Un naufragio personal

Ángel Ortega

The QDGDF Cookbook

This document explains some tricks to be used with the Quick and Dirty Game Development Framework (QDGDF) libraries.

Blending two images into one

You can easily blend two images into one by using the qdgdfv_blend_color() function. This code uses it:

 /* these buffers hold the screen images */
 unsigned char image1[SCREEN_SIZE];
 unsigned char image2[SCREEN_SIZE];
 void blend_two_screens(void)
	int n;
	for(n = 0; n < SCREEN_SIZE; n++) {
		_qdgdfv_virtual_screen[n] =
			qdgdfv_blend_color(image1[n], image2[n], 50);


In this case, the final image is a 50% blend of both images. You can also implement a fade out / fade in (as those shown in films) by looping the percent argument from 0 to 100. Don't forget though that qdgdfv_blend_color() is slow; you can pre-generate lookup tables of blended colors (for, say, 20%, 40%, 60% and 80% blend factors) as in the following example:

 /* pre-generated blend tables */
 unsigned char _blend_tbl_20[256][256];
 unsigned char _blend_tbl_40[256][256];
 unsigned char _blend_tbl_60[256][256];
 unsigned char _blend_tbl_80[256][256];
 void pregen_blend_tables(void)
	 int x, y;
	 for(x = 0; x < 256; x++)
		 for(y = 0; y < 256; y++)
			 _blend_tbl_20[x][y] = qdgdfv_blend_color(x, y, 20);
	 for(x = 0; x < 256; x++)
		 for(y = 0; y < 256; y++)
			 _blend_tbl_40[x][y] = qdgdfv_blend_color(x, y, 40);
	 for(x = 0; x < 256; x++)
		 for(y = 0; y < 256; y++)
			 _blend_tbl_60[x][y] = qdgdfv_blend_color(x, y, 60);
	 for(x = 0; x < 256; x++)
		 for(y = 0; y < 256; y++)
			 _blend_tbl_80[x][y] = qdgdfv_blend_color(x, y, 80);

Then you can use those tables instead of qdgdfv_blend_color(). Consider also pregenerating all 100 lookup tables if you plan to intesively use color blending (they will use 256 * 256 * 100 bytes, 6400 Kb).

You can achieve similar fading and smoothing effects by using the accumulation buffer; see below.

Building and using light tables

Light tables are built to easily apply a light level to a group of colors. QDGDF includes two functions to create them:

 /* basic light table */
 void qdgdfv_build_light_table(unsigned char * lut, int levels, int mid);
 /* coloured light table */
 void qdgdfv_build_light_table_ext(unsigned char * lut, int levels,
	int mid, int fr, int fg, int fb);

In both cases, the lut argument must point to a buffer like

 #define LIGHT_LEVELS 16
 unsigned char lut[LIGHT_LEVELS][256];

Of course, the light levels can be whatever you want.

The mid argument is the middle light level. This light level is the one in which colors are exactly as defined in the palette, that is, 100% lit. Any light level below mid will contain colors slightly dimmed (to the 0 level, where they are all almost black) and any light level from mid will contain saturated colors. If you set mid to LIGHT_LEVELS - 1, you won't have saturated colors if you don't need them.

So, to create a normal light table you should use

 qdgdfv_build_light_table(unsigned char *)&lut,

and get color with different light levels doing

 unsigned char color;
 unsigned char lighted_color;
 /* pixel in the darkness */
 lighted_color = lut[0][color];
 /* pixel in the twilight zone */
 lighted_color = lut[7][color];
 /* pixel completely lit */
 lighted_color = lut[15][color];

This basic light tables can be used for depth cueing effects if you calculate the light level as inversely proportional to the distance of the pixel.

All these examples use a black light table, that is, colors with light applied on them look as they are, and colors without it fade to black. To achieve interesting effects as fogs or hazes, you can use a light table that fades to another color. If you want to use white fog, you can just use the following function:

 qdgdfv_build_light_table_ext(unsigned char *)&lut,
	LIGHT_LEVELS, LIGHT_LEVELS - 1, 255, 255, 255);

which is the same as qdgdfv_build_light_table(), but telling it what color should have the pixels without light. Of course, all three r, g, b factors can be different. In any other case, this light table can be used exactly the same.

Using the accumulation buffer

The accumulation buffer can be used to implement motion blur or object movement trails. It's set by the function

 void qdgdfv_set_accum_buffer(int percent);

By default, no accumulation buffer is active. When an accumulation buffer is created by calling this function with an argument different from 0, anytime the virtual screen is dumped the previously drawn frame is blended with the current one using percent as a blending ratio. As a result, the previous frame seems to remain blurred, intermixed with the content of the virtual screen. This makes transitions smoother, but can be confusing if percent is high (values over 50% tend to blur too much or saturate to bright colors). Using the accumulation buffer with a high percent when the game's main character is poisoned or motion sick looks great if an overall game speed reduction is also done.

Gamma correction

Sometimes the combination of monitor screen, room ambient light and hardware quality seems to make the game view too dark. The overall brightness of the generated colors can be controlled with the following variable:

 int _qdgdfv_gamma_correct;

This variable ranges from 0 (colors are the same as defined in the palette) to 255 (completely saturate). It's a good idea to ask the user to set this variable with a menu, slider or similar widget. Take into account that you must re-set the palette using qdgdfv_set_palette() for the value in _qdgdfv_gamma_correct to be effective.

KTL font file format

The KTL fonts use an ancient format that I used in KGE (Kaplan Graphical Environment), an old MSDOS library for embedded graphical environments. It had many special purpose resources, focused mainly in small memory footprints and ease of use. KTL fonts have an 8 bytes header like this

 struct ktl_header
	unsigned char height;	/* font height in pixels */
	unsigned char width;	/* font width in bytes (1) */
	unsigned char prop;	/* proportional flag (0) */
	char filler[5]; 	/* filler to 8 bytes */

followed by 256 glyphs of height * width bytes for the 256 available characters, coded in the PC VGA charset. Instead of what it can be told from the header, there are no proportional (prop is always 0) nor width other than 1 (eight pixels) fonts.

QDGDF remaps the charset transparently so that ISO-8859-1 (latin1) encoding is used in the string printing code (qdgdfv_font_print()).

Angel Ortega <angel@triptico.com>