FilterFactory - Symbol "d"

Back to function listing

Direction (angle) of the current pixel from the center of the image, where d is an integer between -512 and 512, inclusive.

Note: FilterFactory defines c2d(x,y) := atan2(y,x), but d := atan2(-y,-x)

Errors in the Filter Factory manuals

The manual of Filter Factory for Photoshop states the range as -511 and 512, inclusive, which is wrong for both implementations (Windows and MacOS).
The correct range is -512 and 512, inclusive.

The manual of Filter Factory for Premiere states the range as 0 and 1024, which is also not true.
At least for the Windows implementation, the range seems to be approx. -512 till 512, but I could not detect the exact range (so it could also be 511 instead of 512).

Another mistake in the Filter Factory manuals is the orientation:
The Filter Factory manual writes:

d=0 corresponds to the 3 o’clock position
d=256 to the 6 o’clock position
d=512 to the 9 o’clock position
d=768 to the 12 o’clock position
d=1024 to the full rotation back to the 3 o’clock position

But this does neither match the Windows implementation of Filter Factory nor its Mac OS implementation! In the original Windows & Mac OS Filter Factory implementations we can observe:

d=-512 is at 9 o’clock position
d=-256 is at 12 o’clock position
d=0 is at 3 o’clock position
d=256 is at 6 o’clock position
d=512 is the full rotation back to the 9 o’clock position

Syntax

d

Synonyms

d0
d1

Original Machine Code (in OPER resource) of Filter Factory 3.0 and 3.0.4 for Photoshop/Win32

C++ code

int Y, ymin, X, xmin;

int factory_d() {
	int eax, ebx, ecx;

	eax = -(Y - ymin) / 2;
	ebx = -(X - xmin) / 2;
	ecx = 0;
	eax += y;
	if (eax < 0) {
		eax = -eax;
		ecx |= 0b0100;
	}
	ebx += x;
	if (ebx < 0) {
		ebx = -ebx;
		ecx |= 0b0011;
	}
	if (eax > ebx) {
		ecx ^= 0b0001;
		int xxx = eax;
		eax = ebx;
		ebx = xxx;
	}
	if (eax > 0) {
		eax = eax << 10;
		eax /= ebx;
		if (eax != 0) { // C2D_LOOKUP[-1] will never be called. Good!
			eax = (eax & 0xFFFF0000) + (C2D_LOOKUP[eax-1] & 0xFFFF);
			eax = eax << 9;
			ebx = 205888; // 205888/65536 == pi
			eax /= ebx;
		}
	}
	if ((ecx & 0b0001) != 0) {
		eax = -eax;
		eax += 256;
	}
	if ((ecx & 0b0010) != 0) {
		eax += 256;
	}
	if ((ecx & 0b0100) != 0) {
		eax = -eax;
	}
	return eax;
}