nick black



notcurses_visual - notcurses multimedia


#include <notcurses/notcurses.h>

typedef enum {
} ncscale_e;

typedef enum {
  NCBLIT_DEFAULT,// let the ncvisual choose its own blitter
  NCBLIT_1x1,    // full block                █
  NCBLIT_2x1,    // full/(upper|left) blocks  ▄█
  NCBLIT_1x1x4,  // shaded full blocks        ▓▒░█
  NCBLIT_2x2,    // quadrants                 ▗▐ ▖▄▟▌▙█
  NCBLIT_4x1,    // four vert/horz levels     █▆▄▂ / ▎▌▊█
  NCBLIT_BRAILLE,// 4x2-way braille      ⡀⡄⡆⡇⢀⣀⣄⣆⣇⢠⣠⣤⣦⣧⢰⣰⣴⣶⣷⢸⣸⣼⣾⣿
  NCBLIT_8x1,    // eight vert/horz levels    █▇▆▅▄▃▂▁ / ▏▎▍▌▋▊▉█
  NCBLIT_SIXEL,  // six rows, 1 column (RGB)
} ncblitter_e;

#define NCVISUAL_OPTION_BLEND     0x0002

struct ncvisual_options {
  struct ncplane* n;
  ncscale_e scaling;
  int y, x;
  int begy, begx; // origin of rendered section
  int leny, lenx; // size of rendered section
  ncblitter_e blitter; // glyph set to use (maps input to output cells)
  uint64_t flags; // bitmask over NCVISUAL_OPTION_*

typedef int (*streamcb)(struct notcurses*, struct ncvisual*, void*);

bool notcurses_canopen_images(const struct notcurses* nc);

bool notcurses_canopen_videos(const struct notcurses* nc);

bool notcurses_cansixel(const struct notcurses* nc);

struct ncvisual* ncvisual_from_file(const char* file, nc_err_e* err);

struct ncvisual* ncvisual_from_rgba(const void* rgba, int rows, int rowstride, int cols);

struct ncvisual* ncvisual_from_bgra(const void* bgra, int rows, int rowstride, int cols);

struct ncvisual* ncvisual_from_plane(struct ncplane* n, ncblitter_e blit, int begy, int begx, int leny, int lenx);

int ncvisual_geom(const struct notcurses* nc, const struct ncvisual* n, const struct ncvisual_options* vopts, int* y, int* x, int* toy, int* tox);

void ncvisual_destroy(struct ncvisual* ncv);

nc_err_e ncvisual_decode(struct ncvisual* nc);

struct ncplane* ncvisual_render(struct notcurses* nc, struct ncvisual* ncv, const struct ncvisual_options* vopts);

int ncvisual_simple_streamer(struct ncplane* n, struct ncvisual* ncv, const struct timespec* disptime, void* curry);

int ncvisual_stream(struct notcurses* nc, struct ncvisual* ncv, nc_err_e* err, float timescale, streamcb streamer, const struct ncvisual_options* vopts, void* curry);

int ncvisual_rotate(struct ncvisual* n, double rads);

int ncvisual_resize(struct ncvisual* n, int rows, int cols);

int ncvisual_polyfill_yx(struct ncvisual* n, int y, int x, uint32_t rgba);

int ncvisual_at_yx(const struct ncvisual* n, int y, int x, uint32_t* pixel);

int ncvisual_set_yx(const struct ncvisual* n, int y, int x, uint32_t pixel);

char* ncvisual_subtitle(const struct ncvisual* ncv);


An ncvisual is a virtual pixel framebuffer. They can be created from RGBA/BGRA data in memory (ncvisual_from_rgba/ncvisual_from_bgra), or from the content of a suitable ncplane (ncvisual_from_ncplane). If Notcurses was built against a multimedia engine (FFMpeg or OpenImageIO), image and video files can be loaded into visuals using ncvisual_from_file. ncvisual_from_file discovers the container and codecs, but does not verify that the entire file is well-formed. ncvisual_decode ought be invoked to recover subsequent frames, once per frame.

Once the visual is loaded, it can be transformed using ncvisual_rotate and ncvisual_resize. These are persistent operations, unlike any scaling that takes place at render time. If a subtitle is associated with the frame, it can be acquired with ncvisual_subtitle.

ncvisual_from_rgba and ncvisual_from_bgra both require a number of rows, a number of image columns cols, and a virtual row length of rowstride / 4 columns. The input data must provide 32 bits per pixel, and thus must be at least rowstride * rows bytes, of which a cols rows 4-byte subset is used. It is not possible to mmap(2) an image file and use it directly–decompressed, decoded data is necessary. The resulting plane will be ceil(rows/2) rows, and cols columns. ncvisual_from_plane requires specification of a rectangle via begy, begx, leny, and lenx. The only valid characters within this region are those used by the NCBLIT_2x2 blitter, though this may change in the future.

ncvisual_rotate executes a rotation of rads radians, in the clockwise (positive) or counterclockwise (negative) direction.

ncvisual_subtitle will return a UTF-8-encoded subtitle corresponding to the current frame if such a subtitle was decoded. Note that a subtitle might be returned for multiple frames, or might not.

ncvisual_render blits the visual to an ncplane, based on the contents of its struct ncvisual_options. If n is not NULL, it specifies the plane on which to render, and y/x specify a location within that plane. Otherwise, a new plane will be created, and placed at y/x relative to the rendering area. begy/begx specify the upper left corner of a subsection of the ncvisual to render, while leny/lenx specify the geometry of same. flags is a bitfield over:


The different ncblitter_e values select from among available glyph sets:


notcurses_canopen_images and notcurses_canopen_videos returns true if images and videos, respecitvely, can be decoded, or false if Notcurses was built with insufficient multimedia support.

ncvisual_from_file returns an ncvisual object on success, or NULL on failure. Success indicates that the specified file was opened, and enough data was read to make a firm codec identification. It does not imply that the entire file is properly-formed. On failure, err will be updated. ncvisual_decode returns NCERR_SUCCESS on success, or NCERR_EOF on end of file, or some other nc_err_e on failure. It likewise updates err in the event of an error. It is only necessary for multimedia-based visuals.

ncvisual_from_plane returns NULL if the ncvisual cannot be created and bound. This is usually due to illegal content in the source ncplane.

ncvisual_render returns NULL on error, and otherwise the plane to which the visual was rendered. If opts->n is provided, this will be opts->n. Otherwise, a plane will be created, perfectly sized for the visual and the specified blitter.

ncvisual_geom returns non-zero if the blitter is invalid.


Multimedia decoding requires that Notcurses be built with either FFmpeg or OpenImageIO support. What formats can be decoded is totally dependent on the linked library. OpenImageIO does not support subtitles.


ncvisual_rotate currently supports only M_PI/2 and -M_PI/2 radians for rads, but this will change soon.


notcurses(3), notcurses_plane(3), utf-8(7)