One lens resolved at one shooting configuration, as a flat block of scalars. More...
#include <lensserious_eval.h>
Collaboration diagram for ls_eval_t:Data Fields | |
| float | norm_scale |
| float | norm_unscale |
| float | center_x |
| float | center_y |
| float | vig_scale |
| float | vig_center_x |
| float | vig_center_y |
| float | scale |
| int | dist_model |
| int | tca_model |
| int | vig_model |
| int | enabled |
| int | geom_from |
| int | geom_to |
| float | geom_focal |
| int | reverse |
| float | dist_terms [3] |
| float | tca_terms [6] |
| float | vig_terms [3] |
| int | knot_axes |
| int | knot_n |
| float | knot_r [3][16] |
| float | knot_c [3][16] |
| int | knot_vn |
| float | knot_vr [16] |
| float | knot_v [16] |
Detailed Description
One lens resolved at one shooting configuration, as a flat block of scalars.
Scalars only, and no vector types: a struct of float and int has the same layout under every host C ABI and under OpenCL C, so this can be memcpy'd into a kernel argument. Anything richer (a float2 centre, say) would lay out differently and corrupt every field after it, silently. tests/parity_lensfun.c pins the size.
Produced by ls_eval_from_modifier(); consumed by ls_eval_map() and ls_eval_vignette_factor(), on either side.
Definition at line 143 of file lensserious_eval.h.
Field Documentation
◆ center_x
| float ls_eval_t::center_x |
Definition at line 147 of file lensserious_eval.h.
Referenced by ls_eval_from_modifier(), and ls_eval_map().
◆ center_y
| float ls_eval_t::center_y |
optical centre, in normalized coordinates
Definition at line 147 of file lensserious_eval.h.
Referenced by ls_eval_from_modifier(), and ls_eval_map().
◆ dist_model
| int ls_eval_t::dist_model |
LS_EVAL_DIST_*
Definition at line 156 of file lensserious_eval.h.
Referenced by ls_eval_coord_chain(), ls_eval_dist(), ls_eval_from_modifier(), ls_eval_map(), and ls_eval_undist_factor().
◆ dist_terms
| float ls_eval_t::dist_terms[3] |
Definition at line 172 of file lensserious_eval.h.
Referenced by ls_eval_dist(), ls_eval_from_modifier(), and ls_eval_undist_factor().
◆ enabled
| int ls_eval_t::enabled |
LS_EVAL_ENABLE_* actually resolved
Definition at line 159 of file lensserious_eval.h.
Referenced by ls_eval_adopt_vignetting(), ls_eval_coord_chain(), ls_eval_from_modifier(), ls_eval_map(), and ls_eval_vignette_factor().
◆ geom_focal
| float ls_eval_t::geom_focal |
focal / LS_EVAL_GEOM_HALF_HEIGHT_MM
Definition at line 162 of file lensserious_eval.h.
Referenced by ls_eval_from_modifier(), and ls_eval_geometry().
◆ geom_from
| int ls_eval_t::geom_from |
the lens's own projection, LS_EVAL_LENS_*
Definition at line 160 of file lensserious_eval.h.
Referenced by ls_eval_from_modifier(), and ls_eval_geometry().
◆ geom_to
| int ls_eval_t::geom_to |
the projection asked for
Definition at line 161 of file lensserious_eval.h.
Referenced by ls_eval_from_modifier(), and ls_eval_geometry().
◆ knot_axes
| int ls_eval_t::knot_axes |
Which axes the knot table serves, as LS_EVAL_ENABLE_* bits.
A maker's curve is distortion and lateral CA together – green IS the distortion, and each channel's departure from green IS the aberration. So a consumer that wants the table's geometry but somebody else's CA asks for DISTORTION alone, and every channel then follows the green curve, leaving the ordinary TCA stage free to run after it. Asking for both is the usual case and keeps the three-curve evaluation.
Definition at line 195 of file lensserious_eval.h.
Referenced by ls_eval_coord_chain(), ls_eval_from_modifier(), and ls_eval_map().
◆ knot_c
| float ls_eval_t::knot_c[3][16] |
Definition at line 198 of file lensserious_eval.h.
Referenced by ls_eval_from_modifier(), and ls_eval_knot_factor().
◆ knot_n
| int ls_eval_t::knot_n |
Definition at line 196 of file lensserious_eval.h.
Referenced by ls_eval_from_modifier(), and ls_eval_knot_factor().
◆ knot_r
| float ls_eval_t::knot_r[3][16] |
Definition at line 197 of file lensserious_eval.h.
Referenced by ls_eval_from_modifier(), and ls_eval_knot_factor().
◆ knot_v
| float ls_eval_t::knot_v[16] |
Definition at line 201 of file lensserious_eval.h.
Referenced by ls_eval_adopt_vignetting(), ls_eval_from_modifier(), and ls_eval_vignette_from_r2().
◆ knot_vn
| int ls_eval_t::knot_vn |
Definition at line 199 of file lensserious_eval.h.
Referenced by ls_eval_adopt_vignetting(), ls_eval_from_modifier(), and ls_eval_vignette_from_r2().
◆ knot_vr
| float ls_eval_t::knot_vr[16] |
Definition at line 200 of file lensserious_eval.h.
Referenced by ls_eval_adopt_vignetting(), ls_eval_from_modifier(), and ls_eval_vignette_from_r2().
◆ norm_scale
| float ls_eval_t::norm_scale |
centred pixel coordinates -> calibration-normalized
Definition at line 145 of file lensserious_eval.h.
Referenced by ls_eval_from_modifier(), and ls_eval_map().
◆ norm_unscale
| float ls_eval_t::norm_unscale |
and back
Definition at line 146 of file lensserious_eval.h.
Referenced by ls_eval_from_modifier(), and ls_eval_map().
◆ reverse
| int ls_eval_t::reverse |
Non-zero when this block describes the REVERSE direction (lensfun's reverse flag).
It is not "the same chain, backwards". Upstream registers different callbacks at different priorities, so the composition order itself changes: forward runs scale(100) -> geometry(500) -> distortion(750), reverse runs undistortion(250) -> geometry(500) -> scale(900). The projection endpoints are swapped and the scale factor is stored un-reciprocated by the resolver, so only the ORDER and the choice of dist/undist live here.
Definition at line 171 of file lensserious_eval.h.
Referenced by ls_eval_coord_chain(), ls_eval_from_modifier(), ls_eval_map(), and ls_eval_vignette_from_r2().
◆ scale
| float ls_eval_t::scale |
linear scaling, already reciprocal; 1.0 when off
Definition at line 155 of file lensserious_eval.h.
Referenced by ls_eval_coord_chain(), and ls_eval_from_modifier().
◆ tca_model
| int ls_eval_t::tca_model |
LS_EVAL_TCA_*
Definition at line 157 of file lensserious_eval.h.
Referenced by ls_eval_from_modifier(), ls_eval_tca(), and ls_eval_untca().
◆ tca_terms
| float ls_eval_t::tca_terms[6] |
Definition at line 173 of file lensserious_eval.h.
Referenced by ls_eval_from_modifier(), ls_eval_tca(), and ls_eval_untca().
◆ vig_center_x
| float ls_eval_t::vig_center_x |
Definition at line 154 of file lensserious_eval.h.
Referenced by ls_eval_adopt_vignetting(), ls_eval_from_modifier(), ls_eval_vignette_factor(), and ls_modifier_apply_vignetting().
◆ vig_center_y
| float ls_eval_t::vig_center_y |
centre / aspect_ratio_correction
Definition at line 154 of file lensserious_eval.h.
Referenced by ls_eval_adopt_vignetting(), ls_eval_from_modifier(), ls_eval_vignette_factor(), and ls_modifier_apply_vignetting().
◆ vig_model
| int ls_eval_t::vig_model |
LS_EVAL_VIG_*
Definition at line 158 of file lensserious_eval.h.
Referenced by ls_eval_adopt_vignetting(), ls_eval_from_modifier(), and ls_eval_vignette_from_r2().
◆ vig_scale
| float ls_eval_t::vig_scale |
norm_scale / aspect_ratio_correction
Definition at line 153 of file lensserious_eval.h.
Referenced by ls_eval_adopt_vignetting(), ls_eval_from_modifier(), ls_eval_vignette_factor(), and ls_modifier_apply_vignetting().
◆ vig_terms
| float ls_eval_t::vig_terms[3] |
Definition at line 174 of file lensserious_eval.h.
Referenced by ls_eval_adopt_vignetting(), ls_eval_from_modifier(), and ls_eval_vignette_from_r2().
The documentation for this struct was generated from the following file:
