Context: I'm currently re-engineering an old (2015) E-paper display tag just for the fun of it. For the curious ones, I'm using Ghidra to disassemble and decompile, and trying to make up some sense of the results. The tag uses an ARM Cortex M0 based MCU and a 1.44 inch EPD, and receives its display content via NFC.
After receiving some (RLE compressed) KB of data and inflating that data, the application interprets the data. This is my current knowledge about the contents:
There is potentially a header of 16 bytes ignored by the application.
After the header there is an undetermined number of elements. The application stops at the end of the received data.
Each element is marked with a single byte (byte 0 in the table), followed by a length byte (byte 1 in the table, counted without these first two bytes)
After these two bytes more bytes can follow, depending on the element's type:
My finding so far byte 0
(element
type)byte 1
(# of bytes
to follow)byte 2 byte 3 byte 4 byte 5 byte 6 byte 7 ... set geometry 0x01 3 x offset y offset rotation
(0..3 = 0°/ 90°/ 180°/ 270°)- - - define colors 0x02 2 color1 color2 - - - - fill completely
with background color0x03 0 - - - - - - invert image 0x04 0 - - - - - - draw line 0x11 4 x1 y1 x2 y2 - - draw text 0x12 # x y font opaque centered text... draw pixels 0x13 # x y w h bits per pixel data... draw rectangle 0x14 5 x1 y1 x2 y2 filled - draw circle 0x15 4 x y radius filled - - stamp image 0x17 3 x y index - - - prepare pixel data 0x20 # data... - - - -
Element type 0x20 is special. Each data chunk is appended to a buffer. When finally element type 0x13 is used, its data are also appended, and then the whole buffer is the image to draw.
I spent several hours in web research but could not find any hint. The "usual suspects" like CGM, EMF, EPS, SVG, WMF or Gerber do not match. I looked into few others, too.
What (file) format of a (simple) vector graphic is this?
The application is a mix of apparently assembler, C, and C++ code. Specifically all drawing functions receive the same pointer, which is certainly this. There is also a function pointer table in the structure pointed to, known as vtable. This makes me conclude that libraries are used.
Currently I'm at a bit-block transfer function that supports 16 colours. It can be that this is a proprietary format, but it makes no sense to support more than the two colours of EPDs of that time. That's why I think it is some more generic format.
