4 bit texture bug

[Bart]

I haven't really followed your progress on 4-bit textures at all, so I'm probably missing something fundamental, but when Nik and I had first looked at them, we thought them to be a 4-bit planar format, packing 4 indpendent images into each 16 bit word. For example, this single decal texture contains four images:






If I understand correctly, you are treating them as merely A4L4?

[HarryTuttle]

Bart, the current implementation in fine, in my personal build I just changed those texture format in <Texture.cpp> to this:

Code: Select all

      //
      // 4 bit texture types - all luminance textures (no alpha), only seem to be enabled when contour is enabled ( white = contour value )
      //

   case 8: // low byte, low nibble
      for (yi = y; yi < (y + height); yi++)
      {
         for (xi = x; xi < (x + width); xi++)
         {
            texel = src[yi * 2048 + xi] & 0xFF;
            c = (texel & 0xF) * 17;
            scratch[i++] = c;
            scratch[i++] = c;
            scratch[i++] = c;
            scratch[i++] = c;
         }
      }
      break;

   case 9:   // low byte, high nibble
      for (yi = y; yi < (y + height); yi++)
      {
         for (xi = x; xi < (x + width); xi++)
         {
            texel = src[yi * 2048 + xi] & 0xFF;
            c = ((texel >> 4) & 0xF) * 17;
            scratch[i++] = c;
            scratch[i++] = c;
            scratch[i++] = c;
            scratch[i++] = c;
         }
      }
      break;

   case 10:   // high byte, low nibble
      for (yi = y; yi < (y + height); yi++)
      {
         for (xi = x; xi < (x + width); xi++)
         {
            texel = src[yi * 2048 + xi] >> 8;
            c = (texel & 0xF) * 17;
            scratch[i++] = c;
            scratch[i++] = c;
            scratch[i++] = c;
            scratch[i++] = c;
         }
      }
      break;

   case 11: // high byte, high nibble

      for (yi = y; yi < (y + height); yi++)
      {
         for (xi = x; xi < (x + width); xi++)
         {
            texel = src[yi * 2048 + xi] >> 8;
            c = ((texel >> 4) & 0xF) * 17;
            scratch[i++] = c;
            scratch[i++] = c;
            scratch[i++] = c;
            scratch[i++] = c;
         }
      }
      break;
   }


and inverted the alpha channel in fragment shader.

[Ian]

Does that work for Harley ?

[Bart]

Ah, ok. So you are indeed interpreting it the same way and you've managed to figure out when the different format codes apply to the different planes, which is something Nik and I had trouble with I recall.

[HarryTuttle]

Does that work for Harley ?

Apparently yes:

hl_02.jpeg (235.67 KiB) Viewed 6707 times

[HarryTuttle]

Ah, ok. So you are indeed interpreting it the same way and you've managed to figure out when the different format codes apply to the different planes, which is something Nik and I had trouble with I recall.

In particular: for types 8-11 I've just changed the alpha channel in <Texture.cpp> from this:

Code: Select all

scratch[i++] = (c == 255 ? 0 : 255);

to this:

Code: Select all

scratch[i++] = c;

[Ian]

Well .. this definitely looks better



Gonna test the other games

[HarryTuttle]

Ian, I don't want to sound pedantic but there's something still wrong: it's missing the black shadow underneath the number and around the text.

However, i just compiled the official build and is fine, don't know what's wrong with that image...

[Ian]

This doesn't though
Maybe there is an actual alpha inversion bit as well ..

[HarryTuttle]

What's the texture type of that fence?