Files
test2/source/blender/blenkernel/intern/data_transfer.cc

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

1813 lines
71 KiB
C++
Raw Normal View History

/* SPDX-License-Identifier: GPL-2.0-or-later
* Copyright 2014 Blender Foundation. All rights reserved. */
/** \file
* \ingroup bke
*/
#include "CLG_log.h"
#include "MEM_guardedalloc.h"
#include "DNA_customdata_types.h"
#include "DNA_mesh_types.h"
#include "DNA_meshdata_types.h"
#include "DNA_object_types.h"
#include "DNA_scene_types.h"
#include "BLI_blenlib.h"
#include "BLI_math.h"
#include "BLI_utildefines.h"
#include "BKE_attribute.h"
#include "BKE_attribute.hh"
#include "BKE_customdata.h"
#include "BKE_data_transfer.h"
#include "BKE_deform.h"
#include "BKE_mesh.hh"
#include "BKE_mesh_mapping.h"
#include "BKE_mesh_remap.h"
#include "BKE_mesh_runtime.h"
#include "BKE_mesh_wrapper.h"
#include "BKE_modifier.h"
#include "BKE_object.h"
#include "BKE_object_deform.h"
#include "BKE_report.h"
#include "data_transfer_intern.h"
static CLG_LogRef LOG = {"bke.data_transfer"};
void BKE_object_data_transfer_dttypes_to_cdmask(const int dtdata_types,
CustomData_MeshMasks *r_data_masks)
{
for (int i = 0; i < DT_TYPE_MAX; i++) {
const int dtdata_type = 1 << i;
int cddata_type;
if (!(dtdata_types & dtdata_type)) {
continue;
}
cddata_type = BKE_object_data_transfer_dttype_to_cdtype(dtdata_type);
if (!(cddata_type & CD_FAKE)) {
if (DT_DATATYPE_IS_VERT(dtdata_type)) {
r_data_masks->vmask |= 1LL << cddata_type;
}
else if (DT_DATATYPE_IS_EDGE(dtdata_type)) {
r_data_masks->emask |= 1LL << cddata_type;
}
else if (DT_DATATYPE_IS_LOOP(dtdata_type)) {
r_data_masks->lmask |= 1LL << cddata_type;
}
else if (DT_DATATYPE_IS_POLY(dtdata_type)) {
r_data_masks->pmask |= 1LL << cddata_type;
}
}
else if (cddata_type == CD_FAKE_MDEFORMVERT) {
r_data_masks->vmask |= CD_MASK_MDEFORMVERT; /* Exception for vgroups :/ */
}
else if (cddata_type == CD_FAKE_UV) {
Mesh: Move UV layers to generic attributes Currently the `MLoopUV` struct stores UV coordinates and flags related to editing UV maps in the UV editor. This patch changes the coordinates to use the generic 2D vector type, and moves the flags into three separate boolean attributes. This follows the design in T95965, with the ultimate intention of simplifying code and improving performance. Importantly, the change allows exporters and renderers to use UVs "touched" by geometry nodes, which only creates generic attributes. It also allows geometry nodes to create "proper" UV maps from scratch, though only with the Store Named Attribute node for now. The new design considers any 2D vector attribute on the corner domain to be a UV map. In the future, they might be distinguished from regular 2D vectors with attribute metadata, which may be helpful because they are often interpolated differently. Most of the code changes deal with passing around UV BMesh custom data offsets and tracking the boolean "sublayers". The boolean layers are use the following prefixes for attribute names: vert selection: `.vs.`, edge selection: `.es.`, pinning: `.pn.`. Currently these are short to avoid using up the maximum length of attribute names. To accommodate for these 4 extra characters, the name length limit is enlarged to 68 bytes, while the maximum user settable name length is still 64 bytes. Unfortunately Python/RNA API access to the UV flag data becomes slower. Accessing the boolean layers directly is be better for performance in general. Like the other mesh SoA refactors, backward and forward compatibility aren't affected, and won't be changed until 4.0. We pay for that by making mesh reading and writing more expensive with conversions. Resolves T85962 Differential Revision: https://developer.blender.org/D14365
2023-01-10 00:47:04 -05:00
r_data_masks->lmask |= CD_MASK_PROP_FLOAT2;
}
else if (cddata_type == CD_FAKE_LNOR) {
r_data_masks->lmask |= CD_MASK_NORMAL | CD_MASK_CUSTOMLOOPNORMAL;
}
}
}
bool BKE_object_data_transfer_get_dttypes_capacity(const int dtdata_types,
bool *r_advanced_mixing,
bool *r_threshold)
{
bool ret = false;
*r_advanced_mixing = false;
*r_threshold = false;
2020-09-09 16:35:20 +02:00
for (int i = 0; (i < DT_TYPE_MAX) && !(ret && *r_advanced_mixing && *r_threshold); i++) {
const int dtdata_type = 1 << i;
if (!(dtdata_types & dtdata_type)) {
continue;
}
switch (dtdata_type) {
2018-04-16 17:08:27 +02:00
/* Vertex data */
case DT_TYPE_MDEFORMVERT:
*r_advanced_mixing = true;
*r_threshold = true;
ret = true;
break;
case DT_TYPE_SKIN:
*r_threshold = true;
ret = true;
break;
case DT_TYPE_BWEIGHT_VERT:
ret = true;
break;
2018-04-16 17:08:27 +02:00
/* Edge data */
case DT_TYPE_SHARP_EDGE:
*r_threshold = true;
ret = true;
break;
case DT_TYPE_SEAM:
*r_threshold = true;
ret = true;
break;
case DT_TYPE_CREASE:
ret = true;
break;
case DT_TYPE_BWEIGHT_EDGE:
ret = true;
break;
case DT_TYPE_FREESTYLE_EDGE:
*r_threshold = true;
ret = true;
break;
2018-04-16 17:08:27 +02:00
/* Loop/Poly data */
case DT_TYPE_UV:
ret = true;
break;
case DT_TYPE_MPROPCOL_VERT:
case DT_TYPE_MLOOPCOL_VERT:
case DT_TYPE_MPROPCOL_LOOP:
case DT_TYPE_MLOOPCOL_LOOP:
*r_advanced_mixing = true;
*r_threshold = true;
ret = true;
break;
case DT_TYPE_LNOR:
*r_advanced_mixing = true;
ret = true;
break;
case DT_TYPE_SHARP_FACE:
*r_threshold = true;
ret = true;
break;
case DT_TYPE_FREESTYLE_FACE:
*r_threshold = true;
ret = true;
break;
}
}
return ret;
}
int BKE_object_data_transfer_get_dttypes_item_types(const int dtdata_types)
{
int i, ret = 0;
for (i = 0; (i < DT_TYPE_MAX) && (ret ^ (ME_VERT | ME_EDGE | ME_LOOP | ME_POLY)); i++) {
const int dtdata_type = 1 << i;
if (!(dtdata_types & dtdata_type)) {
continue;
}
if (DT_DATATYPE_IS_VERT(dtdata_type)) {
ret |= ME_VERT;
}
if (DT_DATATYPE_IS_EDGE(dtdata_type)) {
ret |= ME_EDGE;
}
if (DT_DATATYPE_IS_LOOP(dtdata_type)) {
ret |= ME_LOOP;
}
if (DT_DATATYPE_IS_POLY(dtdata_type)) {
ret |= ME_POLY;
}
}
return ret;
}
int BKE_object_data_transfer_dttype_to_cdtype(const int dtdata_type)
{
switch (dtdata_type) {
case DT_TYPE_MDEFORMVERT:
return CD_FAKE_MDEFORMVERT;
case DT_TYPE_SHAPEKEY:
return CD_FAKE_SHAPEKEY;
case DT_TYPE_SKIN:
return CD_MVERT_SKIN;
case DT_TYPE_BWEIGHT_VERT:
return CD_BWEIGHT;
case DT_TYPE_SHARP_EDGE:
return CD_FAKE_SHARP;
case DT_TYPE_SEAM:
return CD_FAKE_SEAM;
case DT_TYPE_CREASE:
return CD_CREASE;
case DT_TYPE_BWEIGHT_EDGE:
return CD_BWEIGHT;
case DT_TYPE_FREESTYLE_EDGE:
return CD_FREESTYLE_EDGE;
case DT_TYPE_UV:
return CD_FAKE_UV;
case DT_TYPE_SHARP_FACE:
return CD_FAKE_SHARP;
case DT_TYPE_FREESTYLE_FACE:
return CD_FREESTYLE_FACE;
case DT_TYPE_LNOR:
return CD_FAKE_LNOR;
case DT_TYPE_MLOOPCOL_VERT:
case DT_TYPE_MLOOPCOL_LOOP:
return CD_PROP_BYTE_COLOR;
case DT_TYPE_MPROPCOL_VERT:
case DT_TYPE_MPROPCOL_LOOP:
return CD_PROP_COLOR;
default:
BLI_assert_unreachable();
}
return 0; /* Should never be reached! */
}
int BKE_object_data_transfer_dttype_to_srcdst_index(const int dtdata_type)
{
switch (dtdata_type) {
case DT_TYPE_MDEFORMVERT:
return DT_MULTILAYER_INDEX_MDEFORMVERT;
case DT_TYPE_SHAPEKEY:
return DT_MULTILAYER_INDEX_SHAPEKEY;
case DT_TYPE_UV:
return DT_MULTILAYER_INDEX_UV;
case DT_TYPE_MPROPCOL_VERT:
case DT_TYPE_MLOOPCOL_VERT:
Fix T103400: Transfer Mesh Data Layout broken for color attributes This was the case when using the operator outside of the modifiers panel. Caused by {rBeae36be372a6}. In above commit, `DT_layer_items` shared both `DT_TYPE_MPROPCOL_LOOP` | `DT_TYPE_MLOOPCOL_LOOP` in a single EnumPropertyItem value "Colors". This is a bit unusual, but probably allowed. As a consequence, checks for specific datatypes would fail when selecting such EnumPropertyItem: - `DT_DATATYPE_IS_MULTILAYERS` (uses `ELEM` to check distinct entries -- would return false) - `BKE_object_data_transfer_dttype_to_srcdst_index` (would return `DT_MULTILAYER_INDEX_INVALID`) These places have now been corrected to take these "special" values into account. Another issue was that multiple EnumPropertyItems with the same value could be created in dt_add_vcol_layers() if attributes of the same domain, but different color types are in play (could lead to crashes) and that has also been corrected. Also: above commit did not give the choice of transfering color attributes from the vertex domain (only face corner attributes could be chosen), this has now been added. DT_layer_vert_items (used from the modifier) already had this included so this was only an issue when using the operator outside of the modifiers panel. Since we now feature two domains, the single "VCOL" in the enum has been split into "COLOR_VERTEX" and "COLOR_CORNER". This will break existing scripts calling bpy.ops.object.datalayout_transfer and will be marked as a breaking change in the release notes. NOTE: there is another bug here when attributes of the same domain, but different color types are in play and you want to transfer just a single specific layer (but that is for a separate commit) Maniphest Tasks: T103400 Differential Revision: https://developer.blender.org/D16935
2023-01-05 16:05:51 +01:00
case DT_TYPE_MPROPCOL_VERT | DT_TYPE_MLOOPCOL_VERT:
return DT_MULTILAYER_INDEX_VCOL_VERT;
case DT_TYPE_MPROPCOL_LOOP:
case DT_TYPE_MLOOPCOL_LOOP:
Fix T103400: Transfer Mesh Data Layout broken for color attributes This was the case when using the operator outside of the modifiers panel. Caused by {rBeae36be372a6}. In above commit, `DT_layer_items` shared both `DT_TYPE_MPROPCOL_LOOP` | `DT_TYPE_MLOOPCOL_LOOP` in a single EnumPropertyItem value "Colors". This is a bit unusual, but probably allowed. As a consequence, checks for specific datatypes would fail when selecting such EnumPropertyItem: - `DT_DATATYPE_IS_MULTILAYERS` (uses `ELEM` to check distinct entries -- would return false) - `BKE_object_data_transfer_dttype_to_srcdst_index` (would return `DT_MULTILAYER_INDEX_INVALID`) These places have now been corrected to take these "special" values into account. Another issue was that multiple EnumPropertyItems with the same value could be created in dt_add_vcol_layers() if attributes of the same domain, but different color types are in play (could lead to crashes) and that has also been corrected. Also: above commit did not give the choice of transfering color attributes from the vertex domain (only face corner attributes could be chosen), this has now been added. DT_layer_vert_items (used from the modifier) already had this included so this was only an issue when using the operator outside of the modifiers panel. Since we now feature two domains, the single "VCOL" in the enum has been split into "COLOR_VERTEX" and "COLOR_CORNER". This will break existing scripts calling bpy.ops.object.datalayout_transfer and will be marked as a breaking change in the release notes. NOTE: there is another bug here when attributes of the same domain, but different color types are in play and you want to transfer just a single specific layer (but that is for a separate commit) Maniphest Tasks: T103400 Differential Revision: https://developer.blender.org/D16935
2023-01-05 16:05:51 +01:00
case DT_TYPE_MPROPCOL_LOOP | DT_TYPE_MLOOPCOL_LOOP:
return DT_MULTILAYER_INDEX_VCOL_LOOP;
default:
return DT_MULTILAYER_INDEX_INVALID;
}
}
/* ********** */
/**
2023-02-27 20:54:27 +11:00
* When transferring color attributes, also transfer the active color attribute string.
* If a match can't be found, use the first color layer that can be found (to ensure a valid string
* is set).
*/
static void data_transfer_mesh_attributes_transfer_active_color_string(
Mesh *mesh_dst, Mesh *mesh_src, const eAttrDomainMask mask_domain, const int data_type)
{
if (mesh_dst->active_color_attribute) {
return;
}
const char *active_color_src = BKE_id_attributes_active_color_name(&mesh_src->id);
if ((data_type == CD_PROP_COLOR) &&
!BKE_id_attribute_search(
&mesh_src->id, active_color_src, CD_MASK_PROP_COLOR, ATTR_DOMAIN_MASK_COLOR)) {
return;
}
else if ((data_type == CD_PROP_BYTE_COLOR) &&
!BKE_id_attribute_search(
&mesh_src->id, active_color_src, CD_MASK_PROP_BYTE_COLOR, ATTR_DOMAIN_MASK_COLOR)) {
return;
}
if ((data_type == CD_PROP_COLOR) &&
BKE_id_attribute_search(
&mesh_dst->id, active_color_src, CD_MASK_PROP_COLOR, ATTR_DOMAIN_MASK_COLOR)) {
mesh_dst->active_color_attribute = BLI_strdup(active_color_src);
}
else if ((data_type == CD_PROP_BYTE_COLOR) &&
BKE_id_attribute_search(
&mesh_dst->id, active_color_src, CD_MASK_PROP_BYTE_COLOR, ATTR_DOMAIN_MASK_COLOR)) {
mesh_dst->active_color_attribute = BLI_strdup(active_color_src);
}
else {
CustomDataLayer *first_color_layer = BKE_id_attribute_from_index(
&mesh_dst->id, 0, mask_domain, CD_MASK_COLOR_ALL);
if (first_color_layer != nullptr) {
mesh_dst->active_color_attribute = BLI_strdup(first_color_layer->name);
}
}
}
/**
2023-02-27 20:54:27 +11:00
* When transferring color attributes, also transfer the default color attribute string.
* If a match cant be found, use the first color layer that can be found (to ensure a valid string
* is set).
*/
static void data_transfer_mesh_attributes_transfer_default_color_string(
Mesh *mesh_dst, Mesh *mesh_src, const eAttrDomainMask mask_domain, const int data_type)
{
if (mesh_dst->default_color_attribute) {
return;
}
const char *default_color_src = BKE_id_attributes_default_color_name(&mesh_src->id);
if ((data_type == CD_PROP_COLOR) &&
!BKE_id_attribute_search(
&mesh_src->id, default_color_src, CD_MASK_PROP_COLOR, ATTR_DOMAIN_MASK_COLOR)) {
return;
}
else if ((data_type == CD_PROP_BYTE_COLOR) && !BKE_id_attribute_search(&mesh_src->id,
default_color_src,
CD_MASK_PROP_BYTE_COLOR,
ATTR_DOMAIN_MASK_COLOR)) {
return;
}
if ((data_type == CD_PROP_COLOR) &&
BKE_id_attribute_search(
&mesh_dst->id, default_color_src, CD_MASK_PROP_COLOR, ATTR_DOMAIN_MASK_COLOR)) {
mesh_dst->default_color_attribute = BLI_strdup(default_color_src);
}
else if ((data_type == CD_PROP_BYTE_COLOR) && BKE_id_attribute_search(&mesh_dst->id,
default_color_src,
CD_MASK_PROP_BYTE_COLOR,
ATTR_DOMAIN_MASK_COLOR)) {
mesh_dst->default_color_attribute = BLI_strdup(default_color_src);
}
else {
CustomDataLayer *first_color_layer = BKE_id_attribute_from_index(
&mesh_dst->id, 0, mask_domain, CD_MASK_COLOR_ALL);
if (first_color_layer != nullptr) {
mesh_dst->default_color_attribute = BLI_strdup(first_color_layer->name);
}
}
}
/* ********** */
/* Generic pre/post processing, only used by custom loop normals currently. */
static void data_transfer_dtdata_type_preprocess(Mesh *me_src,
Mesh *me_dst,
const int dtdata_type,
const bool dirty_nors_dst)
{
if (dtdata_type == DT_TYPE_LNOR) {
/* Compute custom normals into regular loop normals, which will be used for the transfer. */
CustomData *ldata_dst = &me_dst->ldata;
const bool use_split_nors_dst = (me_dst->flag & ME_AUTOSMOOTH) != 0;
const float split_angle_dst = me_dst->smoothresh;
/* This should be ensured by cddata_masks we pass to code generating/giving us me_src now. */
BLI_assert(CustomData_get_layer(&me_src->ldata, CD_NORMAL) != nullptr);
(void)me_src;
short(*custom_nors_dst)[2] = static_cast<short(*)[2]>(
CustomData_get_layer_for_write(ldata_dst, CD_CUSTOMLOOPNORMAL, me_dst->totloop));
/* Cache loop nors into a temp CDLayer. */
blender::float3 *loop_nors_dst = static_cast<blender::float3 *>(
CustomData_get_layer_for_write(ldata_dst, CD_NORMAL, me_dst->totloop));
const bool do_loop_nors_dst = (loop_nors_dst == nullptr);
if (do_loop_nors_dst) {
loop_nors_dst = static_cast<blender::float3 *>(
CustomData_add_layer(ldata_dst, CD_NORMAL, CD_SET_DEFAULT, me_dst->totloop));
CustomData_set_layer_flag(ldata_dst, CD_NORMAL, CD_FLAG_TEMPORARY);
}
if (dirty_nors_dst || do_loop_nors_dst) {
const bool *sharp_edges = static_cast<const bool *>(
CustomData_get_layer_named(&me_dst->edata, CD_PROP_BOOL, "sharp_edge"));
Mesh: Move face shade smooth flag to a generic attribute Currently the shade smooth status for mesh faces is stored as part of `MPoly::flag`. As described in #95967, this moves that information to a separate boolean attribute. It also flips its status, so the attribute is now called `sharp_face`, which mirrors the existing `sharp_edge` attribute. The attribute doesn't need to be allocated when all faces are smooth. Forward compatibility is kept until 4.0 like the other mesh refactors. This will reduce memory bandwidth requirements for some operations, since the array of booleans uses 12 times less memory than `MPoly`. It also allows faces to be stored more efficiently in the future, since the flag is now unused. It's also possible to use generic functions to process the values. For example, finding whether there is a sharp face is just `sharp_faces.contains(true)`. The `shade_smooth` attribute is no longer accessible with geometry nodes. Since there were dedicated accessor nodes for that data, that shouldn't be a problem. That's difficult to version automatically since the named attribute nodes could be used in arbitrary combinations. **Implementation notes:** - The attribute and array variables in the code use the `sharp_faces` term, to be consistent with the user-facing "sharp faces" wording, and to avoid requiring many renames when #101689 is implemented. - Cycles now accesses smooth face status with the generic attribute, to avoid overhead. - Changing the zero-value from "smooth" to "flat" takes some care to make sure defaults are the same. - Versioning for the edge mode extrude node is particularly complex. New nodes are added by versioning to propagate the attribute in its old inverted state. - A lot of access is still done through the `CustomData` API rather than the attribute API because of a few functions. That can be cleaned up easily in the future. - In the future we would benefit from a way to store attributes as a single value for when all faces are sharp. Pull Request: https://projects.blender.org/blender/blender/pulls/104422
2023-03-08 15:36:18 +01:00
const bool *sharp_faces = static_cast<const bool *>(
CustomData_get_layer_named(&me_dst->pdata, CD_PROP_BOOL, "sharp_face"));
blender::bke::mesh::normals_calc_loop(me_dst->vert_positions(),
me_dst->edges(),
me_dst->polys(),
me_dst->loops(),
{},
me_dst->vert_normals(),
me_dst->poly_normals(),
sharp_edges,
sharp_faces,
use_split_nors_dst,
split_angle_dst,
custom_nors_dst,
nullptr,
{loop_nors_dst, me_dst->totloop});
}
}
}
static void data_transfer_dtdata_type_postprocess(Object * /*ob_src*/,
Object * /*ob_dst*/,
Mesh * /*me_src*/,
Mesh *me_dst,
const int dtdata_type,
const bool changed)
{
using namespace blender;
if (dtdata_type == DT_TYPE_LNOR) {
2019-04-26 17:20:45 +02:00
if (!changed) {
return;
}
/* Bake edited destination loop normals into custom normals again. */
CustomData *ldata_dst = &me_dst->ldata;
blender::float3 *loop_nors_dst = static_cast<blender::float3 *>(
CustomData_get_layer_for_write(ldata_dst, CD_NORMAL, me_dst->totloop));
short(*custom_nors_dst)[2] = static_cast<short(*)[2]>(
CustomData_get_layer_for_write(ldata_dst, CD_CUSTOMLOOPNORMAL, me_dst->totloop));
if (!custom_nors_dst) {
custom_nors_dst = static_cast<short(*)[2]>(
CustomData_add_layer(ldata_dst, CD_CUSTOMLOOPNORMAL, CD_SET_DEFAULT, me_dst->totloop));
}
bke::MutableAttributeAccessor attributes = me_dst->attributes_for_write();
bke::SpanAttributeWriter<bool> sharp_edges = attributes.lookup_or_add_for_write_span<bool>(
"sharp_edge", ATTR_DOMAIN_EDGE);
const bool *sharp_faces = static_cast<const bool *>(
CustomData_get_layer_named(&me_dst->pdata, CD_PROP_BOOL, "sharp_face"));
/* Note loop_nors_dst contains our custom normals as transferred from source... */
blender::bke::mesh::normals_loop_custom_set(me_dst->vert_positions(),
me_dst->edges(),
me_dst->polys(),
me_dst->loops(),
me_dst->vert_normals(),
me_dst->poly_normals(),
sharp_faces,
sharp_edges.span,
{loop_nors_dst, me_dst->totloop},
custom_nors_dst);
sharp_edges.finish();
}
}
/* ********** */
static MeshRemapIslandsCalc data_transfer_get_loop_islands_generator(const int cddata_type)
{
switch (cddata_type) {
case CD_FAKE_UV:
return BKE_mesh_calc_islands_loop_poly_edgeseam;
default:
break;
}
return nullptr;
}
float data_transfer_interp_float_do(const int mix_mode,
const float val_dst,
const float val_src,
const float mix_factor)
{
float val_ret;
if ((mix_mode == CDT_MIX_REPLACE_ABOVE_THRESHOLD && (val_dst < mix_factor)) ||
(mix_mode == CDT_MIX_REPLACE_BELOW_THRESHOLD && (val_dst > mix_factor))) {
return val_dst; /* Do not affect destination. */
}
switch (mix_mode) {
case CDT_MIX_REPLACE_ABOVE_THRESHOLD:
case CDT_MIX_REPLACE_BELOW_THRESHOLD:
return val_src;
case CDT_MIX_MIX:
val_ret = (val_dst + val_src) * 0.5f;
break;
case CDT_MIX_ADD:
val_ret = val_dst + val_src;
break;
case CDT_MIX_SUB:
val_ret = val_dst - val_src;
break;
case CDT_MIX_MUL:
val_ret = val_dst * val_src;
break;
case CDT_MIX_TRANSFER:
default:
val_ret = val_src;
break;
}
return interpf(val_ret, val_dst, mix_factor);
}
/* Helpers to match sources and destinations data layers
* (also handles 'conversions' in CD_FAKE cases). */
void data_transfer_layersmapping_add_item(ListBase *r_map,
const int cddata_type,
const int mix_mode,
const float mix_factor,
const float *mix_weights,
const void *data_src,
void *data_dst,
const int data_src_n,
const int data_dst_n,
const size_t elem_size,
const size_t data_size,
const size_t data_offset,
const uint64_t data_flag,
cd_datatransfer_interp interp,
void *interp_data)
{
CustomDataTransferLayerMap *item = MEM_new<CustomDataTransferLayerMap>(__func__);
BLI_assert(data_dst != nullptr);
item->data_type = cddata_type;
item->mix_mode = mix_mode;
item->mix_factor = mix_factor;
item->mix_weights = mix_weights;
item->data_src = data_src;
item->data_dst = data_dst;
item->data_src_n = data_src_n;
item->data_dst_n = data_dst_n;
item->elem_size = elem_size;
item->data_size = data_size;
item->data_offset = data_offset;
item->data_flag = data_flag;
item->interp = interp;
item->interp_data = interp_data;
BLI_addtail(r_map, item);
}
static void data_transfer_layersmapping_add_item_cd(ListBase *r_map,
const int cddata_type,
const int mix_mode,
const float mix_factor,
const float *mix_weights,
const void *data_src,
void *data_dst,
cd_datatransfer_interp interp,
void *interp_data)
{
uint64_t data_flag = 0;
if (cddata_type == CD_FREESTYLE_EDGE) {
data_flag = FREESTYLE_EDGE_MARK;
}
else if (cddata_type == CD_FREESTYLE_FACE) {
data_flag = FREESTYLE_FACE_MARK;
}
data_transfer_layersmapping_add_item(r_map,
cddata_type,
mix_mode,
mix_factor,
mix_weights,
data_src,
data_dst,
0,
0,
0,
0,
0,
data_flag,
interp,
interp_data);
}
/**
* \note
* All those layer mapping handlers return false *only* if they were given invalid parameters.
* This means that even if they do nothing, they will return true if all given parameters were OK.
* Also, r_map may be nullptr, in which case they will 'only' create/delete destination layers
* according to given parameters.
*/
static bool data_transfer_layersmapping_cdlayers_multisrc_to_dst(ListBase *r_map,
const int cddata_type,
const int mix_mode,
const float mix_factor,
const float *mix_weights,
const int num_elem_dst,
const bool use_create,
const bool use_delete,
const CustomData *cd_src,
CustomData *cd_dst,
const int tolayers,
const bool *use_layers_src,
const int num_layers_src,
cd_datatransfer_interp interp,
void *interp_data)
{
const void *data_src;
void *data_dst = nullptr;
int idx_src = num_layers_src;
int idx_dst, tot_dst = CustomData_number_of_layers(cd_dst, cddata_type);
bool *data_dst_to_delete = nullptr;
if (!use_layers_src) {
/* No source at all, we can only delete all dest if requested... */
if (use_delete) {
idx_dst = tot_dst;
while (idx_dst--) {
CustomData_free_layer(cd_dst, cddata_type, num_elem_dst, idx_dst);
}
}
return true;
}
switch (tolayers) {
case DT_LAYERS_INDEX_DST:
idx_dst = tot_dst;
/* Find last source actually used! */
while (idx_src-- && !use_layers_src[idx_src]) {
/* pass */
}
idx_src++;
if (idx_dst < idx_src) {
if (use_create) {
/* Create as much data layers as necessary! */
for (; idx_dst < idx_src; idx_dst++) {
CustomData_add_layer(
cd_dst, eCustomDataType(cddata_type), CD_SET_DEFAULT, num_elem_dst);
}
}
else {
/* Otherwise, just try to map what we can with existing dst data layers. */
idx_src = idx_dst;
}
}
else if (use_delete && idx_dst > idx_src) {
while (idx_dst-- > idx_src) {
CustomData_free_layer(cd_dst, cddata_type, num_elem_dst, idx_dst);
}
}
if (r_map) {
while (idx_src--) {
if (!use_layers_src[idx_src]) {
continue;
}
data_src = CustomData_get_layer_n(cd_src, cddata_type, idx_src);
data_dst = CustomData_get_layer_n_for_write(cd_dst, cddata_type, idx_src, num_elem_dst);
data_transfer_layersmapping_add_item_cd(r_map,
cddata_type,
mix_mode,
mix_factor,
mix_weights,
data_src,
data_dst,
interp,
interp_data);
}
}
break;
case DT_LAYERS_NAME_DST:
if (use_delete) {
if (tot_dst) {
data_dst_to_delete = static_cast<bool *>(
MEM_mallocN(sizeof(*data_dst_to_delete) * size_t(tot_dst), __func__));
memset(data_dst_to_delete, true, sizeof(*data_dst_to_delete) * size_t(tot_dst));
}
}
while (idx_src--) {
const char *name;
if (!use_layers_src[idx_src]) {
continue;
}
name = CustomData_get_layer_name(cd_src, cddata_type, idx_src);
data_src = CustomData_get_layer_n(cd_src, cddata_type, idx_src);
if ((idx_dst = CustomData_get_named_layer(cd_dst, cddata_type, name)) == -1) {
if (use_create) {
2022-08-30 14:54:53 -05:00
CustomData_add_layer_named(
cd_dst, eCustomDataType(cddata_type), CD_SET_DEFAULT, num_elem_dst, name);
idx_dst = CustomData_get_named_layer(cd_dst, cddata_type, name);
}
else {
/* If we are not allowed to create missing dst data layers,
* just skip matching src one. */
continue;
}
}
else if (data_dst_to_delete) {
data_dst_to_delete[idx_dst] = false;
}
if (r_map) {
data_dst = CustomData_get_layer_n_for_write(cd_dst, cddata_type, idx_dst, num_elem_dst);
data_transfer_layersmapping_add_item_cd(r_map,
cddata_type,
mix_mode,
mix_factor,
mix_weights,
data_src,
data_dst,
interp,
interp_data);
}
}
if (data_dst_to_delete) {
/* NOTE:
* This won't affect newly created layers, if any, since tot_dst has not been updated!
* Also, looping backward ensures us we do not suffer
* from index shifting when deleting a layer. */
for (idx_dst = tot_dst; idx_dst--;) {
if (data_dst_to_delete[idx_dst]) {
CustomData_free_layer(cd_dst, cddata_type, num_elem_dst, idx_dst);
}
}
MEM_freeN(data_dst_to_delete);
}
break;
default:
return false;
}
return true;
}
static bool data_transfer_layersmapping_cdlayers(ListBase *r_map,
const int cddata_type,
const int mix_mode,
const float mix_factor,
const float *mix_weights,
const int num_elem_dst,
const bool use_create,
const bool use_delete,
const CustomData *cd_src,
CustomData *cd_dst,
const int fromlayers,
const int tolayers,
cd_datatransfer_interp interp,
void *interp_data)
{
int idx_src, idx_dst;
const void *data_src;
void *data_dst = nullptr;
if (CustomData_layertype_is_singleton(cddata_type)) {
if (!(data_src = CustomData_get_layer(cd_src, cddata_type))) {
if (use_delete) {
CustomData_free_layer(cd_dst, cddata_type, num_elem_dst, 0);
}
return true;
}
data_dst = CustomData_get_layer_for_write(cd_dst, cddata_type, num_elem_dst);
if (!data_dst) {
if (!use_create) {
return true;
}
data_dst = CustomData_add_layer(
cd_dst, eCustomDataType(cddata_type), CD_SET_DEFAULT, num_elem_dst);
}
if (r_map) {
data_transfer_layersmapping_add_item_cd(r_map,
cddata_type,
mix_mode,
mix_factor,
mix_weights,
data_src,
data_dst,
interp,
interp_data);
}
}
else if (fromlayers == DT_LAYERS_ACTIVE_SRC || fromlayers >= 0) {
/* NOTE: use_delete has not much meaning in this case, ignored. */
if (fromlayers >= 0) { /* Real-layer index */
idx_src = fromlayers;
}
else {
if ((idx_src = CustomData_get_active_layer(cd_src, cddata_type)) == -1) {
return true;
}
}
data_src = CustomData_get_layer_n(cd_src, cddata_type, idx_src);
if (!data_src) {
return true;
}
if (tolayers >= 0) { /* Real-layer index */
idx_dst = tolayers;
data_dst = CustomData_get_layer_n_for_write(cd_dst, cddata_type, idx_dst, num_elem_dst);
}
else if (tolayers == DT_LAYERS_ACTIVE_DST) {
if ((idx_dst = CustomData_get_active_layer(cd_dst, cddata_type)) == -1) {
if (!use_create) {
return true;
}
data_dst = CustomData_add_layer(
cd_dst, eCustomDataType(cddata_type), CD_SET_DEFAULT, num_elem_dst);
}
else {
data_dst = CustomData_get_layer_n_for_write(cd_dst, cddata_type, idx_dst, num_elem_dst);
}
}
else if (tolayers == DT_LAYERS_INDEX_DST) {
int num = CustomData_number_of_layers(cd_dst, cddata_type);
idx_dst = idx_src;
if (num <= idx_dst) {
if (!use_create) {
return true;
}
/* Create as much data layers as necessary! */
for (; num <= idx_dst; num++) {
CustomData_add_layer(cd_dst, eCustomDataType(cddata_type), CD_SET_DEFAULT, num_elem_dst);
}
}
data_dst = CustomData_get_layer_n_for_write(cd_dst, cddata_type, idx_dst, num_elem_dst);
}
else if (tolayers == DT_LAYERS_NAME_DST) {
const char *name = CustomData_get_layer_name(cd_src, cddata_type, idx_src);
if ((idx_dst = CustomData_get_named_layer(cd_dst, cddata_type, name)) == -1) {
if (!use_create) {
return true;
}
CustomData_add_layer_named(
cd_dst, eCustomDataType(cddata_type), CD_SET_DEFAULT, num_elem_dst, name);
idx_dst = CustomData_get_named_layer(cd_dst, cddata_type, name);
}
data_dst = CustomData_get_layer_n_for_write(cd_dst, cddata_type, idx_dst, num_elem_dst);
}
else {
return false;
}
if (!data_dst) {
return false;
}
if (r_map) {
data_transfer_layersmapping_add_item_cd(r_map,
cddata_type,
mix_mode,
mix_factor,
mix_weights,
data_src,
data_dst,
interp,
interp_data);
}
}
else if (fromlayers == DT_LAYERS_ALL_SRC) {
int num_src = CustomData_number_of_layers(cd_src, cddata_type);
bool *use_layers_src = num_src ? static_cast<bool *>(MEM_mallocN(
sizeof(*use_layers_src) * size_t(num_src), __func__)) :
nullptr;
bool ret;
if (use_layers_src) {
memset(use_layers_src, true, sizeof(*use_layers_src) * num_src);
}
ret = data_transfer_layersmapping_cdlayers_multisrc_to_dst(r_map,
cddata_type,
mix_mode,
mix_factor,
mix_weights,
num_elem_dst,
use_create,
use_delete,
cd_src,
cd_dst,
tolayers,
use_layers_src,
num_src,
interp,
interp_data);
if (use_layers_src) {
MEM_freeN(use_layers_src);
}
return ret;
}
else {
return false;
}
return true;
}
static bool data_transfer_layersmapping_generate(ListBase *r_map,
Object *ob_src,
Object *ob_dst,
Mesh *me_src,
Mesh *me_dst,
const int elem_type,
int cddata_type,
int mix_mode,
float mix_factor,
const float *mix_weights,
const int num_elem_dst,
const bool use_create,
const bool use_delete,
const int fromlayers,
const int tolayers,
SpaceTransform *space_transform)
{
CustomData *cd_src, *cd_dst;
cd_datatransfer_interp interp = nullptr;
void *interp_data = nullptr;
if (elem_type == ME_VERT) {
if (!(cddata_type & CD_FAKE)) {
cd_src = &me_src->vdata;
cd_dst = &me_dst->vdata;
if (!data_transfer_layersmapping_cdlayers(r_map,
cddata_type,
mix_mode,
mix_factor,
mix_weights,
num_elem_dst,
use_create,
use_delete,
cd_src,
cd_dst,
fromlayers,
tolayers,
interp,
interp_data)) {
/* We handle specific source selection cases here. */
return false;
}
return true;
}
if (cddata_type == CD_FAKE_MDEFORMVERT) {
bool ret;
cd_src = &me_src->vdata;
cd_dst = &me_dst->vdata;
ret = data_transfer_layersmapping_vgroups(r_map,
mix_mode,
mix_factor,
mix_weights,
num_elem_dst,
use_create,
use_delete,
ob_src,
ob_dst,
cd_src,
cd_dst,
me_dst != ob_dst->data,
fromlayers,
tolayers);
return ret;
}
if (cddata_type == CD_FAKE_SHAPEKEY) {
/* TODO: leaving shape-keys aside for now, quite specific case,
Mesh: Move positions to a generic attribute **Changes** As described in T93602, this patch removes all use of the `MVert` struct, replacing it with a generic named attribute with the name `"position"`, consistent with other geometry types. Variable names have been changed from `verts` to `positions`, to align with the attribute name and the more generic design (positions are not vertices, they are just an attribute stored on the point domain). This change is made possible by previous commits that moved all other data out of `MVert` to runtime data or other generic attributes. What remains is mostly a simple type change. Though, the type still shows up 859 times, so the patch is quite large. One compromise is that now `CD_MASK_BAREMESH` now contains `CD_PROP_FLOAT3`. With the general move towards generic attributes over custom data types, we are removing use of these type masks anyway. **Benefits** The most obvious benefit is reduced memory usage and the benefits that brings in memory-bound situations. `float3` is only 3 bytes, in comparison to `MVert` which was 4. When there are millions of vertices this starts to matter more. The other benefits come from using a more generic type. Instead of writing algorithms specifically for `MVert`, code can just use arrays of vectors. This will allow eliminating many temporary arrays or wrappers used to extract positions. Many possible improvements aren't implemented in this patch, though I did switch simplify or remove the process of creating temporary position arrays in a few places. The design clarity that "positions are just another attribute" brings allows removing explicit copying of vertices in some procedural operations-- they are just processed like most other attributes. **Performance** This touches so many areas that it's hard to benchmark exhaustively, but I observed some areas as examples. * The mesh line node with 4 million count was 1.5x (8ms to 12ms) faster. * The Spring splash screen went from ~4.3 to ~4.5 fps. * The subdivision surface modifier/node was slightly faster RNA access through Python may be slightly slower, since now we need a name lookup instead of just a custom data type lookup for each index. **Future Improvements** * Remove uses of "vert_coords" functions: * `BKE_mesh_vert_coords_alloc` * `BKE_mesh_vert_coords_get` * `BKE_mesh_vert_coords_apply{_with_mat4}` * Remove more hidden copying of positions * General simplification now possible in many areas * Convert more code to C++ to use `float3` instead of `float[3]` * Currently `reinterpret_cast` is used for those C-API functions Differential Revision: https://developer.blender.org/D15982
2023-01-10 00:10:43 -05:00
* since we can't access them from mesh vertices :/ */
return false;
}
}
else if (elem_type == ME_EDGE) {
if (!(cddata_type & CD_FAKE)) { /* Unused for edges, currently... */
cd_src = &me_src->edata;
cd_dst = &me_dst->edata;
if (!data_transfer_layersmapping_cdlayers(r_map,
cddata_type,
mix_mode,
mix_factor,
mix_weights,
num_elem_dst,
use_create,
use_delete,
cd_src,
cd_dst,
fromlayers,
tolayers,
interp,
interp_data)) {
/* We handle specific source selection cases here. */
return false;
}
return true;
}
if (r_map && cddata_type == CD_FAKE_SEAM) {
if (!CustomData_get_layer_named(&me_dst->edata, CD_PROP_BOOL, ".uv_seam")) {
CustomData_add_layer_named(
&me_dst->edata, CD_PROP_BOOL, CD_SET_DEFAULT, me_dst->totedge, ".uv_seam");
}
data_transfer_layersmapping_add_item_cd(
r_map,
CD_PROP_BOOL,
mix_mode,
mix_factor,
mix_weights,
CustomData_get_layer_named(&me_src->edata, CD_PROP_BOOL, ".uv_seam"),
CustomData_get_layer_named_for_write(
&me_dst->edata, CD_PROP_BOOL, ".uv_seam", me_dst->totedge),
interp,
interp_data);
return true;
}
if (r_map && cddata_type == CD_FAKE_SHARP) {
if (!CustomData_get_layer_named(&me_dst->edata, CD_PROP_BOOL, "sharp_edge")) {
CustomData_add_layer_named(
&me_dst->edata, CD_PROP_BOOL, CD_SET_DEFAULT, me_dst->totedge, "sharp_edge");
}
data_transfer_layersmapping_add_item_cd(
r_map,
CD_PROP_BOOL,
mix_mode,
mix_factor,
mix_weights,
CustomData_get_layer_named(&me_src->edata, CD_PROP_BOOL, "sharp_edge"),
CustomData_get_layer_named_for_write(
&me_dst->edata, CD_PROP_BOOL, "sharp_edge", me_dst->totedge),
interp,
interp_data);
return true;
}
return false;
}
else if (elem_type == ME_LOOP) {
if (cddata_type == CD_FAKE_UV) {
Mesh: Move UV layers to generic attributes Currently the `MLoopUV` struct stores UV coordinates and flags related to editing UV maps in the UV editor. This patch changes the coordinates to use the generic 2D vector type, and moves the flags into three separate boolean attributes. This follows the design in T95965, with the ultimate intention of simplifying code and improving performance. Importantly, the change allows exporters and renderers to use UVs "touched" by geometry nodes, which only creates generic attributes. It also allows geometry nodes to create "proper" UV maps from scratch, though only with the Store Named Attribute node for now. The new design considers any 2D vector attribute on the corner domain to be a UV map. In the future, they might be distinguished from regular 2D vectors with attribute metadata, which may be helpful because they are often interpolated differently. Most of the code changes deal with passing around UV BMesh custom data offsets and tracking the boolean "sublayers". The boolean layers are use the following prefixes for attribute names: vert selection: `.vs.`, edge selection: `.es.`, pinning: `.pn.`. Currently these are short to avoid using up the maximum length of attribute names. To accommodate for these 4 extra characters, the name length limit is enlarged to 68 bytes, while the maximum user settable name length is still 64 bytes. Unfortunately Python/RNA API access to the UV flag data becomes slower. Accessing the boolean layers directly is be better for performance in general. Like the other mesh SoA refactors, backward and forward compatibility aren't affected, and won't be changed until 4.0. We pay for that by making mesh reading and writing more expensive with conversions. Resolves T85962 Differential Revision: https://developer.blender.org/D14365
2023-01-10 00:47:04 -05:00
cddata_type = CD_PROP_FLOAT2;
}
else if (cddata_type == CD_FAKE_LNOR) {
/* Pre-process should have generated it,
* Post-process will convert it back to CD_CUSTOMLOOPNORMAL. */
cddata_type = CD_NORMAL;
interp_data = space_transform;
interp = customdata_data_transfer_interp_normal_normals;
}
if (!(cddata_type & CD_FAKE)) {
cd_src = &me_src->ldata;
cd_dst = &me_dst->ldata;
if (!data_transfer_layersmapping_cdlayers(r_map,
cddata_type,
mix_mode,
mix_factor,
mix_weights,
num_elem_dst,
use_create,
use_delete,
cd_src,
cd_dst,
fromlayers,
tolayers,
interp,
interp_data)) {
/* We handle specific source selection cases here. */
return false;
}
return true;
}
return false;
}
else if (elem_type == ME_POLY) {
if (cddata_type == CD_FAKE_UV) {
Mesh: Move UV layers to generic attributes Currently the `MLoopUV` struct stores UV coordinates and flags related to editing UV maps in the UV editor. This patch changes the coordinates to use the generic 2D vector type, and moves the flags into three separate boolean attributes. This follows the design in T95965, with the ultimate intention of simplifying code and improving performance. Importantly, the change allows exporters and renderers to use UVs "touched" by geometry nodes, which only creates generic attributes. It also allows geometry nodes to create "proper" UV maps from scratch, though only with the Store Named Attribute node for now. The new design considers any 2D vector attribute on the corner domain to be a UV map. In the future, they might be distinguished from regular 2D vectors with attribute metadata, which may be helpful because they are often interpolated differently. Most of the code changes deal with passing around UV BMesh custom data offsets and tracking the boolean "sublayers". The boolean layers are use the following prefixes for attribute names: vert selection: `.vs.`, edge selection: `.es.`, pinning: `.pn.`. Currently these are short to avoid using up the maximum length of attribute names. To accommodate for these 4 extra characters, the name length limit is enlarged to 68 bytes, while the maximum user settable name length is still 64 bytes. Unfortunately Python/RNA API access to the UV flag data becomes slower. Accessing the boolean layers directly is be better for performance in general. Like the other mesh SoA refactors, backward and forward compatibility aren't affected, and won't be changed until 4.0. We pay for that by making mesh reading and writing more expensive with conversions. Resolves T85962 Differential Revision: https://developer.blender.org/D14365
2023-01-10 00:47:04 -05:00
cddata_type = CD_PROP_FLOAT2;
}
if (!(cddata_type & CD_FAKE)) {
cd_src = &me_src->pdata;
cd_dst = &me_dst->pdata;
if (!data_transfer_layersmapping_cdlayers(r_map,
cddata_type,
mix_mode,
mix_factor,
mix_weights,
num_elem_dst,
use_create,
use_delete,
cd_src,
cd_dst,
fromlayers,
tolayers,
interp,
interp_data)) {
/* We handle specific source selection cases here. */
return false;
}
return true;
}
if (r_map && cddata_type == CD_FAKE_SHARP) {
Mesh: Move face shade smooth flag to a generic attribute Currently the shade smooth status for mesh faces is stored as part of `MPoly::flag`. As described in #95967, this moves that information to a separate boolean attribute. It also flips its status, so the attribute is now called `sharp_face`, which mirrors the existing `sharp_edge` attribute. The attribute doesn't need to be allocated when all faces are smooth. Forward compatibility is kept until 4.0 like the other mesh refactors. This will reduce memory bandwidth requirements for some operations, since the array of booleans uses 12 times less memory than `MPoly`. It also allows faces to be stored more efficiently in the future, since the flag is now unused. It's also possible to use generic functions to process the values. For example, finding whether there is a sharp face is just `sharp_faces.contains(true)`. The `shade_smooth` attribute is no longer accessible with geometry nodes. Since there were dedicated accessor nodes for that data, that shouldn't be a problem. That's difficult to version automatically since the named attribute nodes could be used in arbitrary combinations. **Implementation notes:** - The attribute and array variables in the code use the `sharp_faces` term, to be consistent with the user-facing "sharp faces" wording, and to avoid requiring many renames when #101689 is implemented. - Cycles now accesses smooth face status with the generic attribute, to avoid overhead. - Changing the zero-value from "smooth" to "flat" takes some care to make sure defaults are the same. - Versioning for the edge mode extrude node is particularly complex. New nodes are added by versioning to propagate the attribute in its old inverted state. - A lot of access is still done through the `CustomData` API rather than the attribute API because of a few functions. That can be cleaned up easily in the future. - In the future we would benefit from a way to store attributes as a single value for when all faces are sharp. Pull Request: https://projects.blender.org/blender/blender/pulls/104422
2023-03-08 15:36:18 +01:00
if (!CustomData_get_layer_named(&me_dst->pdata, CD_PROP_BOOL, "sharp_face")) {
CustomData_add_layer_named(
&me_dst->pdata, CD_PROP_BOOL, CD_SET_DEFAULT, me_dst->totpoly, "sharp_face");
Mesh: Move face shade smooth flag to a generic attribute Currently the shade smooth status for mesh faces is stored as part of `MPoly::flag`. As described in #95967, this moves that information to a separate boolean attribute. It also flips its status, so the attribute is now called `sharp_face`, which mirrors the existing `sharp_edge` attribute. The attribute doesn't need to be allocated when all faces are smooth. Forward compatibility is kept until 4.0 like the other mesh refactors. This will reduce memory bandwidth requirements for some operations, since the array of booleans uses 12 times less memory than `MPoly`. It also allows faces to be stored more efficiently in the future, since the flag is now unused. It's also possible to use generic functions to process the values. For example, finding whether there is a sharp face is just `sharp_faces.contains(true)`. The `shade_smooth` attribute is no longer accessible with geometry nodes. Since there were dedicated accessor nodes for that data, that shouldn't be a problem. That's difficult to version automatically since the named attribute nodes could be used in arbitrary combinations. **Implementation notes:** - The attribute and array variables in the code use the `sharp_faces` term, to be consistent with the user-facing "sharp faces" wording, and to avoid requiring many renames when #101689 is implemented. - Cycles now accesses smooth face status with the generic attribute, to avoid overhead. - Changing the zero-value from "smooth" to "flat" takes some care to make sure defaults are the same. - Versioning for the edge mode extrude node is particularly complex. New nodes are added by versioning to propagate the attribute in its old inverted state. - A lot of access is still done through the `CustomData` API rather than the attribute API because of a few functions. That can be cleaned up easily in the future. - In the future we would benefit from a way to store attributes as a single value for when all faces are sharp. Pull Request: https://projects.blender.org/blender/blender/pulls/104422
2023-03-08 15:36:18 +01:00
}
data_transfer_layersmapping_add_item_cd(
r_map,
CD_PROP_BOOL,
mix_mode,
mix_factor,
mix_weights,
CustomData_get_layer_named(&me_src->pdata, CD_PROP_BOOL, "sharp_face"),
CustomData_get_layer_named_for_write(
&me_dst->pdata, CD_PROP_BOOL, "sharp_face", num_elem_dst),
interp,
interp_data);
return true;
}
return false;
}
return false;
}
void BKE_object_data_transfer_layout(struct Depsgraph *depsgraph,
Scene *scene,
Object *ob_src,
Object *ob_dst,
const int data_types,
const bool use_delete,
const int fromlayers_select[DT_MULTILAYER_INDEX_MAX],
const int tolayers_select[DT_MULTILAYER_INDEX_MAX])
{
Mesh *me_src;
Mesh *me_dst;
const bool use_create = true; /* We always create needed layers here. */
CustomData_MeshMasks me_src_mask = CD_MASK_BAREMESH;
BLI_assert((ob_src != ob_dst) && (ob_src->type == OB_MESH) && (ob_dst->type == OB_MESH));
me_dst = static_cast<Mesh *>(ob_dst->data);
/* Get source evaluated mesh. */
BKE_object_data_transfer_dttypes_to_cdmask(data_types, &me_src_mask);
me_src = mesh_get_eval_final(depsgraph, scene, ob_src, &me_src_mask);
if (!me_src) {
return;
}
/* Check all possible data types. */
2020-09-09 16:35:20 +02:00
for (int i = 0; i < DT_TYPE_MAX; i++) {
const int dtdata_type = 1 << i;
int cddata_type;
int fromlayers, tolayers, fromto_idx;
if (!(data_types & dtdata_type)) {
continue;
}
cddata_type = BKE_object_data_transfer_dttype_to_cdtype(dtdata_type);
fromto_idx = BKE_object_data_transfer_dttype_to_srcdst_index(dtdata_type);
if (fromto_idx != DT_MULTILAYER_INDEX_INVALID) {
fromlayers = fromlayers_select[fromto_idx];
tolayers = tolayers_select[fromto_idx];
}
else {
fromlayers = tolayers = 0;
}
if (DT_DATATYPE_IS_VERT(dtdata_type)) {
const int num_elem_dst = me_dst->totvert;
data_transfer_layersmapping_generate(nullptr,
ob_src,
ob_dst,
me_src,
me_dst,
ME_VERT,
cddata_type,
0,
0.0f,
nullptr,
num_elem_dst,
use_create,
use_delete,
fromlayers,
tolayers,
nullptr);
2023-02-27 20:54:27 +11:00
/* Make sure we have active/default color layers if none existed before.
* Use the active/default from src (if it was transferred), otherwise the first. */
if (ELEM(cddata_type, CD_PROP_COLOR, CD_PROP_BYTE_COLOR)) {
data_transfer_mesh_attributes_transfer_active_color_string(
me_dst, me_src, ATTR_DOMAIN_MASK_POINT, cddata_type);
data_transfer_mesh_attributes_transfer_default_color_string(
me_dst, me_src, ATTR_DOMAIN_MASK_POINT, cddata_type);
}
}
if (DT_DATATYPE_IS_EDGE(dtdata_type)) {
const int num_elem_dst = me_dst->totedge;
data_transfer_layersmapping_generate(nullptr,
ob_src,
ob_dst,
me_src,
me_dst,
ME_EDGE,
cddata_type,
0,
0.0f,
nullptr,
num_elem_dst,
use_create,
use_delete,
fromlayers,
tolayers,
nullptr);
}
if (DT_DATATYPE_IS_LOOP(dtdata_type)) {
const int num_elem_dst = me_dst->totloop;
data_transfer_layersmapping_generate(nullptr,
ob_src,
ob_dst,
me_src,
me_dst,
ME_LOOP,
cddata_type,
0,
0.0f,
nullptr,
num_elem_dst,
use_create,
use_delete,
fromlayers,
tolayers,
nullptr);
2023-02-27 20:54:27 +11:00
/* Make sure we have active/default color layers if none existed before.
* Use the active/default from src (if it was transferred), otherwise the first. */
if (ELEM(cddata_type, CD_PROP_COLOR, CD_PROP_BYTE_COLOR)) {
data_transfer_mesh_attributes_transfer_active_color_string(
me_dst, me_src, ATTR_DOMAIN_MASK_CORNER, cddata_type);
data_transfer_mesh_attributes_transfer_default_color_string(
me_dst, me_src, ATTR_DOMAIN_MASK_CORNER, cddata_type);
}
}
if (DT_DATATYPE_IS_POLY(dtdata_type)) {
const int num_elem_dst = me_dst->totpoly;
data_transfer_layersmapping_generate(nullptr,
ob_src,
ob_dst,
me_src,
me_dst,
ME_POLY,
cddata_type,
0,
0.0f,
nullptr,
num_elem_dst,
use_create,
use_delete,
fromlayers,
tolayers,
nullptr);
}
}
}
bool BKE_object_data_transfer_ex(struct Depsgraph *depsgraph,
Scene *scene,
Object *ob_src,
Object *ob_dst,
Mesh *me_dst,
const int data_types,
bool use_create,
const int map_vert_mode,
const int map_edge_mode,
const int map_loop_mode,
const int map_poly_mode,
SpaceTransform *space_transform,
const bool auto_transform,
const float max_distance,
const float ray_radius,
const float islands_handling_precision,
const int fromlayers_select[DT_MULTILAYER_INDEX_MAX],
const int tolayers_select[DT_MULTILAYER_INDEX_MAX],
const int mix_mode,
const float mix_factor,
const char *vgroup_name,
const bool invert_vgroup,
ReportList *reports)
{
#define VDATA 0
#define EDATA 1
#define LDATA 2
#define PDATA 3
#define DATAMAX 4
SpaceTransform auto_space_transform;
Mesh *me_src;
/* Assumed always true if not using an evaluated mesh as destination. */
bool dirty_nors_dst = true;
const MDeformVert *mdef = nullptr;
int vg_idx = -1;
float *weights[DATAMAX] = {nullptr};
MeshPairRemap geom_map[DATAMAX] = {{0}};
bool geom_map_init[DATAMAX] = {false};
ListBase lay_map = {nullptr};
bool changed = false;
bool is_modifier = false;
const bool use_delete = false; /* We never delete data layers from destination here. */
CustomData_MeshMasks me_src_mask = CD_MASK_BAREMESH;
BLI_assert((ob_src != ob_dst) && (ob_src->type == OB_MESH) && (ob_dst->type == OB_MESH));
if (me_dst) {
dirty_nors_dst = BKE_mesh_vert_normals_are_dirty(me_dst);
/* Never create needed custom layers on passed destination mesh
* (assumed to *not* be ob_dst->data, aka modifier case). */
use_create = false;
is_modifier = true;
}
else {
me_dst = static_cast<Mesh *>(ob_dst->data);
}
if (vgroup_name) {
mdef = static_cast<const MDeformVert *>(CustomData_get_layer(&me_dst->vdata, CD_MDEFORMVERT));
if (mdef) {
vg_idx = BKE_id_defgroup_name_index(&me_dst->id, vgroup_name);
}
}
/* Get source evaluated mesh. */
BKE_object_data_transfer_dttypes_to_cdmask(data_types, &me_src_mask);
BKE_mesh_remap_calc_source_cddata_masks_from_map_modes(
map_vert_mode, map_edge_mode, map_loop_mode, map_poly_mode, &me_src_mask);
if (is_modifier) {
me_src = BKE_modifier_get_evaluated_mesh_from_evaluated_object(ob_src);
if (me_src == nullptr ||
!CustomData_MeshMasks_are_matching(&ob_src->runtime.last_data_mask, &me_src_mask)) {
CLOG_WARN(&LOG, "Data Transfer: source mesh data is not ready - dependency cycle?");
return changed;
}
}
else {
me_src = mesh_get_eval_final(depsgraph, scene, ob_src, &me_src_mask);
}
if (!me_src) {
return changed;
}
BKE_mesh_wrapper_ensure_mdata(me_src);
if (auto_transform) {
if (space_transform == nullptr) {
space_transform = &auto_space_transform;
}
BKE_mesh_remap_find_best_match_from_mesh(
Mesh: Move positions to a generic attribute **Changes** As described in T93602, this patch removes all use of the `MVert` struct, replacing it with a generic named attribute with the name `"position"`, consistent with other geometry types. Variable names have been changed from `verts` to `positions`, to align with the attribute name and the more generic design (positions are not vertices, they are just an attribute stored on the point domain). This change is made possible by previous commits that moved all other data out of `MVert` to runtime data or other generic attributes. What remains is mostly a simple type change. Though, the type still shows up 859 times, so the patch is quite large. One compromise is that now `CD_MASK_BAREMESH` now contains `CD_PROP_FLOAT3`. With the general move towards generic attributes over custom data types, we are removing use of these type masks anyway. **Benefits** The most obvious benefit is reduced memory usage and the benefits that brings in memory-bound situations. `float3` is only 3 bytes, in comparison to `MVert` which was 4. When there are millions of vertices this starts to matter more. The other benefits come from using a more generic type. Instead of writing algorithms specifically for `MVert`, code can just use arrays of vectors. This will allow eliminating many temporary arrays or wrappers used to extract positions. Many possible improvements aren't implemented in this patch, though I did switch simplify or remove the process of creating temporary position arrays in a few places. The design clarity that "positions are just another attribute" brings allows removing explicit copying of vertices in some procedural operations-- they are just processed like most other attributes. **Performance** This touches so many areas that it's hard to benchmark exhaustively, but I observed some areas as examples. * The mesh line node with 4 million count was 1.5x (8ms to 12ms) faster. * The Spring splash screen went from ~4.3 to ~4.5 fps. * The subdivision surface modifier/node was slightly faster RNA access through Python may be slightly slower, since now we need a name lookup instead of just a custom data type lookup for each index. **Future Improvements** * Remove uses of "vert_coords" functions: * `BKE_mesh_vert_coords_alloc` * `BKE_mesh_vert_coords_get` * `BKE_mesh_vert_coords_apply{_with_mat4}` * Remove more hidden copying of positions * General simplification now possible in many areas * Convert more code to C++ to use `float3` instead of `float[3]` * Currently `reinterpret_cast` is used for those C-API functions Differential Revision: https://developer.blender.org/D15982
2023-01-10 00:10:43 -05:00
BKE_mesh_vert_positions(me_dst), me_dst->totvert, me_src, space_transform);
}
/* Check all possible data types.
* Note item mappings and dest mix weights are cached. */
2020-09-09 16:35:20 +02:00
for (int i = 0; i < DT_TYPE_MAX; i++) {
const int dtdata_type = 1 << i;
int cddata_type;
int fromlayers, tolayers, fromto_idx;
if (!(data_types & dtdata_type)) {
continue;
}
data_transfer_dtdata_type_preprocess(me_src, me_dst, dtdata_type, dirty_nors_dst);
cddata_type = BKE_object_data_transfer_dttype_to_cdtype(dtdata_type);
fromto_idx = BKE_object_data_transfer_dttype_to_srcdst_index(dtdata_type);
if (fromto_idx != DT_MULTILAYER_INDEX_INVALID) {
fromlayers = fromlayers_select[fromto_idx];
tolayers = tolayers_select[fromto_idx];
}
else {
fromlayers = tolayers = 0;
}
if (DT_DATATYPE_IS_VERT(dtdata_type)) {
Mesh: Move positions to a generic attribute **Changes** As described in T93602, this patch removes all use of the `MVert` struct, replacing it with a generic named attribute with the name `"position"`, consistent with other geometry types. Variable names have been changed from `verts` to `positions`, to align with the attribute name and the more generic design (positions are not vertices, they are just an attribute stored on the point domain). This change is made possible by previous commits that moved all other data out of `MVert` to runtime data or other generic attributes. What remains is mostly a simple type change. Though, the type still shows up 859 times, so the patch is quite large. One compromise is that now `CD_MASK_BAREMESH` now contains `CD_PROP_FLOAT3`. With the general move towards generic attributes over custom data types, we are removing use of these type masks anyway. **Benefits** The most obvious benefit is reduced memory usage and the benefits that brings in memory-bound situations. `float3` is only 3 bytes, in comparison to `MVert` which was 4. When there are millions of vertices this starts to matter more. The other benefits come from using a more generic type. Instead of writing algorithms specifically for `MVert`, code can just use arrays of vectors. This will allow eliminating many temporary arrays or wrappers used to extract positions. Many possible improvements aren't implemented in this patch, though I did switch simplify or remove the process of creating temporary position arrays in a few places. The design clarity that "positions are just another attribute" brings allows removing explicit copying of vertices in some procedural operations-- they are just processed like most other attributes. **Performance** This touches so many areas that it's hard to benchmark exhaustively, but I observed some areas as examples. * The mesh line node with 4 million count was 1.5x (8ms to 12ms) faster. * The Spring splash screen went from ~4.3 to ~4.5 fps. * The subdivision surface modifier/node was slightly faster RNA access through Python may be slightly slower, since now we need a name lookup instead of just a custom data type lookup for each index. **Future Improvements** * Remove uses of "vert_coords" functions: * `BKE_mesh_vert_coords_alloc` * `BKE_mesh_vert_coords_get` * `BKE_mesh_vert_coords_apply{_with_mat4}` * Remove more hidden copying of positions * General simplification now possible in many areas * Convert more code to C++ to use `float3` instead of `float[3]` * Currently `reinterpret_cast` is used for those C-API functions Differential Revision: https://developer.blender.org/D15982
2023-01-10 00:10:43 -05:00
float(*positions_dst)[3] = BKE_mesh_vert_positions_for_write(me_dst);
const int num_verts_dst = me_dst->totvert;
if (!geom_map_init[VDATA]) {
const int num_verts_src = me_src->totvert;
if ((map_vert_mode == MREMAP_MODE_TOPOLOGY) && (num_verts_dst != num_verts_src)) {
BKE_report(reports,
RPT_ERROR,
"Source and destination meshes do not have the same amount of vertices, "
"'Topology' mapping cannot be used in this case");
continue;
}
if ((map_vert_mode & MREMAP_USE_EDGE) && (me_src->totedge == 0)) {
BKE_report(reports,
RPT_ERROR,
"Source mesh doesn't have any edges, "
"None of the 'Edge' mappings can be used in this case");
continue;
}
if ((map_vert_mode & MREMAP_USE_POLY) && (me_src->totpoly == 0)) {
BKE_report(reports,
RPT_ERROR,
"Source mesh doesn't have any faces, "
"None of the 'Face' mappings can be used in this case");
continue;
}
if (ELEM(0, num_verts_dst, num_verts_src)) {
BKE_report(reports,
RPT_ERROR,
"Source or destination meshes do not have any vertices, cannot transfer "
"vertex data");
continue;
}
BKE_mesh_remap_calc_verts_from_mesh(map_vert_mode,
space_transform,
max_distance,
ray_radius,
Mesh: Move positions to a generic attribute **Changes** As described in T93602, this patch removes all use of the `MVert` struct, replacing it with a generic named attribute with the name `"position"`, consistent with other geometry types. Variable names have been changed from `verts` to `positions`, to align with the attribute name and the more generic design (positions are not vertices, they are just an attribute stored on the point domain). This change is made possible by previous commits that moved all other data out of `MVert` to runtime data or other generic attributes. What remains is mostly a simple type change. Though, the type still shows up 859 times, so the patch is quite large. One compromise is that now `CD_MASK_BAREMESH` now contains `CD_PROP_FLOAT3`. With the general move towards generic attributes over custom data types, we are removing use of these type masks anyway. **Benefits** The most obvious benefit is reduced memory usage and the benefits that brings in memory-bound situations. `float3` is only 3 bytes, in comparison to `MVert` which was 4. When there are millions of vertices this starts to matter more. The other benefits come from using a more generic type. Instead of writing algorithms specifically for `MVert`, code can just use arrays of vectors. This will allow eliminating many temporary arrays or wrappers used to extract positions. Many possible improvements aren't implemented in this patch, though I did switch simplify or remove the process of creating temporary position arrays in a few places. The design clarity that "positions are just another attribute" brings allows removing explicit copying of vertices in some procedural operations-- they are just processed like most other attributes. **Performance** This touches so many areas that it's hard to benchmark exhaustively, but I observed some areas as examples. * The mesh line node with 4 million count was 1.5x (8ms to 12ms) faster. * The Spring splash screen went from ~4.3 to ~4.5 fps. * The subdivision surface modifier/node was slightly faster RNA access through Python may be slightly slower, since now we need a name lookup instead of just a custom data type lookup for each index. **Future Improvements** * Remove uses of "vert_coords" functions: * `BKE_mesh_vert_coords_alloc` * `BKE_mesh_vert_coords_get` * `BKE_mesh_vert_coords_apply{_with_mat4}` * Remove more hidden copying of positions * General simplification now possible in many areas * Convert more code to C++ to use `float3` instead of `float[3]` * Currently `reinterpret_cast` is used for those C-API functions Differential Revision: https://developer.blender.org/D15982
2023-01-10 00:10:43 -05:00
positions_dst,
num_verts_dst,
dirty_nors_dst,
me_src,
me_dst,
&geom_map[VDATA]);
geom_map_init[VDATA] = true;
}
if (mdef && vg_idx != -1 && !weights[VDATA]) {
weights[VDATA] = static_cast<float *>(
MEM_mallocN(sizeof(*(weights[VDATA])) * size_t(num_verts_dst), __func__));
BKE_defvert_extract_vgroup_to_vertweights(
mdef, vg_idx, num_verts_dst, invert_vgroup, weights[VDATA]);
}
if (data_transfer_layersmapping_generate(&lay_map,
ob_src,
ob_dst,
me_src,
me_dst,
ME_VERT,
cddata_type,
mix_mode,
mix_factor,
weights[VDATA],
num_verts_dst,
use_create,
use_delete,
fromlayers,
tolayers,
space_transform)) {
CustomDataTransferLayerMap *lay_mapit;
changed |= (lay_map.first != nullptr);
for (lay_mapit = static_cast<CustomDataTransferLayerMap *>(lay_map.first); lay_mapit;
lay_mapit = lay_mapit->next) {
CustomData_data_transfer(&geom_map[VDATA], lay_mapit);
}
BLI_freelistN(&lay_map);
}
}
if (DT_DATATYPE_IS_EDGE(dtdata_type)) {
Mesh: Move positions to a generic attribute **Changes** As described in T93602, this patch removes all use of the `MVert` struct, replacing it with a generic named attribute with the name `"position"`, consistent with other geometry types. Variable names have been changed from `verts` to `positions`, to align with the attribute name and the more generic design (positions are not vertices, they are just an attribute stored on the point domain). This change is made possible by previous commits that moved all other data out of `MVert` to runtime data or other generic attributes. What remains is mostly a simple type change. Though, the type still shows up 859 times, so the patch is quite large. One compromise is that now `CD_MASK_BAREMESH` now contains `CD_PROP_FLOAT3`. With the general move towards generic attributes over custom data types, we are removing use of these type masks anyway. **Benefits** The most obvious benefit is reduced memory usage and the benefits that brings in memory-bound situations. `float3` is only 3 bytes, in comparison to `MVert` which was 4. When there are millions of vertices this starts to matter more. The other benefits come from using a more generic type. Instead of writing algorithms specifically for `MVert`, code can just use arrays of vectors. This will allow eliminating many temporary arrays or wrappers used to extract positions. Many possible improvements aren't implemented in this patch, though I did switch simplify or remove the process of creating temporary position arrays in a few places. The design clarity that "positions are just another attribute" brings allows removing explicit copying of vertices in some procedural operations-- they are just processed like most other attributes. **Performance** This touches so many areas that it's hard to benchmark exhaustively, but I observed some areas as examples. * The mesh line node with 4 million count was 1.5x (8ms to 12ms) faster. * The Spring splash screen went from ~4.3 to ~4.5 fps. * The subdivision surface modifier/node was slightly faster RNA access through Python may be slightly slower, since now we need a name lookup instead of just a custom data type lookup for each index. **Future Improvements** * Remove uses of "vert_coords" functions: * `BKE_mesh_vert_coords_alloc` * `BKE_mesh_vert_coords_get` * `BKE_mesh_vert_coords_apply{_with_mat4}` * Remove more hidden copying of positions * General simplification now possible in many areas * Convert more code to C++ to use `float3` instead of `float[3]` * Currently `reinterpret_cast` is used for those C-API functions Differential Revision: https://developer.blender.org/D15982
2023-01-10 00:10:43 -05:00
const float(*positions_dst)[3] = BKE_mesh_vert_positions_for_write(me_dst);
const int num_verts_dst = me_dst->totvert;
const blender::Span<MEdge> edges_dst = me_dst->edges();
if (!geom_map_init[EDATA]) {
const int num_edges_src = me_src->totedge;
if ((map_edge_mode == MREMAP_MODE_TOPOLOGY) && (edges_dst.size() != num_edges_src)) {
BKE_report(reports,
RPT_ERROR,
"Source and destination meshes do not have the same amount of edges, "
"'Topology' mapping cannot be used in this case");
continue;
}
if ((map_edge_mode & MREMAP_USE_POLY) && (me_src->totpoly == 0)) {
BKE_report(reports,
RPT_ERROR,
"Source mesh doesn't have any faces, "
"None of the 'Face' mappings can be used in this case");
continue;
}
if (ELEM(0, edges_dst.size(), num_edges_src)) {
BKE_report(
reports,
RPT_ERROR,
"Source or destination meshes do not have any edges, cannot transfer edge data");
continue;
}
BKE_mesh_remap_calc_edges_from_mesh(map_edge_mode,
space_transform,
max_distance,
ray_radius,
Mesh: Move positions to a generic attribute **Changes** As described in T93602, this patch removes all use of the `MVert` struct, replacing it with a generic named attribute with the name `"position"`, consistent with other geometry types. Variable names have been changed from `verts` to `positions`, to align with the attribute name and the more generic design (positions are not vertices, they are just an attribute stored on the point domain). This change is made possible by previous commits that moved all other data out of `MVert` to runtime data or other generic attributes. What remains is mostly a simple type change. Though, the type still shows up 859 times, so the patch is quite large. One compromise is that now `CD_MASK_BAREMESH` now contains `CD_PROP_FLOAT3`. With the general move towards generic attributes over custom data types, we are removing use of these type masks anyway. **Benefits** The most obvious benefit is reduced memory usage and the benefits that brings in memory-bound situations. `float3` is only 3 bytes, in comparison to `MVert` which was 4. When there are millions of vertices this starts to matter more. The other benefits come from using a more generic type. Instead of writing algorithms specifically for `MVert`, code can just use arrays of vectors. This will allow eliminating many temporary arrays or wrappers used to extract positions. Many possible improvements aren't implemented in this patch, though I did switch simplify or remove the process of creating temporary position arrays in a few places. The design clarity that "positions are just another attribute" brings allows removing explicit copying of vertices in some procedural operations-- they are just processed like most other attributes. **Performance** This touches so many areas that it's hard to benchmark exhaustively, but I observed some areas as examples. * The mesh line node with 4 million count was 1.5x (8ms to 12ms) faster. * The Spring splash screen went from ~4.3 to ~4.5 fps. * The subdivision surface modifier/node was slightly faster RNA access through Python may be slightly slower, since now we need a name lookup instead of just a custom data type lookup for each index. **Future Improvements** * Remove uses of "vert_coords" functions: * `BKE_mesh_vert_coords_alloc` * `BKE_mesh_vert_coords_get` * `BKE_mesh_vert_coords_apply{_with_mat4}` * Remove more hidden copying of positions * General simplification now possible in many areas * Convert more code to C++ to use `float3` instead of `float[3]` * Currently `reinterpret_cast` is used for those C-API functions Differential Revision: https://developer.blender.org/D15982
2023-01-10 00:10:43 -05:00
positions_dst,
num_verts_dst,
edges_dst.data(),
edges_dst.size(),
dirty_nors_dst,
me_src,
me_dst,
&geom_map[EDATA]);
geom_map_init[EDATA] = true;
}
if (mdef && vg_idx != -1 && !weights[EDATA]) {
weights[EDATA] = static_cast<float *>(
MEM_mallocN(sizeof(*weights[EDATA]) * size_t(edges_dst.size()), __func__));
BKE_defvert_extract_vgroup_to_edgeweights(mdef,
vg_idx,
num_verts_dst,
edges_dst.data(),
edges_dst.size(),
invert_vgroup,
weights[EDATA]);
}
if (data_transfer_layersmapping_generate(&lay_map,
ob_src,
ob_dst,
me_src,
me_dst,
ME_EDGE,
cddata_type,
mix_mode,
mix_factor,
weights[EDATA],
edges_dst.size(),
use_create,
use_delete,
fromlayers,
tolayers,
space_transform)) {
CustomDataTransferLayerMap *lay_mapit;
changed |= (lay_map.first != nullptr);
for (lay_mapit = static_cast<CustomDataTransferLayerMap *>(lay_map.first); lay_mapit;
lay_mapit = lay_mapit->next) {
CustomData_data_transfer(&geom_map[EDATA], lay_mapit);
}
BLI_freelistN(&lay_map);
}
}
if (DT_DATATYPE_IS_LOOP(dtdata_type)) {
Mesh: Move positions to a generic attribute **Changes** As described in T93602, this patch removes all use of the `MVert` struct, replacing it with a generic named attribute with the name `"position"`, consistent with other geometry types. Variable names have been changed from `verts` to `positions`, to align with the attribute name and the more generic design (positions are not vertices, they are just an attribute stored on the point domain). This change is made possible by previous commits that moved all other data out of `MVert` to runtime data or other generic attributes. What remains is mostly a simple type change. Though, the type still shows up 859 times, so the patch is quite large. One compromise is that now `CD_MASK_BAREMESH` now contains `CD_PROP_FLOAT3`. With the general move towards generic attributes over custom data types, we are removing use of these type masks anyway. **Benefits** The most obvious benefit is reduced memory usage and the benefits that brings in memory-bound situations. `float3` is only 3 bytes, in comparison to `MVert` which was 4. When there are millions of vertices this starts to matter more. The other benefits come from using a more generic type. Instead of writing algorithms specifically for `MVert`, code can just use arrays of vectors. This will allow eliminating many temporary arrays or wrappers used to extract positions. Many possible improvements aren't implemented in this patch, though I did switch simplify or remove the process of creating temporary position arrays in a few places. The design clarity that "positions are just another attribute" brings allows removing explicit copying of vertices in some procedural operations-- they are just processed like most other attributes. **Performance** This touches so many areas that it's hard to benchmark exhaustively, but I observed some areas as examples. * The mesh line node with 4 million count was 1.5x (8ms to 12ms) faster. * The Spring splash screen went from ~4.3 to ~4.5 fps. * The subdivision surface modifier/node was slightly faster RNA access through Python may be slightly slower, since now we need a name lookup instead of just a custom data type lookup for each index. **Future Improvements** * Remove uses of "vert_coords" functions: * `BKE_mesh_vert_coords_alloc` * `BKE_mesh_vert_coords_get` * `BKE_mesh_vert_coords_apply{_with_mat4}` * Remove more hidden copying of positions * General simplification now possible in many areas * Convert more code to C++ to use `float3` instead of `float[3]` * Currently `reinterpret_cast` is used for those C-API functions Differential Revision: https://developer.blender.org/D15982
2023-01-10 00:10:43 -05:00
const float(*positions_dst)[3] = BKE_mesh_vert_positions(me_dst);
const int num_verts_dst = me_dst->totvert;
const blender::Span<MEdge> edges_dst = me_dst->edges();
const blender::Span<MPoly> polys_dst = me_dst->polys();
const blender::Span<MLoop> loops_dst = me_dst->loops();
CustomData *ldata_dst = &me_dst->ldata;
MeshRemapIslandsCalc island_callback = data_transfer_get_loop_islands_generator(cddata_type);
if (!geom_map_init[LDATA]) {
const int num_loops_src = me_src->totloop;
if ((map_loop_mode == MREMAP_MODE_TOPOLOGY) && (loops_dst.size() != num_loops_src)) {
BKE_report(reports,
RPT_ERROR,
"Source and destination meshes do not have the same amount of face corners, "
"'Topology' mapping cannot be used in this case");
continue;
}
if ((map_loop_mode & MREMAP_USE_EDGE) && (me_src->totedge == 0)) {
BKE_report(reports,
RPT_ERROR,
"Source mesh doesn't have any edges, "
"None of the 'Edge' mappings can be used in this case");
continue;
}
if (ELEM(0, loops_dst.size(), num_loops_src)) {
BKE_report(
reports,
RPT_ERROR,
"Source or destination meshes do not have any faces, cannot transfer corner data");
continue;
}
BKE_mesh_remap_calc_loops_from_mesh(map_loop_mode,
space_transform,
max_distance,
ray_radius,
Refactor: Move normals out of MVert, lazy calculation As described in T91186, this commit moves mesh vertex normals into a contiguous array of float vectors in a custom data layer, how face normals are currently stored. The main interface is documented in `BKE_mesh.h`. Vertex and face normals are now calculated on-demand and cached, retrieved with an "ensure" function. Since the logical state of a mesh is now "has normals when necessary", they can be retrieved from a `const` mesh. The goal is to use on-demand calculation for all derived data, but leave room for eager calculation for performance purposes (modifier evaluation is threaded, but viewport data generation is not). **Benefits** This moves us closer to a SoA approach rather than the current AoS paradigm. Accessing a contiguous `float3` is much more efficient than retrieving data from a larger struct. The memory requirements for accessing only normals or vertex locations are smaller, and at the cost of more memory usage for just normals, they now don't have to be converted between float and short, which also simplifies code In the future, the remaining items can be removed from `MVert`, leaving only `float3`, which has similar benefits (see T93602). Removing the combination of derived and original data makes it conceptually simpler to only calculate normals when necessary. This is especially important now that we have more opportunities for temporary meshes in geometry nodes. **Performance** In addition to the theoretical future performance improvements by making `MVert == float3`, I've done some basic performance testing on this patch directly. The data is fairly rough, but it gives an idea about where things stand generally. - Mesh line primitive 4m Verts: 1.16x faster (36 -> 31 ms), showing that accessing just `MVert` is now more efficient. - Spring Splash Screen: 1.03-1.06 -> 1.06-1.11 FPS, a very slight change that at least shows there is no regression. - Sprite Fright Snail Smoosh: 3.30-3.40 -> 3.42-3.50 FPS, a small but observable speedup. - Set Position Node with Scaled Normal: 1.36x faster (53 -> 39 ms), shows that using normals in geometry nodes is faster. - Normal Calculation 1.6m Vert Cube: 1.19x faster (25 -> 21 ms), shows that calculating normals is slightly faster now. - File Size of 1.6m Vert Cube: 1.03x smaller (214.7 -> 208.4 MB), Normals are not saved in files, which can help with large meshes. As for memory usage, it may be slightly more in some cases, but I didn't observe any difference in the production files I tested. **Tests** Some modifiers and cycles test results need to be updated with this commit, for two reasons: - The subdivision surface modifier is not responsible for calculating normals anymore. In master, the modifier creates different normals than the result of the `Mesh` normal calculation, so this is a bug fix. - There are small differences in the results of some modifiers that use normals because they are not converted to and from `short` anymore. **Future improvements** - Remove `ModifierTypeInfo::dependsOnNormals`. Code in each modifier already retrieves normals if they are needed anyway. - Copy normals as part of a better CoW system for attributes. - Make more areas use lazy instead of eager normal calculation. - Remove `BKE_mesh_normals_tag_dirty` in more places since that is now the default state of a new mesh. - Possibly apply a similar change to derived face corner normals. Differential Revision: https://developer.blender.org/D12770
2022-01-13 14:37:58 -06:00
me_dst,
Mesh: Move positions to a generic attribute **Changes** As described in T93602, this patch removes all use of the `MVert` struct, replacing it with a generic named attribute with the name `"position"`, consistent with other geometry types. Variable names have been changed from `verts` to `positions`, to align with the attribute name and the more generic design (positions are not vertices, they are just an attribute stored on the point domain). This change is made possible by previous commits that moved all other data out of `MVert` to runtime data or other generic attributes. What remains is mostly a simple type change. Though, the type still shows up 859 times, so the patch is quite large. One compromise is that now `CD_MASK_BAREMESH` now contains `CD_PROP_FLOAT3`. With the general move towards generic attributes over custom data types, we are removing use of these type masks anyway. **Benefits** The most obvious benefit is reduced memory usage and the benefits that brings in memory-bound situations. `float3` is only 3 bytes, in comparison to `MVert` which was 4. When there are millions of vertices this starts to matter more. The other benefits come from using a more generic type. Instead of writing algorithms specifically for `MVert`, code can just use arrays of vectors. This will allow eliminating many temporary arrays or wrappers used to extract positions. Many possible improvements aren't implemented in this patch, though I did switch simplify or remove the process of creating temporary position arrays in a few places. The design clarity that "positions are just another attribute" brings allows removing explicit copying of vertices in some procedural operations-- they are just processed like most other attributes. **Performance** This touches so many areas that it's hard to benchmark exhaustively, but I observed some areas as examples. * The mesh line node with 4 million count was 1.5x (8ms to 12ms) faster. * The Spring splash screen went from ~4.3 to ~4.5 fps. * The subdivision surface modifier/node was slightly faster RNA access through Python may be slightly slower, since now we need a name lookup instead of just a custom data type lookup for each index. **Future Improvements** * Remove uses of "vert_coords" functions: * `BKE_mesh_vert_coords_alloc` * `BKE_mesh_vert_coords_get` * `BKE_mesh_vert_coords_apply{_with_mat4}` * Remove more hidden copying of positions * General simplification now possible in many areas * Convert more code to C++ to use `float3` instead of `float[3]` * Currently `reinterpret_cast` is used for those C-API functions Differential Revision: https://developer.blender.org/D15982
2023-01-10 00:10:43 -05:00
positions_dst,
num_verts_dst,
edges_dst.data(),
edges_dst.size(),
loops_dst.data(),
loops_dst.size(),
polys_dst.data(),
polys_dst.size(),
ldata_dst,
(me_dst->flag & ME_AUTOSMOOTH) != 0,
me_dst->smoothresh,
dirty_nors_dst,
me_src,
island_callback,
islands_handling_precision,
&geom_map[LDATA]);
geom_map_init[LDATA] = true;
}
if (mdef && vg_idx != -1 && !weights[LDATA]) {
weights[LDATA] = static_cast<float *>(
MEM_mallocN(sizeof(*weights[LDATA]) * size_t(loops_dst.size()), __func__));
BKE_defvert_extract_vgroup_to_loopweights(mdef,
vg_idx,
num_verts_dst,
loops_dst.data(),
loops_dst.size(),
invert_vgroup,
weights[LDATA]);
}
if (data_transfer_layersmapping_generate(&lay_map,
ob_src,
ob_dst,
me_src,
me_dst,
ME_LOOP,
cddata_type,
mix_mode,
mix_factor,
weights[LDATA],
loops_dst.size(),
use_create,
use_delete,
fromlayers,
tolayers,
space_transform)) {
CustomDataTransferLayerMap *lay_mapit;
changed |= (lay_map.first != nullptr);
for (lay_mapit = static_cast<CustomDataTransferLayerMap *>(lay_map.first); lay_mapit;
lay_mapit = lay_mapit->next) {
CustomData_data_transfer(&geom_map[LDATA], lay_mapit);
}
BLI_freelistN(&lay_map);
}
}
if (DT_DATATYPE_IS_POLY(dtdata_type)) {
Mesh: Move positions to a generic attribute **Changes** As described in T93602, this patch removes all use of the `MVert` struct, replacing it with a generic named attribute with the name `"position"`, consistent with other geometry types. Variable names have been changed from `verts` to `positions`, to align with the attribute name and the more generic design (positions are not vertices, they are just an attribute stored on the point domain). This change is made possible by previous commits that moved all other data out of `MVert` to runtime data or other generic attributes. What remains is mostly a simple type change. Though, the type still shows up 859 times, so the patch is quite large. One compromise is that now `CD_MASK_BAREMESH` now contains `CD_PROP_FLOAT3`. With the general move towards generic attributes over custom data types, we are removing use of these type masks anyway. **Benefits** The most obvious benefit is reduced memory usage and the benefits that brings in memory-bound situations. `float3` is only 3 bytes, in comparison to `MVert` which was 4. When there are millions of vertices this starts to matter more. The other benefits come from using a more generic type. Instead of writing algorithms specifically for `MVert`, code can just use arrays of vectors. This will allow eliminating many temporary arrays or wrappers used to extract positions. Many possible improvements aren't implemented in this patch, though I did switch simplify or remove the process of creating temporary position arrays in a few places. The design clarity that "positions are just another attribute" brings allows removing explicit copying of vertices in some procedural operations-- they are just processed like most other attributes. **Performance** This touches so many areas that it's hard to benchmark exhaustively, but I observed some areas as examples. * The mesh line node with 4 million count was 1.5x (8ms to 12ms) faster. * The Spring splash screen went from ~4.3 to ~4.5 fps. * The subdivision surface modifier/node was slightly faster RNA access through Python may be slightly slower, since now we need a name lookup instead of just a custom data type lookup for each index. **Future Improvements** * Remove uses of "vert_coords" functions: * `BKE_mesh_vert_coords_alloc` * `BKE_mesh_vert_coords_get` * `BKE_mesh_vert_coords_apply{_with_mat4}` * Remove more hidden copying of positions * General simplification now possible in many areas * Convert more code to C++ to use `float3` instead of `float[3]` * Currently `reinterpret_cast` is used for those C-API functions Differential Revision: https://developer.blender.org/D15982
2023-01-10 00:10:43 -05:00
const float(*positions_dst)[3] = BKE_mesh_vert_positions(me_dst);
const int num_verts_dst = me_dst->totvert;
const blender::Span<MPoly> polys_dst = me_dst->polys();
const blender::Span<MLoop> loops_dst = me_dst->loops();
if (!geom_map_init[PDATA]) {
const int num_polys_src = me_src->totpoly;
if ((map_poly_mode == MREMAP_MODE_TOPOLOGY) && (polys_dst.size() != num_polys_src)) {
BKE_report(reports,
RPT_ERROR,
"Source and destination meshes do not have the same amount of faces, "
"'Topology' mapping cannot be used in this case");
continue;
}
if ((map_poly_mode & MREMAP_USE_EDGE) && (me_src->totedge == 0)) {
BKE_report(reports,
RPT_ERROR,
"Source mesh doesn't have any edges, "
"None of the 'Edge' mappings can be used in this case");
continue;
}
if (ELEM(0, polys_dst.size(), num_polys_src)) {
BKE_report(
reports,
RPT_ERROR,
"Source or destination meshes do not have any faces, cannot transfer face data");
continue;
}
BKE_mesh_remap_calc_polys_from_mesh(map_poly_mode,
space_transform,
max_distance,
ray_radius,
Refactor: Move normals out of MVert, lazy calculation As described in T91186, this commit moves mesh vertex normals into a contiguous array of float vectors in a custom data layer, how face normals are currently stored. The main interface is documented in `BKE_mesh.h`. Vertex and face normals are now calculated on-demand and cached, retrieved with an "ensure" function. Since the logical state of a mesh is now "has normals when necessary", they can be retrieved from a `const` mesh. The goal is to use on-demand calculation for all derived data, but leave room for eager calculation for performance purposes (modifier evaluation is threaded, but viewport data generation is not). **Benefits** This moves us closer to a SoA approach rather than the current AoS paradigm. Accessing a contiguous `float3` is much more efficient than retrieving data from a larger struct. The memory requirements for accessing only normals or vertex locations are smaller, and at the cost of more memory usage for just normals, they now don't have to be converted between float and short, which also simplifies code In the future, the remaining items can be removed from `MVert`, leaving only `float3`, which has similar benefits (see T93602). Removing the combination of derived and original data makes it conceptually simpler to only calculate normals when necessary. This is especially important now that we have more opportunities for temporary meshes in geometry nodes. **Performance** In addition to the theoretical future performance improvements by making `MVert == float3`, I've done some basic performance testing on this patch directly. The data is fairly rough, but it gives an idea about where things stand generally. - Mesh line primitive 4m Verts: 1.16x faster (36 -> 31 ms), showing that accessing just `MVert` is now more efficient. - Spring Splash Screen: 1.03-1.06 -> 1.06-1.11 FPS, a very slight change that at least shows there is no regression. - Sprite Fright Snail Smoosh: 3.30-3.40 -> 3.42-3.50 FPS, a small but observable speedup. - Set Position Node with Scaled Normal: 1.36x faster (53 -> 39 ms), shows that using normals in geometry nodes is faster. - Normal Calculation 1.6m Vert Cube: 1.19x faster (25 -> 21 ms), shows that calculating normals is slightly faster now. - File Size of 1.6m Vert Cube: 1.03x smaller (214.7 -> 208.4 MB), Normals are not saved in files, which can help with large meshes. As for memory usage, it may be slightly more in some cases, but I didn't observe any difference in the production files I tested. **Tests** Some modifiers and cycles test results need to be updated with this commit, for two reasons: - The subdivision surface modifier is not responsible for calculating normals anymore. In master, the modifier creates different normals than the result of the `Mesh` normal calculation, so this is a bug fix. - There are small differences in the results of some modifiers that use normals because they are not converted to and from `short` anymore. **Future improvements** - Remove `ModifierTypeInfo::dependsOnNormals`. Code in each modifier already retrieves normals if they are needed anyway. - Copy normals as part of a better CoW system for attributes. - Make more areas use lazy instead of eager normal calculation. - Remove `BKE_mesh_normals_tag_dirty` in more places since that is now the default state of a new mesh. - Possibly apply a similar change to derived face corner normals. Differential Revision: https://developer.blender.org/D12770
2022-01-13 14:37:58 -06:00
me_dst,
Mesh: Move positions to a generic attribute **Changes** As described in T93602, this patch removes all use of the `MVert` struct, replacing it with a generic named attribute with the name `"position"`, consistent with other geometry types. Variable names have been changed from `verts` to `positions`, to align with the attribute name and the more generic design (positions are not vertices, they are just an attribute stored on the point domain). This change is made possible by previous commits that moved all other data out of `MVert` to runtime data or other generic attributes. What remains is mostly a simple type change. Though, the type still shows up 859 times, so the patch is quite large. One compromise is that now `CD_MASK_BAREMESH` now contains `CD_PROP_FLOAT3`. With the general move towards generic attributes over custom data types, we are removing use of these type masks anyway. **Benefits** The most obvious benefit is reduced memory usage and the benefits that brings in memory-bound situations. `float3` is only 3 bytes, in comparison to `MVert` which was 4. When there are millions of vertices this starts to matter more. The other benefits come from using a more generic type. Instead of writing algorithms specifically for `MVert`, code can just use arrays of vectors. This will allow eliminating many temporary arrays or wrappers used to extract positions. Many possible improvements aren't implemented in this patch, though I did switch simplify or remove the process of creating temporary position arrays in a few places. The design clarity that "positions are just another attribute" brings allows removing explicit copying of vertices in some procedural operations-- they are just processed like most other attributes. **Performance** This touches so many areas that it's hard to benchmark exhaustively, but I observed some areas as examples. * The mesh line node with 4 million count was 1.5x (8ms to 12ms) faster. * The Spring splash screen went from ~4.3 to ~4.5 fps. * The subdivision surface modifier/node was slightly faster RNA access through Python may be slightly slower, since now we need a name lookup instead of just a custom data type lookup for each index. **Future Improvements** * Remove uses of "vert_coords" functions: * `BKE_mesh_vert_coords_alloc` * `BKE_mesh_vert_coords_get` * `BKE_mesh_vert_coords_apply{_with_mat4}` * Remove more hidden copying of positions * General simplification now possible in many areas * Convert more code to C++ to use `float3` instead of `float[3]` * Currently `reinterpret_cast` is used for those C-API functions Differential Revision: https://developer.blender.org/D15982
2023-01-10 00:10:43 -05:00
positions_dst,
num_verts_dst,
loops_dst.data(),
polys_dst.data(),
polys_dst.size(),
me_src,
&geom_map[PDATA]);
geom_map_init[PDATA] = true;
}
if (mdef && vg_idx != -1 && !weights[PDATA]) {
weights[PDATA] = static_cast<float *>(
MEM_mallocN(sizeof(*weights[PDATA]) * size_t(polys_dst.size()), __func__));
BKE_defvert_extract_vgroup_to_polyweights(mdef,
vg_idx,
num_verts_dst,
loops_dst.data(),
loops_dst.size(),
polys_dst.data(),
polys_dst.size(),
invert_vgroup,
weights[PDATA]);
}
if (data_transfer_layersmapping_generate(&lay_map,
ob_src,
ob_dst,
me_src,
me_dst,
ME_POLY,
cddata_type,
mix_mode,
mix_factor,
weights[PDATA],
polys_dst.size(),
use_create,
use_delete,
fromlayers,
tolayers,
space_transform)) {
CustomDataTransferLayerMap *lay_mapit;
changed |= (lay_map.first != nullptr);
for (lay_mapit = static_cast<CustomDataTransferLayerMap *>(lay_map.first); lay_mapit;
lay_mapit = lay_mapit->next) {
CustomData_data_transfer(&geom_map[PDATA], lay_mapit);
}
BLI_freelistN(&lay_map);
}
}
data_transfer_dtdata_type_postprocess(ob_src, ob_dst, me_src, me_dst, dtdata_type, changed);
}
2020-09-09 16:35:20 +02:00
for (int i = 0; i < DATAMAX; i++) {
BKE_mesh_remap_free(&geom_map[i]);
MEM_SAFE_FREE(weights[i]);
}
return changed;
#undef VDATA
#undef EDATA
#undef LDATA
#undef PDATA
#undef DATAMAX
}
bool BKE_object_data_transfer_mesh(struct Depsgraph *depsgraph,
Scene *scene,
Object *ob_src,
Object *ob_dst,
const int data_types,
const bool use_create,
const int map_vert_mode,
const int map_edge_mode,
const int map_loop_mode,
const int map_poly_mode,
SpaceTransform *space_transform,
const bool auto_transform,
const float max_distance,
const float ray_radius,
const float islands_handling_precision,
const int fromlayers_select[DT_MULTILAYER_INDEX_MAX],
const int tolayers_select[DT_MULTILAYER_INDEX_MAX],
const int mix_mode,
const float mix_factor,
const char *vgroup_name,
const bool invert_vgroup,
ReportList *reports)
{
return BKE_object_data_transfer_ex(depsgraph,
scene,
ob_src,
ob_dst,
nullptr,
data_types,
use_create,
map_vert_mode,
map_edge_mode,
map_loop_mode,
map_poly_mode,
space_transform,
auto_transform,
max_distance,
ray_radius,
islands_handling_precision,
fromlayers_select,
tolayers_select,
mix_mode,
mix_factor,
vgroup_name,
invert_vgroup,
reports);
}