LTX-2
Official Python inference and LoRA trainer package for the LTX-2 audio–video generative model.
Discover LTX-2 multi-GPU runners like TI2VidTwoStagesRunner and DistilledRunner. Learn how they use sequence parallelism, tiled data parallelism, and distributed VAE decoding to speed up video generation.
How Gradient Estimation Reduces Inference Steps in LTX-2 Video GenerationDiscover how LTX-2 uses gradient estimation to cut video generation inference steps from 100 to 2. Learn how velocity-corrected sampling maintains quality with fewer steps.
Convolutional vs Diffusion Video VAE in LTX‑2: Architecture, Speed, and Use‑Case ComparisonCompare LTX-2 video VAEs: explore ConvVAE vs DiffVAE architectures, speed, and use cases. Understand their encoder and decoder differences for optimal performance.
How to Configure Model Offloading (CPU vs Disk) for Memory-Constrained Systems in LTX-2Configure LTX-2 model offloading to CPU or disk for memory-constrained systems. Run massive diffusion models on limited VRAM by streaming weights from system RAM or NVMe storage.
LTX-2 Prompt Writing and Enhancement: Best Practices for High-Fidelity Video GenerationMaster LTX-2 prompt writing with best practices for clear, concise prompts and automatic enhancement using --enhance-prompt to achieve high-fidelity video generation.
How Duration Prediction Works to Auto-Determine Frame Count in LTX-2Discover how LTX-2's DurationHead predicts video duration in seconds, automatically setting frame count while respecting VAE temporal constraints for efficient video generation.
DiffVAE Backend Options in LTX-2: NATTEN, Triton, Eager SDPA & Blackwell DSL ExplainedExplore DiffVAE backend options in LTX-2: NATTEN, Triton, Eager SDPA, and Blackwell DSL. Understand which to use for your hardware and configuration.
How Generated Keyframes Work in LTX-2 and Their Token Cost ExplainedLearn how LTX-2 generated keyframes work by adding latent tokens for conditioning frames. Understand their token cost calculated as N / num_latent_frames.
How the Dub-It Pipeline Works for Audio Rephrasing with Lip Sync in LTX-2Discover the Dub-It pipeline in LTX-2 a two stage diffusion system for audio rephrasing with lip sync. Learn how it preserves original audio and maintains perfect sync.
How to Use IC-LoRA for Video-to-Video Transformations in LTX-2: A Complete GuideMaster IC-LoRA for video-to-video transformations in LTX-2. This guide details using ICLoraPipeline and VideoToVideoStrategy for seamless style and structure transfer.
LTX-2 Guidance Parameters: CFG, STG, and Modality Guidance ExplainedMaster LTX-2's guidance parameters: CFG, STG, and modality guidance. Tune these signals independently for advanced video and audio generation. Learn more now.
How to Apply LoRA Adapters (Distilled, IC-LoRA, Detailing) to LTX-2 PipelinesLearn to apply LoRA adapters distilled IC-LoRA and detailing to LTX-2 pipelines Train an adapter save the checkpoint and fuse it into the base model for efficient inference
Have a question about this repo?
These articles cover the highlights, but your codebase questions are specific. Give your agent direct access to the source. Share this with your agent to get started:
curl -s "https://instagit.com/install.md" 
Maintain an open-source project? Get it listed too →