.TH REGION-OPERATIONS 3 "11 April 1990"
.SH NAME
im_region_buffer, im_region_image, im_region_region, im_region_position,
im_region_equalsregion \- 
attach data to region
.SH SYNOPSIS
.B #include <vips/vips.h>

int im_region_buffer( reg, r )
.br
REGION *reg;
.br
Rect *r;

int im_region_image( reg, r )
.br
REGION *reg;
.br
Rect *r;

int im_region_region( reg, treg, r, x, y )
.br
REGION *reg, treg;
.br
Rect *r;
.br
int x, y;

int im_region_position( reg, x, y )
.br
REGION *reg;
.br
int x, y;

int im_region_equalsregion( REGION *reg1, REGION *reg2 )

.SH DESCRIPTION
These functions are used to say where exactly input from or output to a
region goes. Briefly, 
.B im_region_buffer(3) 
gives a region its own piece of local storage, 
.B im_region_image(3) 
links the region to an image, and
.B im_region_region(3)
links the region to another region.

.B im_region_region(3)
is the most powerful of the three --- you can use it to
avoid copy operations. See the source for 
.B im_extract(3), 
.B im_insert(3),
.B im_copy(3), 
.B im_embed(3), 
and 
.B im_lrmerge(3) 
for examples of its use, and see also
.B im_prepare_copy(3).

.B im_region_buffer(3)
clips Rect r against the size of the image upon which reg is
defined, then searches for any calculated pixels which enclose that area. If
there are any suitable pixels already, the region is set to point to them.
Otherwise, a fresh buffer is allocated.

If region->buffer->done is set, then the region is already calculated.
Otherwise, the pixels need to be calculated. 

.B im_region_buffer(3)
is called by 
.B im_prepare(3)
just before it calls the user generate function.

.B im_region_image(3)
attaches reg to the image on which it was defined. The image
has to be SETBUF, MMAPIN, MMAPINRW, or OPENIN --- ie., the types passed by 
.B im_incheck(3).

.B im_region_region(3)
redirects reg to treg. treg can be defined on another
image. r is clipped against the size of reg's image, and against treg->valid.
treg has to have pel data associated with it (ie. memory local to treg, memory
that is part of the image on which treg is defined, or even memory from a
third region), and the pel data has to be in a form which is compatible with
reg, ie. BandFmt and Bands must be the same. 

r becomes the valid field in reg, (x,y) is the top left-hand corner of the
area in treg to which valid is mapped.


.B im_region_position(3)
changes reg->valid, so that reg->valid.left == x and
reg->valid.top == y. width and height are clipped against the size of the
image. If x < 0 or y < 0 or the clipped rect has zero width or height, the
call fails.

This function affects the way that pels are addressed by the 
.B IM_REGION_ADDR(3)
macro. It does not affect the pels themselves! It has the effect of moving
the area of pels a region represents. This call is used by 
.B im_extract(3).

.B im_region_equalsregion(3)
tests for reg1 and reg2 are identical, ie.:

   IM_REGION_ADDR( reg1, x, y ) == 
     IM_REGION_ADDR( reg2, x, y ) &&
   *IM_REGION_ADDR( reg1, x, y ) == 
     *IM_REGION_ADDR( reg2, x, y )

It is used internally by VIPS, but may be useful to applications.

.SH RETURN VALUE
All int-valued functions return zero on success and non-zero on error.
.SH COPYRIGHT
National Gallery, 1993
.SH AUTHOR
J. Cupitt \- 23/7/93