C, (78 + 26 * strlen) codels
This was surprisingly tricky to optimize, mostly because of the possibility of color collisions in neighbouring lines.
The characters are converted to base 12, so each character is a 2-digit number. Every standard line contains the following: pointer (now right for odd lines, left for even lines), duplicate (the number 12, which is first on the stack), push (1st digit), multiply, push (2nd digit), add, outc, push (1 for odd lines, 3 for even lines), duplicate, whitespace, pointer (now down at the end of the line).
To avoid color collisions in neighbouring lines the state after whitespace filling is remembered, and the generation is rolled back to it if a collision happens. The next try starts there with the next color.
Output for "Hello Piet!":

asciipiet2.c
#include "img.h"
#define WIDTH 26
#define OP(op, h, d) int op() { hue += h; dark += d; hue %= 6; dark %= 3; return setp(); }
#define CCMP(c1, c2) (((c1).r == (c2).r) && ((c1).g == (c2).g) && ((c1).b == (c2).b))
#define OPCNT(op) if(op) continue
Color piet[6][2] =
{
{{0xff, 0xc0, 0xc0}, {0xff, 0x00, 0x00}, {0xc0, 0x00, 0x00}},
{{0xff, 0xff, 0xc0}, {0xff, 0xff, 0x00}, {0xc0, 0xc0, 0x00}},
{{0xc0, 0xff, 0xc0}, {0x00, 0xff, 0x00}, {0x00, 0xc0, 0x00}},
{{0xc0, 0xff, 0xff}, {0x00, 0xff, 0xff}, {0x00, 0xc0, 0xc0}},
{{0xc0, 0xc0, 0xff}, {0x00, 0x00, 0xff}, {0x00, 0x00, 0xc0}},
{{0xff, 0xc0, 0xff}, {0xff, 0x00, 0xff}, {0xc0, 0x00, 0xc0}}
};
Color white = {0xff, 0xff, 0xff};
Image img;
int hue, dark, x, y, dx = 1;
void nextline()
{
x -= dx;
dx = -dx;
y += 1;
}
int setp()
{
if(y > 0 && CCMP(piet[hue][dark], imgGetP(img, x, y - 1)))
{
return 1;
}
imgSetP(img, x, y, piet[hue][dark]);
x += dx;
return 0;
}
void whiteto(int to)
{
if(dx == 1)
{
while(x < to) imgSetP(img, x++, y, white);
}
else
{
while(x >= WIDTH - to) imgSetP(img, x--, y, white);
}
}
OP(fill, 0, 0)
OP(pushraw, 0, 1)
OP(pop, 0, 2)
OP(add, 1, 0)
OP(sub, 1, 1)
OP(mul, 1, 2)
OP(divi, 2, 0)
OP(mod, 2, 1)
OP(not, 2, 2)
OP(gt, 3, 0)
OP(pnt, 3, 1)
OP(sw, 3, 2)
OP(dup, 4, 0)
OP(roll, 4, 1)
OP(in, 4, 2)
OP(inc, 5, 0)
OP(out, 5, 1)
OP(outc, 5, 2)
int push(int num);
int pushn(int num) { int i; for(i = 0; i < num - 1; ++i) { if(fill()) return 1; } return pushraw(); }
int push0() { return (push(1) || not()); }
int push8() { return (push(2) || dup() || dup() || mul() || mul()); }
int push9() { return (push(3) || dup() || mul()); }
int push10() { return (push(9) || push(1) || add()); }
int push11() { return (push(9) || push(2) || add()); }
int push(int num)
{
switch(num)
{
case 0: return push0();
case 8: return push8();
case 9: return push9();
case 10: return push10();
case 11: return push11();
default: return pushn(num);
}
}
int main(int argc, char* argv[])
{
char* str;
int len, i;
if(argc != 2)
{
printf("Usage: %s \"string to print\"\n", argv[0]);
return -1;
}
str = argv[1];
len = strlen(str);
imgCreate(img, WIDTH, len + 3);
fill(); push(4); push(3); mul(); push(1); dup(); whiteto(WIDTH - 2);
for(i = 0; i < len; ++i)
{
int var, sx = x, sy = y, sdx = dx, fin = 0, off = rand();
for(var = 0; var < 18 && !fin; var++)
{
x = sx; y = sy; dx = sdx;
hue = ((var + off) % 18) / 3; dark = ((var + off) % 18) % 3;
OPCNT(fill()); OPCNT(pnt());
nextline(); pnt(); dup();
OPCNT(push(str[i] / 12)); OPCNT(mul()); OPCNT(push(str[i] % 12)); OPCNT(add()); OPCNT(outc()); OPCNT(push(2 - dx)); if(i != len - 1) { OPCNT(dup()); }
whiteto(WIDTH - 2);
fin = 1;
}
if (!fin)
{
printf("collision unavoidable\n");
return -1;
}
}
x -= dx;
{
int var, sx = x, sy = y, sdx = dx, fin = 0;
for(var = 0; var < 18 && !fin; var++)
{
x = sx; y = sy; dx = sdx;
hue = var / 3; dark = var % 3;
OPCNT(fill()); OPCNT(pnt()); OPCNT(fill());
fin = 1;
}
if (!fin)
{
printf("collision unavoidable\n");
return -1;
}
}
x -= 2 * dx;
y += 1;
imgSetP(img, x, y, white);
x -= dx;
y += 1;
hue = 0; dark = 1;
fill(); fill(); fill();
imgSave(img, "piet.pnm");
return 0;
}
img.h
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
typedef struct
{
unsigned char r;
unsigned char g;
unsigned char b;
} Color;
typedef struct
{
Color* data;
int width;
int height;
} Image;
#define imgCreate(img, w, h) {\
int length;\
(img).width = (w);\
(img).height = (h);\
length = (img).width * (img).height * sizeof(Color);\
(img).data = malloc(length);\
memset((img).data, 0, length);\
}
#define imgDestroy(img) {\
free((img).data);\
(img).width = 0;\
(img).height = 0;\
}
#define imgGetP(img, x, y) ((img).data[(int)(x) + (int)(y) * (img).width])
#define imgSetP(img, x, y, c) {\
(img).data[(int)(x) + (int)(y) * (img).width] = c;\
}
#define imgLine(img, x, y, xx, yy, c) {\
int x0 = (x), y0 = (y), x1 = (xx), y1 = (yy);\
int dx = abs(x1 - x0), sx = x0 < x1 ? 1 : -1;\
int dy = -abs(y1 - y0), sy = y0 < y1 ? 1 : -1;\
int err = dx + dy, e2;\
\
for(;;)\
{\
imgSetP((img), x0, y0, c);\
if (x0 == x1 && y0 == y1) break;\
e2 = 2 * err;\
if (e2 >= dy) {err += dy; x0 += sx;}\
if (e2 <= dx) {err += dx; y0 += sy;}\
}\
}
#define imgSave(img, fname) {\
FILE* f = fopen((fname), "wb");\
fprintf(f, "P6\n%d %d\n255\n", (img).width, (img).height);\
fwrite((img).data, sizeof(Color), (img).width * (img).height, f);\
fclose(f);\
}
#define imgLoad(img, fname) {\
FILE* f = fopen((fname), "rb");\
char buffer[16];\
int index = 0;\
int field = 0;\
int isP5 = 0;\
unsigned char c = ' ';\
while(field < 4)\
{\
do\
{\
if(c == '#') while(c = fgetc(f), c != '\n');\
} while(c = fgetc(f), isspace(c) || c == '#');\
index = 0;\
do\
{\
buffer[index++] = c;\
} while(c = fgetc(f), !isspace(c) && c != '#' && index < 16);\
buffer[index] = 0;\
switch(field)\
{\
case 0:\
if (strcmp(buffer, "P5") == 0) isP5 = 1;\
else if (strcmp(buffer, "P6") == 0) isP5 = 0;\
else fprintf(stderr, "image format \"%s\" unsupported (not P5 or P6)\n", buffer), exit(1);\
break;\
case 1:\
(img).width = atoi(buffer);\
break;\
case 2:\
(img).height = atoi(buffer);\
break;\
case 3:\
index = atoi(buffer);\
if (index != 255) fprintf(stderr, "image format unsupported (not 255 values per channel)\n"), exit(1);\
break;\
}\
field++;\
}\
imgCreate((img), (img).width, (img).height);\
if (isP5)\
{\
int length = (img).width * (img).height;\
for(index = 0; index < length; ++index)\
{\
(img).data[index].r = (img).data[index].g = (img).data[index].b = fgetc(f);\
}\
}\
else\
{\
fread((img).data, sizeof(Color), (img).width * (img).height, f);\
}\
fclose(f);\
}