Quantum Painter is the standardised API for graphical displays. It currently includes support for basic drawing primitives, as well as custom images, animations, and fonts.
Due to the complexity, there is no support for Quantum Painter on AVR-based boards.
To enable overall Quantum Painter to be built into your firmware, add the following to rules.mk
:
QUANTUM_PAINTER_ENABLE = yes
QUANTUM_PAINTER_DRIVERS += ......
You will also likely need to select an appropriate driver in rules.mk
, which is listed below.
!> Quantum Painter is not currently integrated with system-level operations such as when the keyboard goes into suspend. Users will need to handle this manually at the current time.
The QMK CLI can be used to convert from normal images such as PNG files or animated GIFs, as well as fonts from TTF files.
Supported devices:
Display Panel | Panel Type | Size | Comms Transport | Driver |
---|---|---|---|---|
GC9A01 | RGB LCD (circular) | 240x240 | SPI + D/C + RST | QUANTUM_PAINTER_DRIVERS += gc9a01_spi |
ILI9163 | RGB LCD | 128x128 | SPI + D/C + RST | QUANTUM_PAINTER_DRIVERS += ili9163_spi |
ILI9341 | RGB LCD | 240x320 | SPI + D/C + RST | QUANTUM_PAINTER_DRIVERS += ili9341_spi |
ILI9486 | RGB LCD | 320x480 | SPI + D/C + RST | QUANTUM_PAINTER_DRIVERS += ili9486_spi |
ILI9488 | RGB LCD | 320x480 | SPI + D/C + RST | QUANTUM_PAINTER_DRIVERS += ili9488_spi |
SSD1351 | RGB OLED | 128x128 | SPI + D/C + RST | QUANTUM_PAINTER_DRIVERS += ssd1351_spi |
ST7735 | RGB LCD | 132x162, 80x160 | SPI + D/C + RST | QUANTUM_PAINTER_DRIVERS += st7735_spi |
ST7789 | RGB LCD | 240x320, 240x240 | SPI + D/C + RST | QUANTUM_PAINTER_DRIVERS += st7789_spi |
SH1106 (SPI) | Monochrome OLED | 128x64 | SPI + D/C + RST | QUANTUM_PAINTER_DRIVERS += sh1106_spi |
SH1106 (I2C) | Monochrome OLED | 128x64 | I2C | QUANTUM_PAINTER_DRIVERS += sh1106_i2c |
SSD1306 (SPI) | Monochrome OLED | 128x64 | SPI + D/C + RST | QUANTUM_PAINTER_DRIVERS += sh1106_spi |
SSD1306 (I2C) | Monochrome OLED | 128x32 | I2C | QUANTUM_PAINTER_DRIVERS += sh1106_i2c |
Surface | Virtual | User-defined | None | QUANTUM_PAINTER_DRIVERS += surface |
Option | Default | Purpose |
---|---|---|
QUANTUM_PAINTER_DISPLAY_TIMEOUT |
30000 |
This controls the amount of time (in milliseconds) that all displays will remain on after the last user input. If set to 0 , the display will remain on indefinitely. |
QUANTUM_PAINTER_TASK_THROTTLE |
1 |
This controls the amount of time (in milliseconds) that the Quantum Painter internal task will wait between each execution. Affects animations, display timeout, and LVGL timing if enabled. |
QUANTUM_PAINTER_NUM_IMAGES |
8 |
The maximum number of images/animations that can be loaded at any one time. |
QUANTUM_PAINTER_NUM_FONTS |
4 |
The maximum number of fonts that can be loaded at any one time. |
QUANTUM_PAINTER_CONCURRENT_ANIMATIONS |
4 |
The maximum number of animations that can be executed at the same time. |
QUANTUM_PAINTER_LOAD_FONTS_TO_RAM |
FALSE |
Whether or not fonts should be loaded to RAM. Relevant for fonts stored in off-chip persistent storage, such as external flash. |
QUANTUM_PAINTER_PIXDATA_BUFFER_SIZE |
1024 |
The limit of the amount of pixel data that can be transmitted in one transaction to the display. Higher values require more RAM on the MCU. |
QUANTUM_PAINTER_SUPPORTS_256_PALETTE |
FALSE |
If 256-color palettes are supported. Requires significantly more RAM on the MCU. |
QUANTUM_PAINTER_SUPPORTS_NATIVE_COLORS |
FALSE |
If native color range is supported. Requires significantly more RAM on the MCU. |
QUANTUM_PAINTER_DEBUG |
unset | Prints out significant amounts of debugging information to CONSOLE output. Significant performance degradation, use only for debugging. |
QUANTUM_PAINTER_DEBUG_ENABLE_FLUSH_TASK_OUTPUT |
unset | By default, debug output is disabled while the internal task is flushing the display(s). If you want to keep it enabled, add this to your config.h . Note: Console will get clogged. |
Drivers have their own set of configurable options, and are described in their respective sections.
This command converts images to a format usable by QMK, i.e. the QGF File Format.
Usage:
usage: qmk painter-convert-graphics [-h] [-w] [-d] [-r] -f FORMAT [-o OUTPUT] -i INPUT [-v]
options:
-h, --help show this help message and exit
-w, --raw Writes out the QGF file as raw data instead of c/h combo.
-d, --no-deltas Disables the use of delta frames when encoding animations.
-r, --no-rle Disables the use of RLE when encoding images.
-f FORMAT, --format FORMAT
Output format, valid types: rgb888, rgb565, pal256, pal16, pal4, pal2, mono256, mono16, mono4, mono2
-o OUTPUT, --output OUTPUT
Specify output directory. Defaults to same directory as input.
-i INPUT, --input INPUT
Specify input graphic file.
-v, --verbose Turns on verbose output.
The INPUT
argument can be any image file loadable by Python's Pillow module. Common formats include PNG, or Animated GIF.
The OUTPUT
argument needs to be a directory, and will default to the same directory as the input argument.
The FORMAT
argument can be any of the following:
Format | Meaning |
---|---|
rgb888 |
16,777,216 colors in 8-8-8 RGB format (requires QUANTUM_PAINTER_SUPPORTS_NATIVE_COLORS ) |
rgb565 |
65,536 colors in 5-6-5 RGB format (requires QUANTUM_PAINTER_SUPPORTS_NATIVE_COLORS ) |
pal256 |
256-color palette (requires QUANTUM_PAINTER_SUPPORTS_256_PALETTE ) |
pal16 |
16-color palette |
pal4 |
4-color palette |
pal2 |
2-color palette |
mono256 |
256-shade grayscale (requires QUANTUM_PAINTER_SUPPORTS_256_PALETTE ) |
mono16 |
16-shade grayscale |
mono4 |
4-shade grayscale |
mono2 |
2-shade grayscale |
Examples:
$ cd /home/qmk/qmk_firmware/keyboards/my_keeb
$ qmk painter-convert-graphics -f mono16 -i my_image.gif -o ./generated/
Writing /home/qmk/qmk_firmware/keyboards/my_keeb/generated/my_image.qgf.h...
Writing /home/qmk/qmk_firmware/keyboards/my_keeb/generated/my_image.qgf.c...
This command converts a TTF font to an intermediate format for editing, before converting to the QFF File Format.
Usage:
usage: qmk painter-make-font-image [-h] [-a] [-u UNICODE_GLYPHS] [-n] [-s SIZE] -o OUTPUT -f FONT
optional arguments:
-h, --help show this help message and exit
-a, --no-aa Disable anti-aliasing on fonts.
-u UNICODE_GLYPHS, --unicode-glyphs UNICODE_GLYPHS
Also generate the specified unicode glyphs.
-n, --no-ascii Disables output of the full ASCII character set (0x20..0x7E), exporting only the glyphs specified.
-s SIZE, --size SIZE Specify font size. Default 12.
-o OUTPUT, --output OUTPUT
Specify output image path.
-f FONT, --font FONT Specify input font file.
The FONT
argument is generally a TrueType Font file (TTF).
The OUTPUT
argument is the output image to generate, generally something like my_font.png
.
The UNICODE_GLYPHS
argument allows for specifying extra unicode glyphs to generate, and accepts a string.
Examples:
$ qmk painter-make-font-image --font NotoSans-ExtraCondensedBold.ttf --size 11 -o noto11.png --unicode-glyphs "ĄȽɂɻɣɈʣ"
This command converts an intermediate font image to the QFF File Format.
This command expects an image that conforms to the following format:
- Top-left pixel (at
0,0
) is the "delimiter" color:- Each glyph in the font starts when a pixel of this color is found on the first row
- The first row is discarded when converting to the QFF format
- The number of delimited glyphs must match the supplied arguments to the command:
- The full ASCII set
0x20..0x7E
(if--no-ascii
was not specified) - The corresponding number of unicode glyphs if any were specified with
--unicode-glyphs
- The full ASCII set
- The order of the glyphs matches the ASCII set, if any, followed by the Unicode glyph set, if any.
Usage:
usage: qmk painter-convert-font-image [-h] [-w] [-r] -f FORMAT [-u UNICODE_GLYPHS] [-n] [-o OUTPUT] [-i INPUT]
options:
-h, --help show this help message and exit
-w, --raw Writes out the QFF file as raw data instead of c/h combo.
-r, --no-rle Disable the use of RLE to minimise converted image size.
-f FORMAT, --format FORMAT
Output format, valid types: rgb565, pal256, pal16, pal4, pal2, mono256, mono16, mono4, mono2
-u UNICODE_GLYPHS, --unicode-glyphs UNICODE_GLYPHS
Also generate the specified unicode glyphs.
-n, --no-ascii Disables output of the full ASCII character set (0x20..0x7E), exporting only the glyphs specified.
-o OUTPUT, --output OUTPUT
Specify output directory. Defaults to same directory as input.
-i INPUT, --input INPUT
Specify input graphic file.
The same arguments for --no-ascii
and --unicode-glyphs
need to be specified, as per qmk painter-make-font-image
.
Examples:
$ cd /home/qmk/qmk_firmware/keyboards/my_keeb
$ qmk painter-convert-font-image --input noto11.png -f mono4 --unicode-glyphs "ĄȽɂɻɣɈʣ"
Writing /home/qmk/qmk_firmware/keyboards/my_keeb/generated/noto11.qff.h...
Writing /home/qmk/qmk_firmware/keyboards/my_keeb/generated/noto11.qff.c...
Most TFT display panels use a 5-pin interface -- SPI SCK, SPI MOSI, SPI CS, D/C, and RST pins.
For these displays, QMK's spi_master
must already be correctly configured for the platform you're building for.
The pin assignments for SPI CS, D/C, and RST are specified during device construction.
Enabling support for the GC9A01 in Quantum Painter is done by adding the following to rules.mk
:
QUANTUM_PAINTER_ENABLE = yes
QUANTUM_PAINTER_DRIVERS += gc9a01_spi
Creating a GC9A01 device in firmware can then be done with the following API:
painter_device_t qp_gc9a01_make_spi_device(uint16_t panel_width, uint16_t panel_height, pin_t chip_select_pin, pin_t dc_pin, pin_t reset_pin, uint16_t spi_divisor, int spi_mode);
The device handle returned from the qp_gc9a01_make_spi_device
function can be used to perform all other drawing operations.
The maximum number of displays can be configured by changing the following in your config.h
(default is 1):
// 3 displays:
#define GC9A01_NUM_DEVICES 3
Native color format rgb565 is compatible with GC9A01
Enabling support for the ILI9163 in Quantum Painter is done by adding the following to rules.mk
:
QUANTUM_PAINTER_ENABLE = yes
QUANTUM_PAINTER_DRIVERS += ili9163_spi
Creating a ILI9163 device in firmware can then be done with the following API:
painter_device_t qp_ili9163_make_spi_device(uint16_t panel_width, uint16_t panel_height, pin_t chip_select_pin, pin_t dc_pin, pin_t reset_pin, uint16_t spi_divisor, int spi_mode);
The device handle returned from the qp_ili9163_make_spi_device
function can be used to perform all other drawing operations.
The maximum number of displays can be configured by changing the following in your config.h
(default is 1):
// 3 displays:
#define ILI9163_NUM_DEVICES 3
Native color format rgb565 is compatible with ILI9163
Enabling support for the ILI9341 in Quantum Painter is done by adding the following to rules.mk
:
QUANTUM_PAINTER_ENABLE = yes
QUANTUM_PAINTER_DRIVERS += ili9341_spi
Creating a ILI9341 device in firmware can then be done with the following API:
painter_device_t qp_ili9341_make_spi_device(uint16_t panel_width, uint16_t panel_height, pin_t chip_select_pin, pin_t dc_pin, pin_t reset_pin, uint16_t spi_divisor, int spi_mode);
The device handle returned from the qp_ili9341_make_spi_device
function can be used to perform all other drawing operations.
The maximum number of displays can be configured by changing the following in your config.h
(default is 1):
// 3 displays:
#define ILI9341_NUM_DEVICES 3
Native color format rgb565 is compatible with ILI9341
Enabling support for the ILI9486 in Quantum Painter is done by adding the following to rules.mk
:
QUANTUM_PAINTER_ENABLE = yes
QUANTUM_PAINTER_DRIVERS += ili9486_spi
Creating a ILI9486 device in firmware can then be done with the following API:
painter_device_t qp_ili9486_make_spi_device(uint16_t panel_width, uint16_t panel_height, pin_t chip_select_pin, pin_t dc_pin, pin_t reset_pin, uint16_t spi_divisor, int spi_mode);
There's another variant for this Waveshare module, because it has a quirky SPI->Parallel converter. You can create it with:
painter_device_t qp_ili9486_make_spi_waveshare_device(uint16_t panel_width, uint16_t panel_height, pin_t chip_select_pin, pin_t dc_pin, pin_t reset_pin, uint16_t spi_divisor, int spi_mode);
The device handle returned from these functions can be used to perform all other drawing operations.
The maximum number of displays can be configured by changing the following in your config.h
(default is 1):
// 3 displays:
#define ILI9486_NUM_DEVICES 3
Native color format rgb888 is compatible with ILI9486 Native color format rgb565 is compatible with ILI9486 Waveshare
Enabling support for the ILI9488 in Quantum Painter is done by adding the following to rules.mk
:
QUANTUM_PAINTER_ENABLE = yes
QUANTUM_PAINTER_DRIVERS += ili9488_spi
Creating a ILI9488 device in firmware can then be done with the following API:
painter_device_t qp_ili9488_make_spi_device(uint16_t panel_width, uint16_t panel_height, pin_t chip_select_pin, pin_t dc_pin, pin_t reset_pin, uint16_t spi_divisor, int spi_mode);
The device handle returned from the qp_ili9488_make_spi_device
function can be used to perform all other drawing operations.
The maximum number of displays can be configured by changing the following in your config.h
(default is 1):
// 3 displays:
#define ILI9488_NUM_DEVICES 3
Native color format rgb888 is compatible with ILI9488
Enabling support for the ST7735 in Quantum Painter is done by adding the following to rules.mk
:
QUANTUM_PAINTER_ENABLE = yes
QUANTUM_PAINTER_DRIVERS += st7735_spi
Creating a ST7735 device in firmware can then be done with the following API:
painter_device_t qp_st7735_make_spi_device(uint16_t panel_width, uint16_t panel_height, pin_t chip_select_pin, pin_t dc_pin, pin_t reset_pin, uint16_t spi_divisor, int spi_mode);
The device handle returned from the qp_st7735_make_spi_device
function can be used to perform all other drawing operations.
The maximum number of displays can be configured by changing the following in your config.h
(default is 1):
// 3 displays:
#define ST7735_NUM_DEVICES 3
Native color format rgb565 is compatible with ST7735
!> Some ST7735 devices are known to have different drawing offsets -- despite being a 132x162 pixel display controller internally, some display panels are only 80x160, or smaller. These may require an offset to be applied; see qp_set_viewport_offsets
above for information on how to override the offsets if they aren't correctly rendered.
Enabling support for the ST7789 in Quantum Painter is done by adding the following to rules.mk
:
QUANTUM_PAINTER_ENABLE = yes
QUANTUM_PAINTER_DRIVERS += st7789_spi
Creating a ST7789 device in firmware can then be done with the following API:
painter_device_t qp_st7789_make_spi_device(uint16_t panel_width, uint16_t panel_height, pin_t chip_select_pin, pin_t dc_pin, pin_t reset_pin, uint16_t spi_divisor, int spi_mode);
The device handle returned from the qp_st7789_make_spi_device
function can be used to perform all other drawing operations.
The maximum number of displays can be configured by changing the following in your config.h
(default is 1):
// 3 displays:
#define ST7789_NUM_DEVICES 3
Native color format rgb565 is compatible with ST7789
!> Some ST7789 devices are known to have different drawing offsets -- despite being a 240x320 pixel display controller internally, some display panels are only 240x240, or smaller. These may require an offset to be applied; see qp_set_viewport_offsets
above for information on how to override the offsets if they aren't correctly rendered.
OLED displays tend to use 5-pin SPI when at larger resolutions, or when using color -- SPI SCK, SPI MOSI, SPI CS, D/C, and RST pins. Smaller OLEDs may use I2C instead.
When using these displays, either spi_master
or i2c_master
must already be correctly configured for both the platform and panel you're building for.
For SPI, the pin assignments for SPI CS, D/C, and RST are specified during device construction -- for I2C the panel's address is specified instead.
Enabling support for the SSD1351 in Quantum Painter is done by adding the following to rules.mk
:
QUANTUM_PAINTER_ENABLE = yes
QUANTUM_PAINTER_DRIVERS += ssd1351_spi
Creating a SSD1351 device in firmware can then be done with the following API:
painter_device_t qp_ssd1351_make_spi_device(uint16_t panel_width, uint16_t panel_height, pin_t chip_select_pin, pin_t dc_pin, pin_t reset_pin, uint16_t spi_divisor, int spi_mode);
The device handle returned from the qp_ssd1351_make_spi_device
function can be used to perform all other drawing operations.
The maximum number of displays can be configured by changing the following in your config.h
(default is 1):
// 3 displays:
#define SSD1351_NUM_DEVICES 3
Native color format rgb565 is compatible with SSD1351
Enabling support for the SH1106 in Quantum Painter is done by adding the following to rules.mk
:
QUANTUM_PAINTER_ENABLE = yes
# For SPI:
QUANTUM_PAINTER_DRIVERS += sh1106_spi
# For I2C:
QUANTUM_PAINTER_DRIVERS += sh1106_i2c
Creating a SH1106 device in firmware can then be done with the following APIs:
// SPI-based SH1106:
painter_device_t qp_sh1106_make_spi_device(uint16_t panel_width, uint16_t panel_height, pin_t chip_select_pin, pin_t dc_pin, pin_t reset_pin, uint16_t spi_divisor, int spi_mode);
// I2C-based SH1106:
painter_device_t qp_sh1106_make_i2c_device(uint16_t panel_width, uint16_t panel_height, uint8_t i2c_address);
The device handle returned from the qp_sh1106_make_???_device
function can be used to perform all other drawing operations.
The maximum number of displays of each type can be configured by changing the following in your config.h
(default is 1):
// 3 SPI displays:
#define SH1106_NUM_SPI_DEVICES 3
// 3 I2C displays:
#define SH1106_NUM_I2C_DEVICES 3
Native color format mono2 is compatible with SH1106
SSD1306 and SH1106 are almost entirely identical, to the point of being indisinguishable by Quantum Painter. Enable SH1106 support in Quantum Painter and create SH1106 devices in firmware to perform drawing operations on SSD1306 displays.
Quantum Painter has a surface driver which is able to target a buffer in RAM. In general, surfaces keep track of the "dirty" region -- the area that has been drawn to since the last flush -- so that when transferring to the display they can transfer the minimal amount of data to achieve the end result.
!> These generally require significant amounts of RAM, so at large sizes and/or higher bit depths, they may not be usable on all MCUs.
Enabling support for surfaces in Quantum Painter is done by adding the following to rules.mk
:
QUANTUM_PAINTER_ENABLE = yes
QUANTUM_PAINTER_DRIVERS += surface
Creating a surface in firmware can then be done with the following APIs:
// 16bpp RGB565 surface:
painter_device_t qp_make_rgb565_surface(uint16_t panel_width, uint16_t panel_height, void *buffer);
// 1bpp monochrome surface:
painter_device_t qp_make_mono1bpp_surface(uint16_t panel_width, uint16_t panel_height, void *buffer);
The buffer
is a user-supplied area of memory, which can be statically allocated using SURFACE_REQUIRED_BUFFER_BYTE_SIZE
:
// Buffer required for a 240x80 16bpp surface:
uint8_t framebuffer[SURFACE_REQUIRED_BUFFER_BYTE_SIZE(240, 80, 16)];
The device handle returned from the qp_make_?????_surface
function can be used to perform all other drawing operations.
Example:
static painter_device_t my_surface;
static uint8_t my_framebuffer[SURFACE_REQUIRED_BUFFER_BYTE_SIZE(240, 80, 16)]; // Allocate a buffer for a 16bpp 240x80 RGB565 display
void keyboard_post_init_kb(void) {
my_surface = qp_rgb565_make_surface(240, 80, my_framebuffer);
qp_init(my_surface, QP_ROTATION_0);
keyboard_post_init_user();
}
The maximum number of surfaces can be configured by changing the following in your config.h
(default is 1):
// 3 surfaces:
#define SURFACE_NUM_DEVICES 3
To transfer the contents of the surface to another display of the same pixel format, the following API can be invoked:
bool qp_surface_draw(painter_device_t surface, painter_device_t display, uint16_t x, uint16_t y, bool entire_surface);
The surface
is the surface to copy out from. The display
is the target display to draw into. x
and y
are the target location to draw the surface pixel data. Under normal circumstances, the location should be consistent, as the dirty region is calculated with respect to the x
and y
coordinates -- changing those will result in partial, overlapping draws. entire_surface
whether the entire surface should be drawn, instead of just the dirty region.
!> The surface and display panel must have the same native pixel format.
?> Calling qp_flush()
on the surface resets its dirty region. Copying the surface contents to the display also automatically resets the dirty region.
All APIs require a painter_device_t
object as their first parameter -- this object comes from the specific device initialisation, and instructions on creating it can be found in each driver's respective section.
To use any of the APIs, you need to include qp.h
:
#include <qp.h>
The coordinate system used in Quantum Painter generally accepts left
, top
, right
, and bottom
instead of x/y/width/height, and each coordinate is inclusive of where pixels should be drawn. This is required as some datatypes used by display panels have a maximum value of 255
-- for any value or geometry extent that matches 256
, this would be represented as a 0
, instead.
?> Drawing a horizontal line 8 pixels long, starting from 4 pixels inside the left side of the display, will need left=4
, right=11
.
All color data matches the standard QMK HSV triplet definitions:
- Hue is of the range
0...255
and is internally mapped to 0...360 degrees. - Saturation is of the range
0...255
and is internally mapped to 0...100% saturation. - Value is of the range
0...255
and is internally mapped to 0...100% brightness.
?> Colors used in Quantum Painter are not subject to the RGB lighting CIE curve, if it is enabled.
bool qp_init(painter_device_t device, painter_rotation_t rotation);
The qp_init
function is used to initialise a display device after it has been created. This accepts a rotation parameter (QP_ROTATION_0
, QP_ROTATION_90
, QP_ROTATION_180
, QP_ROTATION_270
), which makes sure that the orientation of what's drawn on the display is correct.
static painter_device_t display;
void keyboard_post_init_kb(void) {
display = qp_make_.......; // Create the display
qp_init(display, QP_ROTATION_0); // Initialise the display
}
bool qp_power(painter_device_t device, bool power_on);
The qp_power
function instructs the display whether or not the display panel should be on or off.
!> If there is a separate backlight controlled through the normal QMK backlight API, this is not controlled by the qp_power
function and needs to be manually handled elsewhere.
static uint8_t last_backlight = 255;
void suspend_power_down_user(void) {
if (last_backlight == 255) {
last_backlight = get_backlight_level();
}
backlight_set(0);
rgb_matrix_set_suspend_state(true);
qp_power(display, false);
}
void suspend_wakeup_init_user(void) {
qp_power(display, true);
rgb_matrix_set_suspend_state(false);
if (last_backlight != 255) {
backlight_set(last_backlight);
}
last_backlight = 255;
}
bool qp_clear(painter_device_t device);
The qp_clear
function clears the display's screen.
bool qp_flush(painter_device_t device);
The qp_flush
function ensures that all drawing operations are "pushed" to the display. This should be done as the last operation whenever a sequence of draws occur, and guarantees that any changes are applied.
!> Some display panels may seem to work even without a call to qp_flush
-- this may be because the driver cannot queue drawing operations and needs to display them immediately when invoked. In general, calling qp_flush
at the end is still considered "best practice".
void housekeeping_task_user(void) {
static uint32_t last_draw = 0;
if (timer_elapsed32(last_draw) > 33) { // Throttle to 30fps
last_draw = timer_read32();
// Draw a rect based off the current RGB color
qp_rect(display, 0, 7, 0, 239, rgb_matrix_get_hue(), 255, 255);
qp_flush(display);
}
}
bool qp_setpixel(painter_device_t device, uint16_t x, uint16_t y, uint8_t hue, uint8_t sat, uint8_t val);
The qp_setpixel
can be used to set a specific pixel on the screen to the supplied color.
?> Using qp_setpixel
for large amounts of drawing operations is inefficient and should be avoided unless they cannot be achieved with other drawing APIs.
void housekeeping_task_user(void) {
static uint32_t last_draw = 0;
if (timer_elapsed32(last_draw) > 33) { // Throttle to 30fps
last_draw = timer_read32();
// Draw a 240px high vertical rainbow line on X=0:
for (int i = 0; i < 239; ++i) {
qp_setpixel(display, 0, i, i, 255, 255);
}
qp_flush(display);
}
}
bool qp_line(painter_device_t device, uint16_t x0, uint16_t y0, uint16_t x1, uint16_t y1, uint8_t hue, uint8_t sat, uint8_t val);
The qp_line
can be used to draw lines on the screen with the supplied color.
void housekeeping_task_user(void) {
static uint32_t last_draw = 0;
if (timer_elapsed32(last_draw) > 33) { // Throttle to 30fps
last_draw = timer_read32();
// Draw 8px-wide rainbow down the left side of the display
for (int i = 0; i < 239; ++i) {
qp_line(display, 0, i, 7, i, i, 255, 255);
}
qp_flush(display);
}
}
bool qp_rect(painter_device_t device, uint16_t left, uint16_t top, uint16_t right, uint16_t bottom, uint8_t hue, uint8_t sat, uint8_t val, bool filled);
The qp_rect
can be used to draw rectangles on the screen with the supplied color, with or without a background fill. If not filled, any pixels inside the rectangle will be left as-is.
void housekeeping_task_user(void) {
static uint32_t last_draw = 0;
if (timer_elapsed32(last_draw) > 33) { // Throttle to 30fps
last_draw = timer_read32();
// Draw 8px-wide rainbow filled rectangles down the left side of the display
for (int i = 0; i < 239; i+=8) {
qp_rect(display, 0, i, 7, i+7, i, 255, 255, true);
}
qp_flush(display);
}
}
bool qp_circle(painter_device_t device, uint16_t x, uint16_t y, uint16_t radius, uint8_t hue, uint8_t sat, uint8_t val, bool filled);
The qp_circle
can be used to draw circles on the screen with the supplied color, with or without a background fill. If not filled, any pixels inside the circle will be left as-is.
void housekeeping_task_user(void) {
static uint32_t last_draw = 0;
if (timer_elapsed32(last_draw) > 33) { // Throttle to 30fps
last_draw = timer_read32();
// Draw r=4 filled circles down the left side of the display
for (int i = 0; i < 239; i+=8) {
qp_circle(display, 4, 4+i, 4, i, 255, 255, true);
}
qp_flush(display);
}
}
bool qp_ellipse(painter_device_t device, uint16_t x, uint16_t y, uint16_t sizex, uint16_t sizey, uint8_t hue, uint8_t sat, uint8_t val, bool filled);
The qp_ellipse
can be used to draw ellipses on the screen with the supplied color, with or without a background fill. If not filled, any pixels inside the ellipses will be left as-is.
void housekeeping_task_user(void) {
static uint32_t last_draw = 0;
if (timer_elapsed32(last_draw) > 33) { // Throttle to 30fps
last_draw = timer_read32();
// Draw 16x8 filled ellipses down the left side of the display
for (int i = 0; i < 239; i+=8) {
qp_ellipse(display, 8, 4+i, 16, 8, i, 255, 255, true);
}
qp_flush(display);
}
}
Making an image available for use requires compiling it into your firmware. To do so, assuming you've created my_image.qgf.c
and my_image.qgf.h
as per the CLI examples above, you'd add the following to your rules.mk
:
SRC += my_image.qgf.c
...and in your keymap.c
, you'd add to the top of the file:
#include "my_image.qgf.h"
painter_image_handle_t qp_load_image_mem(const void *buffer);
The qp_load_image_mem
function loads a QGF image from memory or flash.
qp_load_image_mem
returns a handle to the loaded image, which can then be used to draw to the screen using qp_drawimage
, qp_drawimage_recolor
, qp_animate
, or qp_animate_recolor
. If an image is no longer required, it can be unloaded by calling qp_close_image
below.
See the CLI Commands for instructions on how to convert images to QGF.
?> The total number of images available to load at any one time is controlled by the configurable option QUANTUM_PAINTER_NUM_IMAGES
in the table above. If more images are required, the number should be increased in config.h
.
Image information is available through accessing the handle:
Property | Accessor |
---|---|
Width | image->width |
Height | image->height |
Frame Count | image->frame_count |
bool qp_close_image(painter_image_handle_t image);
The qp_close_image
function releases resources related to the loading of the supplied image.
bool qp_drawimage(painter_device_t device, uint16_t x, uint16_t y, painter_image_handle_t image);
bool qp_drawimage_recolor(painter_device_t device, uint16_t x, uint16_t y, painter_image_handle_t image, uint8_t hue_fg, uint8_t sat_fg, uint8_t val_fg, uint8_t hue_bg, uint8_t sat_bg, uint8_t val_bg);
The qp_drawimage
and qp_drawimage_recolor
functions draw the supplied image to the screen at the supplied location, with the latter function allowing for monochrome-based images to be recolored.
// Draw an image on the bottom-right of the 240x320 display on initialisation
static painter_image_handle_t my_image;
void keyboard_post_init_kb(void) {
my_image = qp_load_image_mem(gfx_my_image);
if (my_image != NULL) {
qp_drawimage(display, (239 - my_image->width), (319 - my_image->height), my_image);
}
}
deferred_token qp_animate(painter_device_t device, uint16_t x, uint16_t y, painter_image_handle_t image);
deferred_token qp_animate_recolor(painter_device_t device, uint16_t x, uint16_t y, painter_image_handle_t image, uint8_t hue_fg, uint8_t sat_fg, uint8_t val_fg, uint8_t hue_bg, uint8_t sat_bg, uint8_t val_bg);
The qp_animate
and qp_animate_recolor
functions draw the supplied image to the screen at the supplied location, with the latter function allowing for monochrome-based animations to be recolored. They also set up internal timing such that each frame is rendered at the correct time as per the animated image.
Once an image has been set to animate, it will loop indefinitely until stopped, with no user intervention required.
Both functions return a deferred_token
, which can then be used to stop the animation, using qp_stop_animation
below.
// Animate an image on the bottom-right of the 240x320 display on initialisation
static painter_image_handle_t my_image;
static deferred_token my_anim;
void keyboard_post_init_kb(void) {
my_image = qp_load_image_mem(gfx_my_image);
if (my_image != NULL) {
my_anim = qp_animate(display, (239 - my_image->width), (319 - my_image->height), my_image);
}
}
void qp_stop_animation(deferred_token anim_token);
The qp_stop_animation
function stops the previously-started animation.
void housekeeping_task_user(void) {
if (some_random_stop_reason) {
qp_stop_animation(my_anim);
}
}
Making a font available for use requires compiling it into your firmware. To do so, assuming you've created my_font.qff.c
and my_font.qff.h
as per the CLI examples above, you'd add the following to your rules.mk
:
SRC += noto11.qff.c
...and in your keymap.c
, you'd add to the top of the file:
#include "noto11.qff.h"
painter_font_handle_t qp_load_font_mem(const void *buffer);
The qp_load_font_mem
function loads a QFF font from memory or flash.
qp_load_font_mem
returns a handle to the loaded font, which can then be measured using qp_textwidth
, or drawn to the screen using qp_drawtext
, or qp_drawtext_recolor
. If a font is no longer required, it can be unloaded by calling qp_close_font
below.
See the CLI Commands for instructions on how to convert TTF fonts to QFF.
?> The total number of fonts available to load at any one time is controlled by the configurable option QUANTUM_PAINTER_NUM_FONTS
in the table above. If more fonts are required, the number should be increased in config.h
.
Font information is available through accessing the handle:
Property | Accessor |
---|---|
Line Height | image->line_height |
bool qp_close_font(painter_font_handle_t font);
The qp_close_font
function releases resources related to the loading of the supplied font.
int16_t qp_textwidth(painter_font_handle_t font, const char *str);
The qp_textwidth
function allows measurement of how many pixels wide the supplied string would result in, for the given font.
int16_t qp_drawtext(painter_device_t device, uint16_t x, uint16_t y, painter_font_handle_t font, const char *str);
int16_t qp_drawtext_recolor(painter_device_t device, uint16_t x, uint16_t y, painter_font_handle_t font, const char *str, uint8_t hue_fg, uint8_t sat_fg, uint8_t val_fg, uint8_t hue_bg, uint8_t sat_bg, uint8_t val_bg);
The qp_drawtext
and qp_drawtext_recolor
functions draw the supplied string to the screen at the given location using the font supplied, with the latter function allowing for monochrome-based fonts to be recolored.
// Draw a text message on the bottom-right of the 240x320 display on initialisation
static painter_font_handle_t my_font;
void keyboard_post_init_kb(void) {
my_font = qp_load_font_mem(font_noto11);
if (my_font != NULL) {
static const char *text = "Hello from QMK!";
int16_t width = qp_textwidth(my_font, text);
qp_drawtext(display, (239 - width), (319 - my_font->line_height), my_font, text);
}
}
These functions allow external code to retrieve the current width, height, rotation, and drawing offsets.
uint16_t qp_get_width(painter_device_t device);
uint16_t qp_get_height(painter_device_t device);
painter_rotation_t qp_get_rotation(painter_device_t device);
uint16_t qp_get_offset_x(painter_device_t device);
uint16_t qp_get_offset_y(painter_device_t device);
Convenience function to call all the previous ones at once.
Note: You can pass NULL
for the values you are not interested in.
void qp_get_geometry(painter_device_t device, uint16_t *width, uint16_t *height, painter_rotation_t *rotation, uint16_t *offset_x, uint16_t *offset_y);
void qp_set_viewport_offsets(painter_device_t device, uint16_t offset_x, uint16_t offset_y);
The qp_set_viewport_offsets
function can be used to offset all subsequent drawing operations. For example, if a display controller is internally 240x320, but the display panel is 240x240 and has a Y offset of 80 pixels, you could invoke qp_set_viewport_offsets(display, 0, 80);
and the drawing positioning would be corrected.
bool qp_viewport(painter_device_t device, uint16_t left, uint16_t top, uint16_t right, uint16_t bottom);
The qp_viewport
function controls where raw pixel data is written to.
bool qp_pixdata(painter_device_t device, const void *pixel_data, uint32_t native_pixel_count);
The qp_pixdata
function allows raw pixel data to be streamed to the display. It requires a native pixel count rather than the number of bytes to transfer, to ensure display panel data alignment is respected. E.g. for display panels using RGB565 internal format, sending 10 pixels will result in 20 bytes of transfer.
!> Under normal circumstances, users will not need to manually call either qp_viewport
or qp_pixdata
. These allow for writing of raw pixel information, in the display panel's native format, to the area defined by the viewport.