feat: add ltx2.3 support (#1463)

* add GemmaTokenizer

* add basic ltx2.3 support

* change vocab file encoding

* fix ci

* fix ubuntu build

* add temporal tiling support

* add ltx audio support

* update ggml submodule url

* fix generate_video

* add i2v support

* minify bundled Gemma tokenizer vocab sources

* pass video fps into temporal rope embeddings

* fix av_ca_timestep_scale_multiplier

* add LTX2Scheduler support

* update docs

* fix ci
This commit is contained in:
leejet authored and GitHub committed 2026-05-17 16:46:20 +08:00
1 parent 3b4d26f3d9
commit 67dda3f897
44 files changed
+6393 -240

No files matched your search

+10 -3
View File
@@ -231,16 +231,21 @@ bool execute_vid_gen_job(ServerRuntime& runtime,
sd_vid_gen_params_t params = job.vid_gen.to_sd_vid_gen_params_t();
SDImageVec results;
int num_results = 0;
int num_results = 0;
sd_audio_t* generated_audio = nullptr;
{
std::lock_guard<std::mutex> lock(*runtime.sd_ctx_mutex);
sd_image_t* raw_results = generate_video(runtime.sd_ctx, &params, &num_results);
sd_image_t* raw_results = nullptr;
if (!generate_video(runtime.sd_ctx, &params, &raw_results, &num_results, &generated_audio)) {
raw_results = nullptr;
}
results.adopt(raw_results, num_results);
}
num_results = results.count();
if (num_results <= 0) {
free_sd_audio(generated_audio);
error_message = "generate_video returned no results";
return false;
}
@@ -249,7 +254,9 @@ bool execute_vid_gen_job(ServerRuntime& runtime,
results.data(),
num_results,
job.vid_gen.gen_params.fps,
job.vid_gen.output_compression);
job.vid_gen.output_compression,
generated_audio);
free_sd_audio(generated_audio);
if (video_bytes.empty()) {
error_message = "failed to encode generated video container";
return false;