The current implementation for lossy HTJ2K is passing in -1.0f as qstep param when invoking openJPH's encoding, this will in turn use the OpenJPH's default value. Would it be possible to expose the qstep param as one of the HTJ2K encoding option parameters, so we can have more control of the compression ratio of the output?
The hardcoded qstep value can be found at Native/Common/OpenJPH/interface/ojph_interface.cpp, line 20.
void HTJpeg2000EncodeStream(Htj2k_outdata *j2c_outinfo, const unsigned char *source, size_t sourceLength, const struct Frameinfo *finfo, ojph::PROGRESSION_ORDER progression_order)
{
size_t decompositions_ = 5;
bool request_tlm_marker_ = true;
bool set_tilepart_divisions_at_components_ = false;
bool set_tilepart_divisions_at_resolutions_ = true;
float quantizationStep_ = -1.0f;
Thank you in advance.
The current implementation for lossy HTJ2K is passing in -1.0f as qstep param when invoking openJPH's encoding, this will in turn use the OpenJPH's default value. Would it be possible to expose the qstep param as one of the HTJ2K encoding option parameters, so we can have more control of the compression ratio of the output?
The hardcoded qstep value can be found at Native/Common/OpenJPH/interface/ojph_interface.cpp, line 20.
void HTJpeg2000EncodeStream(Htj2k_outdata *j2c_outinfo, const unsigned char *source, size_t sourceLength, const struct Frameinfo *finfo, ojph::PROGRESSION_ORDER progression_order){size_t decompositions_ = 5;bool request_tlm_marker_ = true;bool set_tilepart_divisions_at_components_ = false;bool set_tilepart_divisions_at_resolutions_ = true;float quantizationStep_ = -1.0f;Thank you in advance.