2cb45f698e
fix two small bugs
...
VipsPool was not linking to the parent class correctly, ifthenelse had
tangled images.
2011-06-21 14:42:39 +01:00
3126e5100d
get interpolate new from string working
...
vips_type_map() has to see abstract types now so we can get the
baseclass for interpolate. interpolate class has to set the nickname.
2011-06-21 11:08:56 +01:00
642305327e
oops
2011-06-20 10:29:00 +01:00
d3b29f9cd2
start gobject-introspection experiments
...
messing about with gi for python bindings
2011-06-20 08:31:20 +01:00
f85655052e
build xyz2lab tables in im_col_*
...
added calls to build xyz2lab tables, thanks hkaimio
2011-06-19 17:15:10 +01:00
863325a257
cleanups before attempting gobject base
2011-06-18 17:47:56 +01:00
4918a67e14
init more types on startup
...
create VipsImage and VipsRegion types on vips_init()
2011-06-18 17:42:06 +01:00
e346187043
new python cleanups
...
seems to work now, try an operation next
2011-06-18 13:34:52 +01:00
3628692799
add experimental ctypes Python binding
...
start hacking on a new Python interface
2011-06-17 14:50:14 +01:00
3920f5dc7e
get im_system_image() working again
...
chop stuff about to get system image working again.
2011-06-16 13:54:13 +01:00
6a582ae1a9
test for res-unit meta before fetching
...
if the res-unit meta is missing, don't generate an error message
2011-06-16 11:34:20 +01:00
f85c36edee
oop double free() in system_image on error
2011-06-15 22:17:08 +01:00
5efdcd9b76
Merge remote-tracking branch 'origin/master'
...
Conflicts:
libvips/convolution/im_aconv.c
2011-06-15 17:53:24 +01:00
4650234e20
add VIPS_EXEEXT to the exported API
...
EXEEXT was just internal, now it's part of the API we export
2011-06-15 14:42:49 +01:00
7bd79e3ef4
16-bit intermediates for im_aconv()
...
im_aconv() can use 16-bit intermediates for 8-bit images, if the mask has
small enough hlines (all less than 256 elements)
also, work for all types
2011-06-14 12:36:07 +01:00
85e1359e4a
Merge remote-tracking branch 'origin/master'
2011-06-13 21:36:05 +01:00
ef8fee1722
-ore im_aconv() tweaks
...
common-up hlines where we can
2011-06-13 14:09:19 +01:00
c248b7203d
more im_aconv() cluster polishing
...
larger edge buffer, strike out all involved nodes on merge
2011-06-10 17:21:32 +01:00
702ffd3d36
oops
2011-06-10 14:33:29 +01:00
c605153852
much faster clustering in im_aconv()
...
cluster lines in batches to avoid rescanning the whole space each time
2011-06-10 12:02:33 +01:00
953a315755
more im_aconv() speedups
...
now uses rolling boxes for the vertical pass too
2011-06-09 17:53:53 +01:00
4e3e0cee6c
refactor im_aconv
...
get ready for rolling vertical sums too
2011-06-09 15:21:42 +01:00
21fce2ab9e
im_aconv() works for a large 2d mask
...
needs some more tuning, we should be able to speed it up still
2011-06-09 11:39:31 +01:00
a9e305787b
longer lines for rw_mask
...
rw_mask failed for lines over 4096 characters
2011-06-09 09:35:57 +01:00
d0cb80b341
more testing
...
now fails for large square masks
2011-06-09 00:57:55 +01:00
ab06ba8a1b
woo, works
2011-06-09 00:45:43 +01:00
ecf344ec0f
more tweaks, still not working
...
argh, can't see the problem
2011-06-08 22:13:30 +01:00
788a43b8b1
im_aconv() almost working
2011-06-08 14:49:28 +01:00
cca955136d
more space for image headers
...
up to 32kb for image header print, but a true expanding string would be
better
2011-06-08 13:57:59 +01:00
69331fab9f
new im_aconv() compiles
...
compiled and linked in, needs testing though
2011-06-08 13:37:03 +01:00
2d717ec578
more aconv hackery
2011-06-08 10:30:18 +01:00
b1c1a3a65d
more hacking on im_aconv()
...
got the splitter more-or-less done, the convolver next
2011-06-07 14:50:32 +01:00
0aa9f81c30
started im_aconv()
...
started hacking non-separable version of im_aconvsep()
2011-06-06 17:35:32 +01:00
7b6bc4267b
better vips error msg
...
"vips mpouikh" makes a better error message
2011-06-06 12:31:12 +01:00
20a65cca59
aconvsep supports all formats
...
float/int/complex etc. all work
2011-06-04 21:12:55 +01:00
710294f889
rename aconv as aconvsep
2011-06-04 17:44:54 +01:00
70dc170d97
much better accuracy for small masks
2011-06-04 12:39:54 +01:00
56d8f75500
fix an edge problem in aconv
...
aconv no longer has tiling artifacts
2011-06-04 10:01:52 +01:00
3fab0cbbd4
fix vips8 CLI
...
the vips8 CLI needed updating for the new refcount strategy
2011-06-02 15:53:30 +01:00
35a7515e3f
small bugfixes
...
various bugfixes ... nip2 runs reliably again now
2011-06-02 15:34:57 +01:00
d28a7059cf
oops
2011-06-02 12:54:56 +01:00
971615d6ae
back to fixed refs, add a pool system instead
...
floating refs were very confusing and annoying, instead use simple fixed
refs and add a pool system to track refs for you
also fixed a couple of dumb errors in close callbacks
2011-06-02 12:23:56 +01:00
75e5804e3c
small fixes
...
also bumped smalltile to 512x512 for tetsing aconv, this needs checking
2011-06-01 17:56:28 +01:00
6819919f0c
im_aconv() works
...
got im_aconv() working, woo
2011-06-01 17:36:17 +01:00
f1b8b7e735
oop, test for vips_object_all NULL
2011-06-01 15:20:50 +01:00
45b582678f
first paste of aconv
2011-06-01 09:17:07 +01:00
40a8c8f95d
always check mask->scale is non-zero
2011-05-31 12:24:09 +01:00
770398753c
add vips_check_dmask_1d()
...
add a check-mask-is-1d function
2011-05-31 12:21:05 +01:00
c0cbbbf415
free im_malloc() mem in postclose
2011-05-30 19:40:56 +01:00
a55e8d15bb
fix -ve lobes on laplacian for large sigma
2011-05-30 12:14:20 +01:00
John Cupitt