xref: /third_party/ffmpeg/libavcodec/libjxlenc.c (revision cabdff1a) 
1/*
2 * JPEG XL encoding support via libjxl
3 * Copyright (c) 2021 Leo Izen <leo.izen@gmail.com>
4 *
5 * This file is part of FFmpeg.
6 *
7 * FFmpeg is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
11 *
12 * FFmpeg is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
15 * Lesser General Public License for more details.
16 *
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with FFmpeg; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20 */
21
22/**
23 * @file
24 * JPEG XL encoder using libjxl
25 */
26
27#include <string.h>
28
29#include "libavutil/avutil.h"
30#include "libavutil/csp.h"
31#include "libavutil/error.h"
32#include "libavutil/frame.h"
33#include "libavutil/libm.h"
34#include "libavutil/opt.h"
35#include "libavutil/pixdesc.h"
36#include "libavutil/pixfmt.h"
37#include "libavutil/version.h"
38
39#include "avcodec.h"
40#include "encode.h"
41#include "codec_internal.h"
42
43#include <jxl/encode.h>
44#include <jxl/thread_parallel_runner.h>
45#include "libjxl.h"
46
47typedef struct LibJxlEncodeContext {
48    AVClass *class;
49    void *runner;
50    JxlEncoder *encoder;
51    JxlEncoderFrameSettings *options;
52    int effort;
53    float distance;
54    int modular;
55    uint8_t *buffer;
56    size_t buffer_size;
57} LibJxlEncodeContext;
58
59/**
60 * Map a quality setting for -qscale roughly from libjpeg
61 * quality numbers to libjxl's butteraugli distance for
62 * photographic content.
63 *
64 * Setting distance explicitly is preferred, but this will
65 * allow qscale to be used as a fallback.
66 *
67 * This function is continuous and injective on [0, 100] which
68 * makes it monotonic.
69 *
70 * @param  quality 0.0 to 100.0 quality setting, libjpeg quality
71 * @return         Butteraugli distance between 0.0 and 15.0
72 */
73static float quality_to_distance(float quality)
74{
75    if (quality >= 100.0)
76        return 0.0;
77    else if (quality >= 90.0)
78        return (100.0 - quality) * 0.10;
79    else if (quality >= 30.0)
80        return 0.1 + (100.0 - quality) * 0.09;
81    else if (quality > 0.0)
82        return 15.0 + (59.0 * quality - 4350.0) * quality / 9000.0;
83    else
84        return 15.0;
85}
86
87/**
88 * Initalize the decoder on a per-frame basis. All of these need to be set
89 * once each time the decoder is reset, which it must be each frame to make
90 * the image2 muxer work.
91 *
92 * @return       0 upon success, negative on failure.
93 */
94static int libjxl_init_jxl_encoder(AVCodecContext *avctx)
95{
96    LibJxlEncodeContext *ctx = avctx->priv_data;
97
98    /* reset the encoder every frame for image2 muxer */
99    JxlEncoderReset(ctx->encoder);
100
101    ctx->options = JxlEncoderFrameSettingsCreate(ctx->encoder, NULL);
102    if (!ctx->options) {
103        av_log(avctx, AV_LOG_ERROR, "Failed to create JxlEncoderOptions\n");
104        return AVERROR_EXTERNAL;
105    }
106
107    /* This needs to be set each time the decoder is reset */
108    if (JxlEncoderSetParallelRunner(ctx->encoder, JxlThreadParallelRunner, ctx->runner)
109            != JXL_ENC_SUCCESS) {
110        av_log(avctx, AV_LOG_ERROR, "Failed to set JxlThreadParallelRunner\n");
111        return AVERROR_EXTERNAL;
112    }
113
114    /* these shouldn't fail, libjxl bug notwithstanding */
115    if (JxlEncoderFrameSettingsSetOption(ctx->options, JXL_ENC_FRAME_SETTING_EFFORT, ctx->effort)
116            != JXL_ENC_SUCCESS) {
117        av_log(avctx, AV_LOG_ERROR, "Failed to set effort to: %d\n", ctx->effort);
118        return AVERROR_EXTERNAL;
119    }
120
121    /* check for negative, our default */
122    if (ctx->distance < 0.0) {
123        /* use ffmpeg.c -q option if passed */
124        if (avctx->flags & AV_CODEC_FLAG_QSCALE)
125            ctx->distance = quality_to_distance((float)avctx->global_quality / FF_QP2LAMBDA);
126        else
127            /* default 1.0 matches cjxl */
128            ctx->distance = 1.0;
129    }
130
131    /*
132     * 0.01 is the minimum distance accepted for lossy
133     * interpreting any positive value less than this as minimum
134     */
135    if (ctx->distance > 0.0 && ctx->distance < 0.01)
136        ctx->distance = 0.01;
137    if (JxlEncoderSetFrameDistance(ctx->options, ctx->distance) != JXL_ENC_SUCCESS) {
138        av_log(avctx, AV_LOG_ERROR, "Failed to set distance: %f\n", ctx->distance);
139        return AVERROR_EXTERNAL;
140    }
141
142    /*
143     * In theory the library should automatically enable modular if necessary,
144     * but it appears it won't at the moment due to a bug. This will still
145     * work even if that is patched.
146     */
147    if (JxlEncoderFrameSettingsSetOption(ctx->options, JXL_ENC_FRAME_SETTING_MODULAR,
148            ctx->modular || ctx->distance <= 0.0 ? 1 : -1) != JXL_ENC_SUCCESS) {
149        av_log(avctx, AV_LOG_ERROR, "Failed to set modular\n");
150        return AVERROR_EXTERNAL;
151    }
152
153    return 0;
154}
155
156/**
157 * Global encoder initialization. This only needs to be run once,
158 * not every frame.
159 */
160static av_cold int libjxl_encode_init(AVCodecContext *avctx)
161{
162    LibJxlEncodeContext *ctx = avctx->priv_data;
163    JxlMemoryManager manager;
164
165    ff_libjxl_init_memory_manager(&manager);
166    ctx->encoder = JxlEncoderCreate(&manager);
167    if (!ctx->encoder) {
168        av_log(avctx, AV_LOG_ERROR, "Failed to create JxlEncoder\n");
169        return AVERROR_EXTERNAL;
170    }
171
172    ctx->runner = JxlThreadParallelRunnerCreate(&manager, ff_libjxl_get_threadcount(avctx->thread_count));
173    if (!ctx->runner) {
174        av_log(avctx, AV_LOG_ERROR, "Failed to create JxlThreadParallelRunner\n");
175        return AVERROR_EXTERNAL;
176    }
177
178    ctx->buffer_size = 4096;
179    ctx->buffer = av_realloc(NULL, ctx->buffer_size);
180
181    if (!ctx->buffer) {
182        av_log(avctx, AV_LOG_ERROR, "Could not allocate encoding buffer\n");
183        return AVERROR(ENOMEM);
184    }
185
186    return 0;
187}
188
189/**
190 * Populate a JxlColorEncoding with the given enum AVColorPrimaries.
191 * @return < 0 upon failure, >= 0 upon success
192 */
193static int libjxl_populate_primaries(void *avctx, JxlColorEncoding *jxl_color, enum AVColorPrimaries prm)
194{
195    const AVColorPrimariesDesc *desc;
196
197    switch (prm) {
198    case AVCOL_PRI_BT709:
199        jxl_color->primaries = JXL_PRIMARIES_SRGB;
200        jxl_color->white_point = JXL_WHITE_POINT_D65;
201        return 0;
202    case AVCOL_PRI_BT2020:
203        jxl_color->primaries = JXL_PRIMARIES_2100;
204        jxl_color->white_point = JXL_WHITE_POINT_D65;
205        return 0;
206    case AVCOL_PRI_SMPTE431:
207        jxl_color->primaries = JXL_PRIMARIES_P3;
208        jxl_color->white_point = JXL_WHITE_POINT_DCI;
209        return 0;
210    case AVCOL_PRI_SMPTE432:
211        jxl_color->primaries = JXL_PRIMARIES_P3;
212        jxl_color->white_point = JXL_WHITE_POINT_D65;
213        return 0;
214    case AVCOL_PRI_UNSPECIFIED:
215        av_log(avctx, AV_LOG_WARNING, "Unknown primaries, assuming BT.709/sRGB. Colors may be wrong.\n");
216        jxl_color->primaries = JXL_PRIMARIES_SRGB;
217        jxl_color->white_point = JXL_WHITE_POINT_D65;
218        return 0;
219    }
220
221    desc = av_csp_primaries_desc_from_id(prm);
222    if (!desc)
223        return AVERROR(EINVAL);
224
225    jxl_color->primaries = JXL_PRIMARIES_CUSTOM;
226    jxl_color->white_point = JXL_WHITE_POINT_CUSTOM;
227
228    jxl_color->primaries_red_xy[0] = av_q2d(desc->prim.r.x);
229    jxl_color->primaries_red_xy[1] = av_q2d(desc->prim.r.y);
230    jxl_color->primaries_green_xy[0] = av_q2d(desc->prim.g.x);
231    jxl_color->primaries_green_xy[1] = av_q2d(desc->prim.g.y);
232    jxl_color->primaries_blue_xy[0] = av_q2d(desc->prim.b.x);
233    jxl_color->primaries_blue_xy[1] = av_q2d(desc->prim.b.y);
234    jxl_color->white_point_xy[0] = av_q2d(desc->wp.x);
235    jxl_color->white_point_xy[1] = av_q2d(desc->wp.y);
236
237    return 0;
238}
239
240/**
241 * Encode an entire frame. Currently animation, is not supported by
242 * this encoder, so this will always reinitialize a new still image
243 * and encode a one-frame image (for image2 and image2pipe).
244 */
245static int libjxl_encode_frame(AVCodecContext *avctx, AVPacket *pkt, const AVFrame *frame, int *got_packet)
246{
247    LibJxlEncodeContext *ctx = avctx->priv_data;
248    AVFrameSideData *sd;
249    const AVPixFmtDescriptor *pix_desc = av_pix_fmt_desc_get(frame->format);
250    JxlBasicInfo info;
251    JxlColorEncoding jxl_color;
252    JxlPixelFormat jxl_fmt;
253    JxlEncoderStatus jret;
254    int ret;
255    size_t available = ctx->buffer_size;
256    size_t bytes_written = 0;
257    uint8_t *next_out = ctx->buffer;
258
259    ret = libjxl_init_jxl_encoder(avctx);
260    if (ret) {
261        av_log(avctx, AV_LOG_ERROR, "Error frame-initializing JxlEncoder\n");
262        return ret;
263    }
264
265    /* populate the basic info settings */
266    JxlEncoderInitBasicInfo(&info);
267    jxl_fmt.num_channels = pix_desc->nb_components;
268    info.xsize = frame->width;
269    info.ysize = frame->height;
270    info.num_extra_channels = (jxl_fmt.num_channels + 1) % 2;
271    info.num_color_channels = jxl_fmt.num_channels - info.num_extra_channels;
272    info.bits_per_sample = av_get_bits_per_pixel(pix_desc) / jxl_fmt.num_channels;
273    info.alpha_bits = (info.num_extra_channels > 0) * info.bits_per_sample;
274    if (pix_desc->flags & AV_PIX_FMT_FLAG_FLOAT) {
275        info.exponent_bits_per_sample = info.bits_per_sample > 16 ? 8 : 5;
276        info.alpha_exponent_bits = info.alpha_bits ? info.exponent_bits_per_sample : 0;
277        jxl_fmt.data_type = info.bits_per_sample > 16 ? JXL_TYPE_FLOAT : JXL_TYPE_FLOAT16;
278    } else {
279        info.exponent_bits_per_sample = 0;
280        info.alpha_exponent_bits = 0;
281        jxl_fmt.data_type = info.bits_per_sample <= 8 ? JXL_TYPE_UINT8 : JXL_TYPE_UINT16;
282    }
283
284    /* JPEG XL format itself does not support limited range */
285    if (avctx->color_range == AVCOL_RANGE_MPEG ||
286        avctx->color_range == AVCOL_RANGE_UNSPECIFIED && frame->color_range == AVCOL_RANGE_MPEG)
287        av_log(avctx, AV_LOG_WARNING, "This encoder does not support limited (tv) range, colors will be wrong!\n");
288    else if (avctx->color_range != AVCOL_RANGE_JPEG && frame->color_range != AVCOL_RANGE_JPEG)
289        av_log(avctx, AV_LOG_WARNING, "Unknown color range, assuming full (pc)\n");
290
291    /* bitexact lossless requires there to be no XYB transform */
292    info.uses_original_profile = ctx->distance == 0.0;
293
294    if (JxlEncoderSetBasicInfo(ctx->encoder, &info) != JXL_ENC_SUCCESS) {
295        av_log(avctx, AV_LOG_ERROR, "Failed to set JxlBasicInfo\n");
296        return AVERROR_EXTERNAL;
297    }
298
299    /* rendering intent doesn't matter here
300     * but libjxl will whine if we don't set it */
301    jxl_color.rendering_intent = JXL_RENDERING_INTENT_RELATIVE;
302
303    switch (frame->color_trc && frame->color_trc != AVCOL_TRC_UNSPECIFIED
304            ? frame->color_trc : avctx->color_trc) {
305    case AVCOL_TRC_BT709:
306        jxl_color.transfer_function = JXL_TRANSFER_FUNCTION_709;
307        break;
308    case AVCOL_TRC_LINEAR:
309        jxl_color.transfer_function = JXL_TRANSFER_FUNCTION_LINEAR;
310        break;
311    case AVCOL_TRC_IEC61966_2_1:
312        jxl_color.transfer_function = JXL_TRANSFER_FUNCTION_SRGB;
313        break;
314    case AVCOL_TRC_SMPTE428:
315        jxl_color.transfer_function = JXL_TRANSFER_FUNCTION_DCI;
316        break;
317    case AVCOL_TRC_SMPTE2084:
318        jxl_color.transfer_function = JXL_TRANSFER_FUNCTION_PQ;
319        break;
320    case AVCOL_TRC_ARIB_STD_B67:
321        jxl_color.transfer_function = JXL_TRANSFER_FUNCTION_HLG;
322        break;
323    case AVCOL_TRC_GAMMA22:
324        jxl_color.transfer_function = JXL_TRANSFER_FUNCTION_GAMMA;
325        jxl_color.gamma = 2.2;
326        break;
327    case AVCOL_TRC_GAMMA28:
328        jxl_color.transfer_function = JXL_TRANSFER_FUNCTION_GAMMA;
329        jxl_color.gamma = 2.8;
330        break;
331    default:
332        if (pix_desc->flags & AV_PIX_FMT_FLAG_FLOAT) {
333            av_log(avctx, AV_LOG_WARNING, "Unknown transfer function, assuming Linear Light. Colors may be wrong.\n");
334            jxl_color.transfer_function = JXL_TRANSFER_FUNCTION_LINEAR;
335        } else {
336            av_log(avctx, AV_LOG_WARNING, "Unknown transfer function, assuming IEC61966-2-1/sRGB. Colors may be wrong.\n");
337            jxl_color.transfer_function = JXL_TRANSFER_FUNCTION_SRGB;
338        }
339    }
340
341    /* This should be implied to be honest
342     * but a libjxl bug makes it fail otherwise */
343    if (info.num_color_channels == 1)
344        jxl_color.color_space = JXL_COLOR_SPACE_GRAY;
345    else
346        jxl_color.color_space = JXL_COLOR_SPACE_RGB;
347
348    ret = libjxl_populate_primaries(avctx, &jxl_color,
349            frame->color_primaries && frame->color_primaries != AVCOL_PRI_UNSPECIFIED
350            ? frame->color_primaries : avctx->color_primaries);
351    if (ret < 0)
352        return ret;
353
354    sd = av_frame_get_side_data(frame, AV_FRAME_DATA_ICC_PROFILE);
355    if (sd && sd->size && JxlEncoderSetICCProfile(ctx->encoder, sd->data, sd->size) != JXL_ENC_SUCCESS)
356        av_log(avctx, AV_LOG_WARNING, "Could not set ICC Profile\n");
357    if (JxlEncoderSetColorEncoding(ctx->encoder, &jxl_color) != JXL_ENC_SUCCESS)
358        av_log(avctx, AV_LOG_WARNING, "Failed to set JxlColorEncoding\n");
359
360    /* depending on basic info, level 10 might
361     * be required instead of level 5 */
362    if (JxlEncoderGetRequiredCodestreamLevel(ctx->encoder) > 5) {
363        if (JxlEncoderSetCodestreamLevel(ctx->encoder, 10) != JXL_ENC_SUCCESS)
364            av_log(avctx, AV_LOG_WARNING, "Could not increase codestream level\n");
365    }
366
367    jxl_fmt.endianness = JXL_NATIVE_ENDIAN;
368    jxl_fmt.align = frame->linesize[0];
369
370    if (JxlEncoderAddImageFrame(ctx->options, &jxl_fmt, frame->data[0], jxl_fmt.align * info.ysize) != JXL_ENC_SUCCESS) {
371        av_log(avctx, AV_LOG_ERROR, "Failed to add Image Frame\n");
372        return AVERROR_EXTERNAL;
373    }
374
375    /*
376     * Run this after the last frame in the image has been passed.
377     * TODO support animation
378     */
379    JxlEncoderCloseInput(ctx->encoder);
380
381    while (1) {
382        jret = JxlEncoderProcessOutput(ctx->encoder, &next_out, &available);
383        if (jret == JXL_ENC_ERROR) {
384            av_log(avctx, AV_LOG_ERROR, "Unspecified libjxl error occurred\n");
385            return AVERROR_EXTERNAL;
386        }
387        bytes_written = ctx->buffer_size - available;
388        /* all data passed has been encoded */
389        if (jret == JXL_ENC_SUCCESS)
390            break;
391        if (jret == JXL_ENC_NEED_MORE_OUTPUT) {
392            /*
393             * at the moment, libjxl has no way to
394             * tell us how much space it actually needs
395             * so we need to malloc loop
396             */
397            uint8_t *temp;
398            size_t new_size = ctx->buffer_size * 2;
399            temp = av_realloc(ctx->buffer, new_size);
400            if (!temp)
401                return AVERROR(ENOMEM);
402            ctx->buffer = temp;
403            ctx->buffer_size = new_size;
404            next_out = ctx->buffer + bytes_written;
405            available = new_size - bytes_written;
406            continue;
407        }
408        av_log(avctx, AV_LOG_ERROR, "Bad libjxl event: %d\n", jret);
409        return AVERROR_EXTERNAL;
410    }
411
412    ret = ff_get_encode_buffer(avctx, pkt, bytes_written, 0);
413    if (ret < 0)
414        return ret;
415
416    memcpy(pkt->data, ctx->buffer, bytes_written);
417    *got_packet = 1;
418
419    return 0;
420}
421
422static av_cold int libjxl_encode_close(AVCodecContext *avctx)
423{
424    LibJxlEncodeContext *ctx = avctx->priv_data;
425
426    if (ctx->runner)
427        JxlThreadParallelRunnerDestroy(ctx->runner);
428    ctx->runner = NULL;
429
430    /*
431     * destroying the decoder also frees
432     * ctx->options so we don't need to
433     */
434    if (ctx->encoder)
435        JxlEncoderDestroy(ctx->encoder);
436    ctx->encoder = NULL;
437
438    av_freep(&ctx->buffer);
439
440    return 0;
441}
442
443#define OFFSET(x) offsetof(LibJxlEncodeContext, x)
444#define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
445
446static const AVOption libjxl_encode_options[] = {
447    { "effort",        "Encoding effort",                                  OFFSET(effort),     AV_OPT_TYPE_INT,    { .i64 =    7 },    1,     9, VE },
448    { "distance",      "Maximum Butteraugli distance (quality setting, "
449                        "lower = better, zero = lossless, default 1.0)",   OFFSET(distance),   AV_OPT_TYPE_FLOAT,  { .dbl = -1.0 }, -1.0,  15.0, VE },
450    { "modular",       "Force modular mode",                               OFFSET(modular),    AV_OPT_TYPE_INT,    { .i64 =    0 },    0,     1, VE },
451    { NULL },
452};
453
454static const AVClass libjxl_encode_class = {
455    .class_name = "libjxl",
456    .item_name  = av_default_item_name,
457    .option     = libjxl_encode_options,
458    .version    = LIBAVUTIL_VERSION_INT,
459};
460
461const FFCodec ff_libjxl_encoder = {
462    .p.name           = "libjxl",
463    .p.long_name      = NULL_IF_CONFIG_SMALL("libjxl JPEG XL"),
464    .p.type           = AVMEDIA_TYPE_VIDEO,
465    .p.id             = AV_CODEC_ID_JPEGXL,
466    .priv_data_size   = sizeof(LibJxlEncodeContext),
467    .init             = libjxl_encode_init,
468    FF_CODEC_ENCODE_CB(libjxl_encode_frame),
469    .close            = libjxl_encode_close,
470    .p.capabilities   = AV_CODEC_CAP_OTHER_THREADS,
471    .caps_internal    = FF_CODEC_CAP_AUTO_THREADS | FF_CODEC_CAP_INIT_CLEANUP,
472    .p.pix_fmts       = (const enum AVPixelFormat[]) {
473        AV_PIX_FMT_RGB24, AV_PIX_FMT_RGBA,
474        AV_PIX_FMT_RGB48, AV_PIX_FMT_RGBA64,
475        AV_PIX_FMT_GRAY8, AV_PIX_FMT_YA8,
476        AV_PIX_FMT_GRAY16, AV_PIX_FMT_YA16,
477        AV_PIX_FMT_GRAYF32,
478        AV_PIX_FMT_NONE
479    },
480    .p.priv_class     = &libjxl_encode_class,
481    .p.wrapper_name   = "libjxl",
482};
483