(avctx, AV_LOG_ERROR, "too large quantiser, maximum is 64\n");
            return AVERROR_INVALIDDATA;
        }

        for (j = 0; j < 64; j++) {
            ctx->quants[0][j] = ctx->quant_mat[j] * ctx->force_quant;
            ls += av_log2((1 << 11)  / ctx->quants[0][j]) * 2 + 1;
        }

        ctx->bits_per_mb = ls * 8;
        if (ctx->chroma_factor == CFACTOR_Y444)
            ctx->bits_per_mb += ls * 4;
        if (ctx->num_planes == 4)
            ctx->bits_per_mb += ls * 4;
    }

    ctx->frame_size_upper_bound = ctx->pictures_per_frame *
                                  ctx->slices_per_picture *
                                  (2 + 2 * ctx->num_planes +
                                   (mps * ctx->bits_per_mb) / 8)
                                  + 200;

    avctx->codec_tag   = ctx->profile_info->tag;

    av_log(avctx, AV_LOG_DEBUG,
           "profile %d, %d slices, interlacing: %s, %d bits per MB\n",
           ctx->profile, ctx->slices_per_picture * ctx->pictures_per_frame,
           interlaced ? "yes" : "no", ctx->bits_per_mb);
    av_log(avctx, AV_LOG_DEBUG, "frame size upper bound: %d\n",
           ctx->frame_size_upper_bound);

    return 0;
}

#define OFFSET(x) offsetof(ProresContext, x)
#define VE     AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM

static const AVOption options[] = {
    { "mbs_per_slice", "macroblocks per slice", OFFSET(mbs_per_slice),
        AV_OPT_TYPE_INT, { .i64 = 8 }, 1, MAX_MBS_PER_SLICE, VE },
    { "profile",       NULL, OFFSET(profile), AV_OPT_TYPE_INT,
        { .i64 = PRORES_PROFILE_STANDARD },
        PRORES_PROFILE_PROXY, PRORES_PROFILE_HQ, VE, "profile" },
    { "proxy",         NULL, 0, AV_OPT_TYPE_CONST, { .i64 = PRORES_PROFILE_PROXY },
        0, 0, VE, "profile" },
    { "lt",            NULL, 0, AV_OPT_TYPE_CONST, { .i64 = PRORES_PROFILE_LT },
        0, 0, VE, "profile" },
    { "standard",      NULL, 0, AV_OPT_TYPE_CONST, { .i64 = PRORES_PROFILE_STANDARD },
        0, 0, VE, "profile" },
    { "hq",            NULL, 0, AV_OPT_TYPE_CONST, { .i64 = PRORES_PROFILE_HQ },
        0, 0, VE, "profile" },
    { "vendor", "vendor ID", OFFSET(vendor),
        AV_OPT_TYPE_STRING, { .str = "Lavc" }, CHAR_MIN, CHAR_MAX, VE },
    { "bits_per_mb", "desired bits per macroblock", OFFSET(bits_per_mb),
        AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 8192, VE },
    { "quant_mat", "quantiser matrix", OFFSET(quant_sel), AV_OPT_TYPE_INT,
        { .i64 = -1 }, -1, QUANT_MAT_DEFAULT, VE, "quant_mat" },
    { "auto",          NULL, 0, AV_OPT_TYPE_CONST, { .i64 = -1 },
        0, 0, VE, "quant_mat" },
    { "proxy",         NULL, 0, AV_OPT_TYPE_CONST, { .i64 = QUANT_MAT_PROXY },
        0, 0, VE, "quant_mat" },
    { "lt",            NULL, 0, AV_OPT_TYPE_CONST, { .i64 = QUANT_MAT_LT },
        0, 0, VE, "quant_mat" },
    { "standard",      NULL, 0, AV_OPT_TYPE_CONST, { .i64 = QUANT_MAT_STANDARD },
        0, 0, VE, "quant_mat" },
    { "hq",            NULL, 0, AV_OPT_TYPE_CONST, { .i64 = QUANT_MAT_HQ },
        0, 0, VE, "quant_mat" },
    { "default",       NULL, 0, AV_OPT_TYPE_CONST, { .i64 = QUANT_MAT_DEFAULT },
        0, 0, VE, "quant_mat" },
    { NULL }
};

static const AVClass proresenc_class = {
    .class_name = "ProRes encoder",
    .item_name  = av_default_item_name,
    .option     = options,
    .version    = LIBAVUTIL_VERSION_INT,
};

AVCodec ff_prores_kostya_encoder = {
    .name           = "prores_kostya",
    .type           = AVMEDIA_TYPE_VIDEO,
    .id             = AV_CODEC_ID_PRORES,
    .priv_data_size = sizeof(ProresContext),
    .init           = encode_init,
    .close          = encode_close,
    .encode2        = encode_frame,
    .capabilities   = CODEC_CAP_SLICE_THREADS,
    .long_name      = NULL_IF_CONFIG_SMALL("Apple ProRes (iCodec Pro)"),
    .pix_fmts       = (const enum PixelFormat[]) {
                          PIX_FMT_YUV422P10, PIX_FMT_YUV444P10, PIX_FMT_NONE
                      },
    .priv_class     = &proresenc_class,
};