Hi!

I am attempting to write an sub-routine that will read a pixel from a bitmap based on the position of the mouse and give me a return value based on the color of that pixel. here is what I have so far:

function get_pixel_color()
{var x_coord = integer(mouse_pos.x*(bmap_width(my_bmap)/screen_size.x));
var y_coord = integer(mouse_pos.y*(bmap_height(my_bmap)/screen_size.y));
var pixel;
COLOR* pixel_color;
if(x_coord>bmap_width(my_bmap)||x_coord<0) { return(-1); }
if(y_coord>bmap_height(my_bmap)||y_coord<0) { return(-1); }
bmap_lock(my_bmap,0);
pixel = pixel_for_bmap(my_bmap,x_coord,y_coord);
bmap_unlock(my_bmap);
pixel_to_vec(pixel_color,NULL,8888,pixel);
if(pixel_color.red)
{ if(pixel_color.red>245&&pixel_color.red<255)
{ return(0);
}
}
return(-1)
}

The routine works fine - x_coord and y_coord are always co-ordinates within the bmap and pixel_for_bmap() is returning the pixel it is supposed to. However, when I execute the pixel_to_vec() command, the game crashes.

I tested the return value from bmap_lock and it is 8888, indicating that it is locking a 32-bit tga file with an alpha channel - which is curious becuse the bmap im reading from was created in photshop and saved as a 24-bit tga file with no alpha channel.

Am I feeding pixel_to_vec a bad pixel value? or maybe im feeding some bad value for one of the other parameters? Is bmap_lock misreading the bmap and thats whats causing problems? why would pixel_to_vec() crash the engine like this?

Have you tried it also with bmp?

I will try that and let you know

Yes...my_bmap.bmp crashes on the pixel_to_vec() just the same as my_bmap.tga.

1st guess... you've using .red instead of .blue (i know... thank you Conitec!) i'll have a look more if that doesn't solve it

Sorry....i dont understand, .red instead of .blue? I'm feeding pixel_color to the function, not pixel_color.red or pixel_color.blue. I use if(pixel_color.red) after calling pixel_to_vec() to determine if there is any red in the pixel that is read from the bitmap. The program is crashing when it calls pixel_to_vec(), not when it tries to evaluate if(pixel_color.red).

sorry only read half the thread.

Should
var x_coord = integer(mouse_pos.x*(bmap_width(my_bmap)/screen_size.x));
be
var x_coord = integer(mouse_pos.x+[(bmap_width(my_bmap)/screen_size.x));

I'm using a function (hopefully) similar to what you're trying to achieve1, 2 and 3 chooses the panel clicked and the limits of the bmap while holding mouse_left

i'm reading from a bmap that is not displayed on screen, therefore the mouse co-ordinates have to be converted to co-ordinates on the bmap. the bmap i am reading from is 744x600 pixels, the mouse position x (lets say the video_mode is 8 and the screen size therefore 1024x768) is between 0 and 1024 and the mouse position y is between 0 and 768. Dividing bmap_width(my_bmap) by screen_size.x gives me the ratio i need to convert the mouse pos_x to the equivalent x co-ordinate on the bmap. lets say the mouse is halfway across the screen - mouse_pos.x = 512.

x_coord = integer(mouse_pos.x*(bmap_width(my_bmap)/screen_size.x));
x_coord = integer(512*(744/1024));
x_coord = integer(512*0.727);
x_coord = 372;

mouse_pos.x of 512 = bitmap_pos.x of 372. Even when this formula generates a number outside the bounds of the bitmap i am still boundary checking and force an early return in this instance. (if(x_coord>bmap_width(my_bmap)||x_coord<0) { return(-1); }). The same is for y_coord.

I know that x_coord and y_coord are giving me the numbers im expecting. I know that bmap_lock is sucessful - it returns 8888. I know that the pixel for bmap is returning the pixel read from the bmap....well, i know it is returing *A* value...perhaps the value returned by pixel_for_bmap is the problem? or perhaps the format returned by bmap_lock is the problem?

I mentioned before that the bmap was made in photoshop and saved as a 24 bit tga file with no alpha channel - a return code of 8888 means it is reading the bmap as a 32 bit tga file with an alpha channel. I recall reading that the bmap_lock "upconverts" a bmap to the next compatible mode if its native mode isnt compatible or something like that.

I notice you save the return value of bmap_lock to format and pass format to the pixel_to_vec function instead of passing 8888. I will try that and see if that solves the problem

Ah....found the problem.

"COLOR* pixel_color;" needs to be "COLOR pixel_color;". I'm passing a pointer to a color instead of the color vector itself - THATS what was crashing pixel_to_vec().

Dumb, dumb, dumb me. every other capitalised variable declaration requires a * (PANEL*, TEXT*, BMAP*, FONT*, etc etc). I know not ALL of them do though, I need to learn to check these things.

Thank you everyone for your time and input.