pub trait DrawTarget: Dimensions {
    type Color: PixelColor;
    type Error;

    // Required method
    fn draw_iter<I>(&mut self, pixels: I) -> Result<(), Self::Error>
       where I: IntoIterator<Item = Pixel<Self::Color>>;

    // Provided methods
    fn fill_contiguous<I>(
        &mut self,
        area: &Rectangle,
        colors: I
    ) -> Result<(), Self::Error>
       where I: IntoIterator<Item = Self::Color> { ... }
    fn fill_solid(
        &mut self,
        area: &Rectangle,
        color: Self::Color
    ) -> Result<(), Self::Error> { ... }
    fn clear(&mut self, color: Self::Color) -> Result<(), Self::Error> { ... }
}
Expand description

A target for embedded-graphics drawing operations.

The DrawTarget trait is used to add embedded-graphics support to a display driver or similar targets like framebuffers or image files. Targets are required to at least implement the draw_iter method and the Dimensions trait. All other methods provide default implementations which use these methods internally.

Because the default implementations cannot use features specific to the target hardware they can be overridden to improve performance. These target specific implementations might, for example, use hardware accelerated drawing operations provided by a display controller or specialized hardware modules in a microcontroller.

Note that some displays require a “flush” operation to write changes from a framebuffer to the display. See docs associated with the chosen display driver for details on how to update the display.

Examples

Minimum implementation

In this example DrawTarget is implemented for an an imaginary 64px x 64px 8-bit grayscale display that is connected using a simplified SPI interface. Because the hardware doesn’t support any acceleration only the draw_iter method and OriginDimensions trait need to be implemented.

To reduce the overhead caused by communicating with the display for each drawing operation the display driver uses and framebuffer to store the pixel data in memory. This way all drawing operations can be executed in local memory and the actual display is only updated on demand by calling the flush method.

Because all drawing operations are using a local framebuffer no communication error can occur while they are executed and the Error type can be set to core::convert::Infallible.

use core::convert::TryInto;
use embedded_graphics::{
    pixelcolor::{Gray8, GrayColor},
    prelude::*,
    primitives::{Circle, PrimitiveStyle},
};

/// SPI communication error
#[derive(Debug)]
struct CommError;

/// A fake 64px x 64px display.
struct ExampleDisplay {
    /// The framebuffer with one `u8` value per pixel.
    framebuffer: [u8; 64 * 64],

    /// The interface to the display controller.
    iface: SPI1,
}

impl ExampleDisplay {
    /// Updates the display from the framebuffer.
    pub fn flush(&self) -> Result<(), CommError> {
        self.iface.send_bytes(&self.framebuffer)
    }
}

impl DrawTarget for ExampleDisplay {
    type Color = Gray8;
    // `ExampleDisplay` uses a framebuffer and doesn't need to communicate with the display
    // controller to draw pixel, which means that drawing operations can never fail. To reflect
    // this the type `Infallible` was chosen as the `Error` type.
    type Error = core::convert::Infallible;

    fn draw_iter<I>(&mut self, pixels: I) -> Result<(), Self::Error>
    where
        I: IntoIterator<Item = Pixel<Self::Color>>,
    {
        for Pixel(coord, color) in pixels.into_iter() {
            // Check if the pixel coordinates are out of bounds (negative or greater than
            // (63,63)). `DrawTarget` implementation are required to discard any out of bounds
            // pixels without returning an error or causing a panic.
            if let Ok((x @ 0..=63, y @ 0..=63)) = coord.try_into() {
                // Calculate the index in the framebuffer.
                let index: u32 = x + y * 64;
                self.framebuffer[index as usize] = color.luma();
            }
        }

        Ok(())
    }
}

impl OriginDimensions for ExampleDisplay {
    fn size(&self) -> Size {
        Size::new(64, 64)
    }
}

let mut display = ExampleDisplay {
    framebuffer: [0; 4096],
    iface: SPI1,
};

// Draw a circle with top-left at `(22, 22)` with a diameter of `20` and a white stroke
let circle = Circle::new(Point::new(22, 22), 20)
    .into_styled(PrimitiveStyle::with_stroke(Gray8::WHITE, 1));

circle.draw(&mut display)?;

// Update the display
display.flush().unwrap();

Hardware acceleration - solid rectangular fill

This example uses an imaginary display with 16bpp RGB565 colors and hardware support for filling of rectangular areas with a solid color. A real display controller that supports this operation is the SSD1331 with it’s “Draw Rectangle” (22h) command which this example is loosely based on.

To leverage this feature in a DrawTarget, the default implementation of fill_solid can be overridden by a custom implementation. Instead of drawing individual pixels, this target specific version will only send a single command to the display controller in one transaction. Because the command size is independent of the filled area, all fill_solid calls will only transmit 8 bytes to the display, which is far less then what is required to transmit each pixel color inside the filled area.

use core::convert::TryInto;
use embedded_graphics::{
    pixelcolor::{raw::RawU16, Rgb565, RgbColor},
    prelude::*,
    primitives::{Circle, Rectangle, PrimitiveStyle, PrimitiveStyleBuilder},
};

/// SPI communication error
#[derive(Debug)]
struct CommError;

/// An example display connected over SPI.
struct ExampleDisplay {
    iface: SPI1,
}

impl ExampleDisplay {
    /// Send a single pixel to the display
    pub fn set_pixel(&self, x: u32, y: u32, color: u16) -> Result<(), CommError> {
        // ...

        Ok(())
    }

    /// Send commands to the display
    pub fn send_commands(&self, commands: &[u8]) -> Result<(), CommError> {
        // Send data marked as commands to the display.

        Ok(())
    }
}

impl DrawTarget for ExampleDisplay {
    type Color = Rgb565;
    type Error = CommError;

    fn draw_iter<I>(&mut self, pixels: I) -> Result<(), Self::Error>
    where
        I: IntoIterator<Item = Pixel<Self::Color>>,
    {
        for Pixel(coord, color) in pixels.into_iter() {
            // Check if the pixel coordinates are out of bounds (negative or greater than
            // (63,63)). `DrawTarget` implementation are required to discard any out of bounds
            // pixels without returning an error or causing a panic.
            if let Ok((x @ 0..=63, y @ 0..=63)) = coord.try_into() {
                self.set_pixel(x, y, RawU16::from(color).into_inner())?;
            }
        }

        Ok(())
    }

    fn fill_solid(&mut self, area: &Rectangle, color: Self::Color) -> Result<(), Self::Error> {
        // Clamp the rectangle coordinates to the valid range by determining
        // the intersection of the fill area and the visible display area
        // by using Rectangle::intersection.
        let area = area.intersection(&self.bounding_box());

        // Do not send a draw rectangle command if the intersection size if zero.
        // The size is checked by using `Rectangle::bottom_right`, which returns `None`
        // if the size is zero.
        let bottom_right = if let Some(bottom_right) = area.bottom_right() {
            bottom_right
        } else {
            return Ok(());
        };

        self.send_commands(&[
            // Draw rectangle command
            0x22,
            // Top left X coordinate
            area.top_left.x as u8,
            // Top left Y coordinate
            area.top_left.y as u8,
            // Bottom right X coordinate
            bottom_right.x as u8,
            // Bottom right Y coordinate
            bottom_right.y as u8,
            // Fill color red channel
            color.r(),
            // Fill color green channel
            color.g(),
            // Fill color blue channel
            color.b(),
        ])
    }
}

impl OriginDimensions for ExampleDisplay {
    fn size(&self) -> Size {
        Size::new(64, 64)
    }
}

let mut display = ExampleDisplay { iface: SPI1 };

// Draw a rectangle with 5px red stroke and green fill.
// The stroke and fill can be broken down into multiple individual rectangles,
// so this uses `fill_solid` internally.
Rectangle::new(Point::new(20, 20), Size::new(50, 40))
    .into_styled(
        PrimitiveStyleBuilder::new()
            .stroke_color(Rgb565::RED)
            .stroke_width(5)
            .fill_color(Rgb565::GREEN)
            .build(),
    )
    .draw(&mut display)?;

// Draw a circle with top-left at `(5, 5)` with a diameter of `10` and a magenta stroke with
// cyan fill. This shape cannot be optimized by calls to `fill_solid` as it contains transparent
// pixels as well as pixels of different colors. It will instead delegate to `draw_iter`
// internally.
Circle::new(Point::new(5, 5), 10)
    .into_styled(
        PrimitiveStyleBuilder::new()
            .stroke_color(Rgb565::MAGENTA)
            .stroke_width(1)
            .fill_color(Rgb565::CYAN)
            .build(),
    )
    .draw(&mut display)?;

Required Associated Types§

source

type Color: PixelColor

The pixel color type the targetted display supports.

source

type Error

Error type to return when a drawing operation fails.

This error is returned if an error occurred during a drawing operation. This mainly applies to drivers that need to communicate with the display for each drawing operation, where a communication error can occur. For drivers that use an internal framebuffer where drawing operations can never fail, core::convert::Infallible can instead be used as the Error type.

Required Methods§

source

fn draw_iter<I>(&mut self, pixels: I) -> Result<(), Self::Error>where I: IntoIterator<Item = Pixel<Self::Color>>,

Draw individual pixels to the display without a defined order.

Due to the unordered nature of the pixel iterator, this method is likely to be the slowest drawing method for a display that writes data to the hardware immediately. If possible, the other methods in this trait should be implemented to improve performance when rendering more contiguous pixel patterns.

Provided Methods§

source

fn fill_contiguous<I>( &mut self, area: &Rectangle, colors: I ) -> Result<(), Self::Error>where I: IntoIterator<Item = Self::Color>,

Fill a given area with an iterator providing a contiguous stream of pixel colors.

Use this method to fill an area with contiguous, non-transparent pixel colors. Pixel coordinates are iterated over from the top left to the bottom right corner of the area in row-first order. The provided iterator must provide pixel color values based on this ordering to produce correct output.

As seen in the example below, the PointsIter::points method can be used to get an iterator over all points in the provided area.

The provided iterator is not required to provide width * height pixels to completely fill the area. In this case, fill_contiguous should return without error.

This method should not attempt to draw any pixels that fall outside the drawable area of the target display. The area argument can be clipped to the drawable area using the Rectangle::intersection method.

The default implementation of this method delegates to draw_iter.

Examples

This is an example implementation of fill_contiguous that delegates to draw_iter. This delegation behaviour is undesirable in a real application as it will be as slow as the default trait implementation, however is shown here for demonstration purposes.

The example demonstrates the usage of Rectangle::intersection on the passed area argument to only draw visible pixels. If there is no intersection between area and the display area, no pixels will be drawn.

use embedded_graphics::{
    pixelcolor::{Gray8, GrayColor},
    prelude::*,
    primitives::{ContainsPoint, Rectangle},
};

struct ExampleDisplay;

impl DrawTarget for ExampleDisplay {
    type Color = Gray8;
    type Error = core::convert::Infallible;

    fn draw_iter<I>(&mut self, pixels: I) -> Result<(), Self::Error>
    where
        I: IntoIterator<Item = Pixel<Self::Color>>,
    {
        // Draw pixels to the display

        Ok(())
    }

    fn fill_contiguous<I>(&mut self, area: &Rectangle, colors: I) -> Result<(), Self::Error>
    where
        I: IntoIterator<Item = Self::Color>,
    {
        // Clamp area to drawable part of the display target
        let drawable_area = area.intersection(&self.bounding_box());

        // Check that there are visible pixels to be drawn
        if drawable_area.size != Size::zero() {
            self.draw_iter(
                area.points()
                    .zip(colors)
                    .filter(|(pos, _color)| drawable_area.contains(*pos))
                    .map(|(pos, color)| Pixel(pos, color)),
            )
        } else {
            Ok(())
        }
    }
}

impl OriginDimensions for ExampleDisplay {
    fn size(&self) -> Size {
        Size::new(64, 64)
    }
}
source

fn fill_solid( &mut self, area: &Rectangle, color: Self::Color ) -> Result<(), Self::Error>

Fill a given area with a solid color.

If the target display provides optimized hardware commands for filling a rectangular area of the display with a solid color, this method should be overridden to use those commands to improve performance.

The default implementation of this method calls fill_contiguous with an iterator that repeats the given color for every point in area.

source

fn clear(&mut self, color: Self::Color) -> Result<(), Self::Error>

Fill the entire display with a solid color.

If the target hardware supports a more optimized way of filling the entire display with a solid color, this method should be overridden to use those commands.

The default implementation of this method delegates to fill_solid to fill the bounding_box returned by the Dimensions implementation.

Implementors§