Fabio
c36510c21e
Some checks failed
Linux Build / Linux cpu (push) Has been cancelled
Linux Build / Linux cpu-clang (push) Has been cancelled
Linux Build / Linux vulkan (push) Has been cancelled
mac-build / macOS CPU (push) Has been cancelled
Nix Build / Nix (cpu) (push) Has been cancelled
Nix Build / Nix (vulkan) (push) Has been cancelled
windows-build / Windows CPU (push) Has been cancelled
71 lines
2.2 KiB
C++
71 lines
2.2 KiB
C++
#pragma once
|
|
|
|
#include "engine/models/dramabox/types.h"
|
|
|
|
#include <cstdint>
|
|
#include <string>
|
|
#include <vector>
|
|
|
|
namespace engine::models::dramabox {
|
|
|
|
struct DramaBoxEncodedReferenceLatents;
|
|
|
|
struct DramaBoxLatentShape {
|
|
int64_t batch = 1;
|
|
int64_t channels = 8;
|
|
int64_t frames = 0;
|
|
int64_t mel_bins = 16;
|
|
|
|
int64_t token_count() const noexcept;
|
|
int64_t latent_value_count() const noexcept;
|
|
};
|
|
|
|
struct DramaBoxLatentState {
|
|
DramaBoxLatentShape target_shape;
|
|
int64_t tokens = 0;
|
|
std::vector<float> latent;
|
|
std::vector<float> clean_latent;
|
|
std::vector<float> denoise_mask;
|
|
std::vector<float> positions;
|
|
};
|
|
|
|
struct DramaBoxPromptChunk {
|
|
std::string text;
|
|
double estimated_duration_seconds = 0.0;
|
|
};
|
|
|
|
double estimate_dramabox_speech_duration_seconds(const std::string & prompt);
|
|
double estimate_dramabox_duration_seconds(const std::string & prompt, float duration_scale);
|
|
std::vector<DramaBoxPromptChunk> chunk_prompt_for_duration(
|
|
const std::string & prompt,
|
|
float max_duration_seconds,
|
|
float target_duration_seconds,
|
|
float duration_scale);
|
|
int64_t dramabox_aligned_pixel_frames(double duration_seconds, float fps);
|
|
DramaBoxLatentShape dramabox_target_latent_shape(double duration_seconds, const DramaBoxConfig & config);
|
|
DramaBoxLatentState create_dramabox_initial_state(const DramaBoxLatentShape & shape, const DramaBoxConfig & config);
|
|
std::vector<float> make_dramabox_ltx2_sigmas(int64_t steps, int64_t latent_tokens);
|
|
float auto_rescale_for_cfg(float cfg);
|
|
void append_reference_latents(
|
|
DramaBoxLatentState & state,
|
|
const DramaBoxEncodedReferenceLatents & ref,
|
|
const DramaBoxConfig & config);
|
|
void guided_prediction_from_velocity(
|
|
const std::vector<float> & velocity,
|
|
const DramaBoxLatentState & state,
|
|
int64_t branch_count,
|
|
float sigma,
|
|
float cfg_scale,
|
|
float spatio_temporal_guidance_scale,
|
|
float guidance_rescale,
|
|
bool cfg_enabled,
|
|
bool stg_enabled,
|
|
std::vector<float> & cond,
|
|
std::vector<float> & pred);
|
|
void post_process_and_euler_step(
|
|
DramaBoxLatentState & state,
|
|
std::vector<float> & denoised,
|
|
float sigma,
|
|
float sigma_next);
|
|
|
|
} // namespace engine::models::dramabox
|