Sprite Image (.pl8)

.pl8 files are basically bitmaps, with some additional features bolted on. There are multiple images (sprites) within one file. The height, width, and data offset are located in the header.

The color table is stored in a separate file.

A type of “transparency” is supported, despite the color table not supporting an alpha channel.

The image data is located at the ‘Data Offset’, and is Width x Height long. Each byte corresponds to a color in the color table (.256 file).

File Offset (bytes)
0 - 1	0bit indicates RLS encoding, the rest is unknown
2 - 3	Number of Sprites
4 - 7	Unknown
	Sprite 1
8 - 9	Width
10 - 12	Height
13 - 16	Data Offset
17 - 18	x coordinate
20 - 21	y coordinate
22	Tile type 0 - simple bitmap like encoding 1 - isometric (ISO) encoding 2 - ISO encoding with additional rows 3 - ISO encoding additional rows on the left side only 4 - ISO encoding additional rows on the right side only
23	Indicates how many extra row is stored in ISO encoded tile
24 - 25	Unknown
	Sprite 2

Simple bitmap like encoding

Each byte from given offset, is simply index to pallet and no more manipulation is needed. Just take data from tile.offset until tile.offset + tile.width * tile.height

Run-length encoding (RLE)

RLE uses small chunks to encode number of multiple consequent transparent or number of consequent opaque pixels.

First byte in each chunk indicates how many opaque pixels is in this chunk. If the number of opaque pixels is zero, next byte indicates how many transparent pixels should be filed in the image.

File Offset (bytes)	Color
1	Number of opaque pixels
2	Number of transparent pixels or index to pallet
3-N	Indices to pallet

Isometric encoding (ISO)

Isometric images are stored by rows. Each row is centred and it’s width is dependant on the Y position.

Row width can be calculated this way:

row_width = y < half_heing ? (y * 4) + 2 : (height - half_heing) * 5 + 2;

Then there are additional rows. They add additional row on the top of the image, but they are stored diagonally from the middle of left side to the middle of top side and from the middle of top to to middle of the right side. See following image for better understanding.

Based on Tile type the extra rows can be on left side, right or on both sides.

Iso tile in its entirety can be parsed with code similar to this:

std::ifstrem file;

...

file.seekg(tileHeader->offset);

// Fill top half
for (int y = 0; y < halfHeight; ++y)
{
    int rowStart = (halfHeight - 1 - y) * 2;
    int rowStop = rowStart + (y * 4) + 2;

    for (int x = rowStart; x < rowStop; ++x)
    {
        uint8_t palletIndex = file.get();
        tile.pixels[((y + tileHeader->extraRows) * tileHeader->width) + x] = palletIndex;
    }
}
// Fill bottom half
for (int y = halfHeight; y < tileHeader->height; ++y)
{
    int rowStart = (halfHeight - 1 - (tileHeader->height - y - 1)) * 2;
    int rowStop = rowStart + ((tileHeader->height - y -1) * 4) + 2;

    for (int x = rowStart; x < rowStop; ++x)
    {
        uint8_t palletIndex = file.get();
        tile.pixels[((y + tileHeader->extraRows) * tileHeader->width) + x] = palletIndex;
    }
}

// Fill extra rows
for (int y_ = tileHeader->extraRows; y_ > 0; --y_)
{
    int rightOffset = tileHeader->extraType == 3
        ? halfWidth + 1 : tileHeader->width;
    int leftOffset = tileHeader->extraType == 4
        ? halfWidth - 1
        : 0;

    for (int x = leftOffset; x < rightOffset; ++x)
    {
        int y = x <= halfWidth
            ?  y_ + (halfHeight - 1) - (x / 2)
            : y_ + (x / 2) - (halfHeight - 1);

        uint8_t palletIndex = file.get();
        tile.pixels[(y * tile.width) + x] = palletIndex;
    }
}