7-color e-paper dithering

Seven-color panels add orange to black, white, green, blue, red and yellow. The extra warm color helps with skin tones, sunsets and wood.

Palette and color codes

The code is the value DitherInk writes for each pixel in .bin, .h, .png and .bmp files. The preview color is the brightened color used for matching and on-screen previews. The index color is the pure color stored in the PNG and BMP palette.

Frame buffer format

4 bits per pixel (color_4bpp). 2 pixels per byte, the high nibble is the left pixel. The value is the color code. Rows are stored top to bottom and each row is padded to a whole byte.

Supported sizes

Convert an image to 7-color →

Tips

Each pixel is a 4-bit color code, two pixels per byte. Check your driver's color code table: DitherInk writes the codes listed below, and you can change them under Advanced output options.

Example code (7.3" 800×480)

Arduino / ESP-IDF with the .h file

#include "my_image.h"  // downloaded from DitherInk

// MY_IMAGE_WIDTH x MY_IMAGE_HEIGHT = 800 x 480, color_4bpp
// Color codes: black=0, white=1, green=2, blue=3, red=4, yellow=5, orange=6
// The buffer is already in the panel's native orientation.

void showImage() {
  epd_write_frame(my_image, MY_IMAGE_SIZE);  // your driver's full-frame write
  epd_refresh();
}

Replace epd_write_frame and epd_refresh with the functions your display driver provides. Most drivers have a function that takes a full frame buffer.

ESP32 loading the .bin file from flash

#include <LittleFS.h>

constexpr size_t FRAME_BYTES = 192000;
// Large buffer: allocate in PSRAM on ESP32 boards that have it.
uint8_t* frame = (uint8_t*)ps_malloc(FRAME_BYTES);

bool loadFrame(const char* path) {
  File f = LittleFS.open(path, "r");
  if (!f || f.size() != FRAME_BYTES) return false;  // wrong size = wrong panel settings
  f.read(frame, FRAME_BYTES);
  f.close();
  return true;
}

Python (Raspberry Pi) reading the .bin file

import numpy as np

data = np.fromfile("my_image.bin", dtype=np.uint8)
assert data.size == 192000
codes = np.stack([data >> 4, data & 0x0F], axis=-1).reshape(480, -1)[:, :800]