loading a 256 color image

[ironwolfe]

I'm trying to load a 256 color picture into a 1289,720,256 screen but i keep getting "illegal function call" when i _putimage. I suspect its the picture itself, what is the best way to convert images to 256 colors so they are compatible with QB64.

here is a copy of the code.

DIM pic&

SCREEN _NEWIMAGE(1280, 720, 256)





temp


SUB temp


    pic& = _LOADIMAGE("demon.bmp")
    '_COPYPALETTE backgrnd1, 0
    _PUTIMAGE (3, 3), pic&



    SLEEP

END SUB

[Petr]

Hi. This is a known (and long-term) limitation of the LOADIMAGE command. Do you expect that if you load an image on your 256 color screen, will it change its color palette? There is a way despite the limitations of LOADIMAGE to get a 256-color image on a 256 color screen. It's called the Floyd-Steinberg method, Ashish once wrote a beautiful sample and beautifully functional program that translates it from 32 bits to 256 colors (8 bit) or more, from 256 colors (8 bit) to 16 colors (4 bit).


This what I give here, is for images that are already converted to 256 colors!

Code: QB64: [Select]

1. a = _LOADIMAGE("cat.bmp", 256)
2. PRINT a
3. PRINT "As you see, LOADIMAGE return -1, this is invalid handle error flag."
4. PRINT "Try using my nethod: (press key)"
5. SLEEP
6.  
7. SCREEN _NEWIMAGE(1024, 768, 256)
8. i32to256 "cat.bmp", 100, 100 'filename, X position, Y position
9.  
10.  
11.  
12.  
13. SUB i32to256 (image AS STRING, x AS INTEGER, y AS INTEGER) '     this is already on the .NET forum writed by me. As example how show pictures in 256 colors.
14.     IF _FILEEXISTS(image$) THEN
15.         image& = _LOADIMAGE(image$, 32) '                       256 colors image can be loaded as 32 bit image
16.         TYPE colors
17.             ClrVal AS LONG '                                    this contais color number in long format (_RGB32)
18.             ClrNmbr AS LONG '                                   this contais number for color. How much is this one color used in picture. Is for future use, if 32bit image contais more than 256 colors, then
19.         END TYPE '                                               i will use the most used only.
20.         REDIM colors(256) AS colors
21.         REDIM scn AS LONG, col AS LONG, scan AS LONG, control AS LONG, TotalColors AS LONG
22.         REDIM m AS _MEM
23.         m = _MEMIMAGE(image&)
24.  
25.         FOR scan = 0 TO (_WIDTH(image&) * _HEIGHT(image&) * 4) - 4 STEP 4 ' use 32 bit, step is 4 byt * 8 bit = 32 bit, i read 4 bytes (LONG) in one loop, so STEP 4
26.             _MEMGET m, m.OFFSET + scan, col&
27.             FOR control = 0 TO TotalColors&
28.                 IF col& = colors(control).ClrVal THEN colors(control).ClrNmbr = colors(control).ClrNmbr + 1: col& = 0: EXIT FOR
29.             NEXT
30.             IF col& <> 0 THEN colors(control + 1).ClrVal = col&: colors(control + 1).ClrNmbr = 1: TotalColors& = TotalColors& + 1: col& = 0
31.             IF TotalColors& > 255 THEN EXIT FOR
32.         NEXT scan
33.         IF TotalColors& <= 256 THEN
34.             image256& = _NEWIMAGE(_WIDTH(image&), _HEIGHT(image&), 256)
35.             _DEST image256&
36.             DIM m2 AS _MEM
37.             m2 = _MEMIMAGE(image256&)
38.             FOR MESecam = 255 - TotalColors& TO 255
39.                 _DEST 0
40.                 _PALETTECOLOR MESecam, colors(255 - MESecam).ClrVal
41.             NEXT
42.  
43.             REDIM SelectColor AS _UNSIGNED _BYTE
44.             FOR scan = 0 TO (_WIDTH(image&) * _HEIGHT(image&) * 4) - 4 STEP 4
45.                 _MEMGET m, m.OFFSET + scan, Value&
46.                 FOR SelectColor = 255 - TotalColors& TO 255
47.                     IF colors(255 - SelectColor).ClrVal& = Value& THEN _MEMPUT m2, m2.OFFSET + position256, SelectColor
48.                 NEXT SelectColor
49.                 position256 = position256 + 1
50.             NEXT scan
51.             _PUTIMAGE (x, y), image256&, 0
52.             _MEMFREE m: _MEMFREE m2: _FREEIMAGE image&: _FREEIMAGE image256&
53.         ELSE PRINT "Image contains more than 256 colors."
54.         END IF
55.     ELSE PRINT "File "; image$; " not exists.": SLEEP 5
56.     END IF
57. END SUB
58.  

[RhoSigma]

1. _LOADIMAGE currently always loads images in 32-bits mode, even 256 color images (256 color loading is disabled since a while, as it didn't work reliable since the transition from SDL to OpenGL back in 2012/13). You cannot _PUTIMAGE a 32-bits image to a 256 (8-bits) SCREEN, obviously.

2. Are you sure you want a 256 colors SCREEN? If it's only to be in line with the image's colordepth, then forget it, simply use a 32-bits SCREEN instead, then you can _PUTIMAGE to it as is.

3. If you really want a 256 colors SCREEN, then let me know and i'll provide you with my Floyd-Steinberg remapper function, which can reduce any image to a given target palette.

BTW  - Just see, Petr was faster than me, but the example he posted is NOT Floyd-Steinberg, it's a simple quantitative color reduction, which gives much less display quality than real F-S remapping. See the screen shot pictures here: https://www.qb64.org/forum/index.php?topic=809.msg100182#msg100182, that's real F-S quality.

[Petr]

Here is Floyd - Steinberg dithering by Vince: https://www.qb64.org/forum/index.php?topic=338.msg2271#msg2271


As you can see, the resulting image is stored on a 32-bit screen. To save it after converting to 256 color screen, you need to set 256 color screen to the correct color palette.

[SMcNeill]

Aren’t you guys overthinking this?  If you load a 256 color image onto a 32-bit screen (which is what _LOADIMAGE current does), then all you need to do is a simple read-and-convert loop to put it on a 256 color screen.  No need for dithering, or anything fancy...  Just (pseudocode follows as I’m not at my PC currently)

Load the 256 color image to a 32-bit screen.
Use the 32-bit screen as a source, a 256 color screen as a destination
DIM Palette(255) ‘to hold the 255 color image palette
FOR x = 0 TO _WIDTH - 1
    FOR y = 0 TO _HEIGHT -1
        p = POINT(x,y)
        FOR i = 0 to 255
             IF p = Palette(i) THEN
                  PSET (x,y), i
                  EXIT FOR
             ELSEIF Palette(i) = 0 THEN ‘new palette color found
                   _PALETTECOLOR i, p
                   Palette(i) = p ‘store the new color
                   PSET (x,y), i
                   EXIT FOR
             END IF
         NEXT
    NEXT
NEXT

[RhoSigma]

Aren’t you guys overthinking this?  If you load a 256 color image onto a 32-bit screen (which is what _LOADIMAGE current does), then all you need to do is a simple read-and-convert loop to put it on a 256 color screen.  No need for dithering, or anything fancy...  Just (pseudocode follows as I’m not at my PC currently)

In fact, you are true here. We are thinking to far according to our own skills and experiences. If we limit the whole thing to 256 color images, as the original question was about, then no dithering or color reduction is needed, as 256 colors always remain to be 256 colors, even if put to a 32-bit screen. Thanks that you focus us back to the original question.

[Petr]

Aren’t you guys overthinking this?  If you load a 256 color image onto a 32-bit screen (which is what _LOADIMAGE current does), then all you need to do is a simple read-and-convert loop to put it on a 256 color screen.  No need for dithering, or anything fancy...  Just (pseudocode follows as I’m not at my PC currently)

:-D Hi. About which condition i write in first reply? How works my source? Dithering is need for images, which use more than 256 colors. I add here Dithering link just for interest.

[Raven_Singularity]

It would be nice if QB64 supported loading 8bit images to an 8bit screen.  The two simplest ways to do this are adopting the palette of the image, or when loading the image rematching the palette to the closest colour that already exists in the screen palette.  These require minimal coding.

There's also 7, 6, 5, 4, 3, 2, 1bit images, which could leave part of the screen palette alone.

That said, I can't imagine why a new programmer like ironwolfe is wanting to use a 256 colour palette screen on a modern PC.  ironwolfe, just use a 32bit screen and be done with it!  :-)

[Cobalt]

It would be nice if QB64 supported loading 8bit images to an 8bit screen.  The two simplest ways to do this are adopting the palette of the image, or when loading the image rematching the palette to the closest colour that already exists in the screen palette.  These require minimal coding.

There's also 7, 6, 5, 4, 3, 2, 1bit images, which could leave part of the screen palette alone.

That said, I can't imagine why a new programmer like ironwolfe is wanting to use a 256 colour palette screen on a modern PC.  ironwolfe, just use a 32bit screen and be done with it!  :-)

there are a few old skool tricks you can't do with 32bit, like palette shifting for one.

[Raven_Singularity]

Palette shifting is the only advantage I know of, but ironwolfe was not trying to do that.

[moises1953]

There is another posible solution: map the 32 bit colors to a 256 color sorted palette,
so no need to store readed colors, and can load full coloured images (more or less).
In this example the 0-15 color remains untouched for compatibility with old VGA mode 12.

Code: QB64: [Select]

1. CONST resx = 1024, resy = 768
2. scr& = _NEWIMAGE(resx, resy, 256)
3. SCREEN scr&
4. _FULLSCREEN
5. CALL ImgCreate8bitPalette
6. imgfile$ = "D:\FQB64\QB64.PNG"
7. px1 = 384: py1 = 80
8. IF ImgLOAD(imgfile$, scr&, px1, py1) = 0 THEN PRINT "Can't load "; imgfile$
9. END
10.  
11. FUNCTION ImgLOAD% (file$, scr&, px, py)
12.   DIM r AS INTEGER, g AS INTEGER, b AS INTEGER, p AS _UNSIGNED LONG
13.   tmp& = _LOADIMAGE(file$, 32)
14.   IF tmp& >= -1 THEN
15.     ImgLOAD = 0
16.   ELSE
17.     ImgLOAD = -1
18.     _SOURCE tmp&
19.     '_DEST scr&
20.     FOR ix = 0 TO _WIDTH(tmp&) - 1
21.       FOR iy = 0 TO _HEIGHT(tmp&) - 1
22.         p = POINT(ix, iy)
23.         r = _RED32(p)
24.         g = _GREEN32(p)
25.         b = _BLUE32(p)
26.         IF r OR g OR b THEN
27.           map = (b + 25) \ 51 + ((g + 25) \ 51) * 6 + ((r + 25) \ 51) * 36 ' map to sorted palette
28.           ipal = map + 16
29.         ELSE
30.           ipal = 0
31.         END IF
32.         PSET (px + ix, py + iy), ipal
33.       NEXT iy
34.     NEXT ix
35.     _SOURCE scr&
36.     _FREEIMAGE tmp&
37.   END IF
38. END FUNCTION
39.  
40. SUB ImgCreate8bitPalette
41.   '.standard 16 colors VGA palette untouched (0-15)
42.   '.Map 256->6 colors for 6^3=216 new colors, spaced 255\5=51: 0,51,102,153,204,255
43.   '.Final palette: 15 + 216 = 231 last palette color
44.   i = 15
45.   FOR ir = 0 TO 255 STEP 51
46.     FOR ig = 0 TO 255 STEP 51
47.       FOR ib = 0 TO 255 STEP 51
48.         i = i + 1
49.         _PALETTECOLOR i, _RGB32(ir, ig, ib)
50.       NEXT
51.     NEXT
52.   NEXT
53. END SUB
54.