griefith/test
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

VSE: Fix various "off by half a pixel" issues in image transform

Browse Source 
Code inside `IMB_transform` (which is pretty much only used inside VSE
to do translation/rotation/scale of image or movie strips) was not
correctly doing mapping between pixel and texel spaces. This is similar
to e.g. GPU rasterization rules and has to do with whether some
coordinate refers to pixel/texel "corner" or "center" etc. It's a long
topic, but short summary would be:

- Coordinates refer to pixel/texel corner,
- Do sampling at pixel centers,
- Bilinear filter should use floor(x-0.5) and floor(x-0.5)+1 texels.

Also, there was a sign error introduced in Subsampling 3x3 filter, in
commit b3fd169259.

After making the PR, I found out that this seems to fix #90785, #112923
and possibly some others.

Long explanation with lots of images is in the PR.

Pull Request: https://projects.blender.org/blender/blender/pulls/116628
This commit is contained in:
Aras Pranckevicius
2024-01-08 16:51:35 +01:00
committed by Aras Pranckevicius
parent 694cb43e59
commit 877d9c596a
1 changed files with 18 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

31
source/blender/imbuf/intern/transform.cc
View File

@@ -81,16 +81,12 @@ struct TransformUserData {
void init_add_x(const float4x4 &transform_matrix)
{
const double width = src->x;
add_x = double2(transform_matrix.x_axis()) * width + double2(transform_matrix.location());
add_x = (add_x - start_uv) * (1.0 / width);
add_x = double2(transform_matrix.x_axis());
}
void init_add_y(const float4x4 &transform_matrix)
{
const double height = src->y;
add_y = double2(transform_matrix.y_axis()) * height + double2(transform_matrix.location());
add_y = (add_y - start_uv) * (1.0 / height);
add_y = double2(transform_matrix.y_axis());
}
void init_subsampling(const int num_subsamples)
@@ -102,7 +98,7 @@ struct TransformUserData {
for (int y : IndexRange(0, num_subsamples)) {
for (int x : IndexRange(0, num_subsamples)) {
double2 delta_uv = -offset_x - offset_y;
double2 delta_uv = offset_x + offset_y;
delta_uv += x * subsample_add_x;
delta_uv += y * subsample_add_y;
subsampling.delta_uvs.append(delta_uv);
@@ -313,6 +309,13 @@ class Sampler {
u = wrap_uv(u, source->x);
v = wrap_uv(v, source->y);
}
/* BLI_bilinear_interpolation functions use `floor(uv)` and `floor(uv)+1`
* texels. For proper mapping between pixel and texel spaces, need to
* subtract 0.5. */
if constexpr (Filter == IMB_FILTER_BILINEAR) {
u -= 0.5f;
v -= 0.5f;
}
if constexpr (Filter == IMB_FILTER_BILINEAR && std::is_same_v<StorageType, float> &&
NumChannels == 4)
{
@@ -505,9 +508,10 @@ class ScanlineProcessor {
private:
void process_one_sample_per_pixel(const TransformUserData *user_data, int scanline)
{
double2 uv = user_data->start_uv +
user_data->destination_region.x_range.first() * user_data->add_x +
user_data->add_y * scanline;
/* Note: sample at pixel center for proper filtering. */
double pixel_x = user_data->destination_region.x_range.first() + 0.5;
double pixel_y = scanline + 0.5;
double2 uv = user_data->start_uv + user_data->add_x * pixel_x + user_data->add_y * pixel_y;
output.init_pixel_pointer(user_data->dst,
int2(user_data->destination_region.x_range.first(), scanline));
@@ -526,9 +530,10 @@ class ScanlineProcessor {
void process_with_subsampling(const TransformUserData *user_data, int scanline)
{
double2 uv = user_data->start_uv +
user_data->destination_region.x_range.first() * user_data->add_x +
user_data->add_y * scanline;
/* Note: sample at pixel center for proper filtering. */
double pixel_x = user_data->destination_region.x_range.first() + 0.5;
double pixel_y = scanline + 0.5;
double2 uv = user_data->start_uv + user_data->add_x * pixel_x + user_data->add_y * pixel_y;
output.init_pixel_pointer(user_data->dst,
int2(user_data->destination_region.x_range.first(), scanline));