ref: e1b1b5422c00453455c47e71ff976adfc3ba6f53
dir: /src/sdf/ftsdfcommon.c/
/****************************************************************************
*
* ftsdfcommon.c
*
* Auxiliary data for Signed Distance Field support (body).
*
* Copyright (C) 2020-2021 by
* David Turner, Robert Wilhelm, and Werner Lemberg.
*
* Written by Anuj Verma.
*
* This file is part of the FreeType project, and may only be used,
* modified, and distributed under the terms of the FreeType project
* license, LICENSE.TXT. By continuing to use, modify, or distribute
* this file you indicate that you have read the license and
* understand and accept it fully.
*
*/
#include "ftsdf.h"
#include "ftsdfcommon.h"
/**************************************************************************
*
* common functions
*
*/
/*
* Original algorithm:
*
* https://github.com/chmike/fpsqrt
*
* Use this to compute the square root of a 16.16 fixed point number.
*/
FT_LOCAL_DEF( FT_16D16 )
square_root( FT_16D16 val )
{
FT_ULong t, q, b, r;
r = (FT_ULong)val;
b = 0x40000000L;
q = 0;
while ( b > 0x40L )
{
t = q + b;
if ( r >= t )
{
r -= t;
q = t + b;
}
r <<= 1;
b >>= 1;
}
q >>= 8;
return (FT_16D16)q;
}
/**************************************************************************
*
* format and sign manipulating functions
*
*/
/*
* Convert 16.16 fixed point values to the desired output format.
* In this case we reduce 16.16 fixed point values to normalized
* 8-bit values.
*
* The `max_value` in the parameter is the maximum value in the
* distance field map and is equal to the spread. We normalize
* the distances using this value instead of computing the maximum
* value for the entire bitmap.
*
* You can use this function to map the 16.16 signed values to any
* format required. Do note that the output buffer is 8-bit, so only
* use an 8-bit format for `FT_SDFFormat`, or increase the buffer size in
* `ftsdfrend.c`.
*/
FT_LOCAL_DEF( FT_SDFFormat )
map_fixed_to_sdf( FT_16D16 dist,
FT_16D16 max_value )
{
FT_SDFFormat out;
FT_16D16 udist;
/* normalize the distance values */
dist = FT_DivFix( dist, max_value );
udist = dist < 0 ? -dist : dist;
/* Reduce the distance values to 8 bits. */
/* */
/* Since +1/-1 in 16.16 takes the 16th bit, we right-shift */
/* the number by 9 to make it fit into the 7-bit range. */
/* */
/* One bit is reserved for the sign. */
udist >>= 9;
/* Since `char` can only store a maximum positive value */
/* of 127 we need to make sure it does not wrap around and */
/* give a negative value. */
if ( dist > 0 && udist > 127 )
udist = 127;
if ( dist < 0 && udist > 128 )
udist = 128;
/* Output the data; negative values are from [0, 127] and positive */
/* from [128, 255]. One important thing is that negative values */
/* are inverted here, that means [0, 128] maps to [-128, 0] linearly. */
/* More on that in `freetype.h` near the documentation of */
/* `FT_RENDER_MODE_SDF`. */
out = dist < 0 ? 128 - (FT_SDFFormat)udist
: (FT_SDFFormat)udist + 128;
return out;
}
/*
* Invert the signed distance packed into the corresponding format.
* So if the values are negative they will become positive in the
* chosen format.
*
* [Note]: This function should only be used after converting the
* 16.16 signed distance values to `FT_SDFFormat`. If that
* conversion has not been done, then simply invert the sign
* and use the above function to pack the values.
*/
FT_LOCAL_DEF( FT_SDFFormat )
invert_sign( FT_SDFFormat dist )
{
return 255 - dist;
}
/* END */