This repository has been archived on 2023-10-09. You can view files and clone it. You cannot open issues or pull requests or push a commit.
Files
blender-archive/source/blender/compositor/operations/COM_IDMaskOperation.cc
Campbell Barton c434782e3a File headers: SPDX License migration
Use a shorter/simpler license convention, stops the header taking so
much space.

Follow the SPDX license specification: https://spdx.org/licenses

- C/C++/objc/objc++
- Python
- Shell Scripts
- CMake, GNUmakefile

While most of the source tree has been included

- `./extern/` was left out.
- `./intern/cycles` & `./intern/atomic` are also excluded because they
  use different header conventions.

doc/license/SPDX-license-identifiers.txt has been added to list SPDX all
used identifiers.

See P2788 for the script that automated these edits.

Reviewed By: brecht, mont29, sergey

Ref D14069
2022-02-11 09:14:36 +11:00

50 lines
1.5 KiB
C++

/* SPDX-License-Identifier: GPL-2.0-or-later
* Copyright 2011 Blender Foundation. */
#include "COM_IDMaskOperation.h"
namespace blender::compositor {
IDMaskOperation::IDMaskOperation()
{
this->add_input_socket(DataType::Value);
this->add_output_socket(DataType::Value);
flags_.complex = true;
flags_.can_be_constant = true;
}
void *IDMaskOperation::initialize_tile_data(rcti *rect)
{
void *buffer = get_input_operation(0)->initialize_tile_data(rect);
return buffer;
}
void IDMaskOperation::execute_pixel(float output[4], int x, int y, void *data)
{
MemoryBuffer *input_buffer = (MemoryBuffer *)data;
const int buffer_width = input_buffer->get_width();
float *buffer = input_buffer->get_buffer();
int buffer_index = (y * buffer_width + x);
output[0] = (roundf(buffer[buffer_index]) == object_index_) ? 1.0f : 0.0f;
}
void IDMaskOperation::update_memory_buffer_partial(MemoryBuffer *output,
const rcti &area,
Span<MemoryBuffer *> inputs)
{
const MemoryBuffer *input = inputs[0];
const int width = BLI_rcti_size_x(&area);
for (int y = area.ymin; y < area.ymax; y++) {
float *out = output->get_elem(area.xmin, y);
const float *in = input->get_elem(area.xmin, y);
const float *row_end = out + width * output->elem_stride;
while (out < row_end) {
out[0] = (roundf(in[0]) == object_index_) ? 1.0f : 0.0f;
in += input->elem_stride;
out += output->elem_stride;
}
}
}
} // namespace blender::compositor