Emil
22012c1a80
Native and manual checks / native (windows-2025) (push) Waiting to run
Native and manual checks / native (ubuntu-24.04) (push) Failing after 36s
Native and manual checks / manual (push) Successful in 28s
Windows editor and software Vulkan / windows-graphics (push) Canceled after 0s
95 lines
4.1 KiB
C++
95 lines
4.1 KiB
C++
#pragma once
|
|
#include <faset/render/renderer.hpp>
|
|
#include <array>
|
|
#include <cstddef>
|
|
#include <cstdint>
|
|
#include <span>
|
|
#include <string>
|
|
#include <string_view>
|
|
#include <unordered_map>
|
|
#include <unordered_set>
|
|
#include <vector>
|
|
|
|
namespace faset::render {
|
|
struct Bounds {
|
|
Vec3 min{};
|
|
Vec3 max{};
|
|
};
|
|
|
|
// Mesh bounds are computed from all vertices; world bounds transform all eight
|
|
// corners of the local AABB, including rotation, reflection and nonuniform scale.
|
|
// Empty or nonfinite geometry/matrices are rejected instead of being culled.
|
|
Bounds local_bounds(const Mesh& mesh);
|
|
Bounds transformed_bounds(const Bounds& local, const Mat4& model);
|
|
Bounds transformed_bounds(const Mesh& mesh, const Mat4& model);
|
|
|
|
// Resolve a requested mode against the current device and target capabilities.
|
|
// Missing HZB retains GPU frustum culling, but callers must report that fallback.
|
|
VisibilityMode select_effective_visibility_mode(VisibilityMode requested,
|
|
bool gpu_available,
|
|
bool hzb_available) noexcept;
|
|
|
|
struct InstanceUpdate {
|
|
std::uint32_t slot{};
|
|
std::uint64_t generation{};
|
|
Mat4 previous_model{identity};
|
|
Bounds previous_bounds{};
|
|
bool previous_valid{};
|
|
};
|
|
|
|
// GPU instance metadata: history valid, stable slot, then generation low/high.
|
|
// A zero generation identifies an anonymous, untracked draw.
|
|
constexpr std::array<std::uint32_t, 4>
|
|
gpu_instance_metadata(const InstanceUpdate& update, bool history_compatible) noexcept {
|
|
return {update.previous_valid && history_compatible ? 1U : 0U, update.slot,
|
|
static_cast<std::uint32_t>(update.generation),
|
|
static_cast<std::uint32_t>(update.generation >> 32)};
|
|
}
|
|
|
|
// One update per logical instance per rendered frame. finish_frame() promotes the
|
|
// current state to *rendered* history and retires keys absent from that frame.
|
|
// Reordering the input never renumbers live instances. Mesh changes retain the
|
|
// slot but advance its generation; slot reuse also advances its generation.
|
|
class InstanceTracker {
|
|
public:
|
|
// Prefer this overload for engine-owned meshes. Retaining the owner prevents
|
|
// allocator address reuse from impersonating the previously rendered mesh.
|
|
InstanceUpdate update(std::string_view key, std::shared_ptr<const Mesh> mesh_identity,
|
|
const Mat4& model, const Bounds& world_bounds,
|
|
std::string_view view_id);
|
|
// Non-owning overload for meshes whose lifetime is guaranteed by the caller.
|
|
InstanceUpdate update(std::string_view key, const Mesh* mesh_identity, const Mat4& model,
|
|
const Bounds& world_bounds, std::string_view view_id);
|
|
void finish_frame();
|
|
void invalidate_view(std::string_view view_id);
|
|
|
|
private:
|
|
struct Record {
|
|
std::uint32_t slot{};
|
|
std::uint64_t generation{};
|
|
const Mesh* mesh{};
|
|
std::shared_ptr<const Mesh> mesh_owner{};
|
|
Mat4 current_model{identity}, previous_model{identity};
|
|
Bounds current_bounds{}, previous_bounds{};
|
|
std::string current_view, previous_view;
|
|
std::uint64_t seen_frame{};
|
|
bool committed{};
|
|
};
|
|
std::unordered_map<std::string, Record> records_;
|
|
std::vector<std::uint32_t> free_slots_;
|
|
std::vector<std::uint64_t> slot_generations_;
|
|
std::unordered_set<std::string> invalid_views_;
|
|
std::uint64_t frame_{1};
|
|
};
|
|
|
|
// thresholds[i] is the projected-pixel boundary between levels i and i+1,
|
|
// ordered strictly high-to-low. Moving toward a coarser level requires size <
|
|
// threshold*(1-hysteresis); toward a finer level requires size >
|
|
// threshold*(1+hysteresis). SIZE_MAX means no prior level and uses raw thresholds.
|
|
// available is optional; when given it has thresholds.size()+1 entries, with 0
|
|
// marking missing prepared geometry. Nearest available wins; ties prefer finer.
|
|
std::size_t select_lod(float projected_pixels, std::size_t previous_level,
|
|
std::span<const float> thresholds, float hysteresis,
|
|
std::span<const std::uint8_t> available = {});
|
|
} // namespace faset::render
|