dnrops
/
gimp
mirror of https://github.com/GNOME/gimp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Bug 786844 - Wavelet decompose plugin changes image slightly 

(AKA Issue with grain merge layer mode)

Fall back to using the legacy grain-extract/merge modes when the
input image precision is integer, and only use the non-legacy modes
when the precision is floating point.  The midpoint intensity value
of the non-legacy modes is 0.5, which isn't representable exactly
using integer precision, causing small deviations between the input
image and the reconstructed image.
This commit is contained in:
Ell 2017-10-29 11:20:54 -04:00
parent e4440e3fdb
commit 70881728f2
1 changed files with 44 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

50
plug-ins/common/wavelet-decompose.c
View File

@ -167,10 +167,12 @@ run (const gchar *name,
if (status == GIMP_PDB_SUCCESS)
{
gint32 *scale_ids;
gint32 new_scale_id;
gint32 parent_id;
gint id;
gint32 *scale_ids;
gint32 new_scale_id;
gint32 parent_id;
GimpLayerMode grain_extract_mode = GIMP_LAYER_MODE_GRAIN_EXTRACT;
GimpLayerMode grain_merge_mode = GIMP_LAYER_MODE_GRAIN_MERGE;
gint id;
gimp_progress_init (_("Wavelet-Decompose"));
@ -196,6 +198,42 @@ run (const gchar *name,
gimp_image_get_item_position (image_id,
drawable_id));
/* the exact result of the grain-extract and grain-merge modes depends on
* the choice of (gamma-corrected) midpoint intensity value. for the
* non-legacy modes, the midpoint value is 0.5, which isn't representable
* exactly using integer precision. for the legacy modes, the midpoint
* value is 128/255 (i.e., 0x80), which is representable exactly using
* (gamma-corrected) integer precision. we therefore use the legacy
* modes when the input image precision is integer, and only use the
* (preferable) non-legacy modes when the input image precision is
* floating point.
*
* this avoids imperfect reconstruction of the image when using integer
* precision. see bug #786844.
*/
switch (gimp_image_get_precision (image_id))
{
case GIMP_PRECISION_U8_LINEAR:
case GIMP_PRECISION_U8_GAMMA:
case GIMP_PRECISION_U16_LINEAR:
case GIMP_PRECISION_U16_GAMMA:
case GIMP_PRECISION_U32_LINEAR:
case GIMP_PRECISION_U32_GAMMA:
grain_extract_mode = GIMP_LAYER_MODE_GRAIN_EXTRACT_LEGACY;
grain_merge_mode = GIMP_LAYER_MODE_GRAIN_MERGE_LEGACY;
break;
case GIMP_PRECISION_HALF_LINEAR:
case GIMP_PRECISION_HALF_GAMMA:
case GIMP_PRECISION_FLOAT_LINEAR:
case GIMP_PRECISION_FLOAT_GAMMA:
case GIMP_PRECISION_DOUBLE_LINEAR:
case GIMP_PRECISION_DOUBLE_GAMMA:
grain_extract_mode = GIMP_LAYER_MODE_GRAIN_EXTRACT;
grain_merge_mode = GIMP_LAYER_MODE_GRAIN_MERGE;
break;
}
for (id = 0 ; id < wavelet_params.scales; ++id)
{
gint32 blur_id, tmp_id;
@ -219,7 +257,7 @@ run (const gchar *name,
gimp_image_get_item_position (image_id,
tmp_id));
gimp_layer_set_mode (tmp_id, GIMP_LAYER_MODE_GRAIN_EXTRACT);
gimp_layer_set_mode (tmp_id, grain_extract_mode);
new_scale_id = gimp_image_merge_down (image_id, tmp_id,
GIMP_EXPAND_AS_NECESSARY);
scale_ids[id] = new_scale_id;
@ -236,7 +274,7 @@ run (const gchar *name,
gimp_image_reorder_item (image_id, scale_ids[id], parent_id,
gimp_image_get_item_position (image_id,
new_scale_id));
gimp_layer_set_mode (scale_ids[id], GIMP_LAYER_MODE_GRAIN_MERGE);
gimp_layer_set_mode (scale_ids[id], grain_merge_mode);
if (wavelet_params.create_masks)
{