Skip to content
Video

LTX-2.5 Stage 1 Distill 2-Step LoRA

видео со звуком за два шага

2 min readupdated

Это LoRA для дистиллированной LTX-2.5 от lvladikov, которая сокращает первую стадию рендера с 8 шагов до 2 — сразу для картинки и звука, в text-to-video и image-to-video. Ранний кандидат на релиз (чекпоинт 6000), обучение продолжается; автор предупреждает, что это экспериментальная версия для быстрых превью, а не для продакшн-видео.

Ключевые особенности

  • Сокращает stage 1 с 8 шагов до 2 при strength 1.0.
  • Работает для text-to-video и image-to-video, покрывает изображение, звук и cross-attention между ними.
  • В комплекте ComfyUI workflows для локального запуска и Comfy Cloud; автор сочетает LoRA с one-step refiner и ускоренным VAE-декодером.
  • Обучена на одном NVIDIA H200, rank 64, метод DMD2; teacher — LTX-2.5 distilled на стандартных 8 шагах stage 1.
  • Только для stage 1: не применяйте LoRA к refiner. Обучена на 576x320 и 320x576, после refine — 1152x640 и 640x1152, длина 5 секунд.

Пример

Пример запуска в ComfyUI: LTX-2.5 distilled + эта LoRA (strength 1.0), 2 steps, sigmas 1.0, 0.909375, 0.0, Euler Ancestral, CFG 1, half size. Затем one-step refiner SoL-Refiner-LTX-2.3-One-Step: x2 latent upsample, AdaIN к stage 1, 1 step, sigmas 0.725, 0.0, CFG 1.

ПодробнееОбучение: DMD2, один NVIDIA H200, rank 64; teacher — LTX-2.5 distilled на 8 шагах stage 1.
Ограничения: экспериментальный чекпоинт 6000, обучение продолжается; LoRA только для stage 1, не для refiner.
Тренировочные разрешения stage 1: 576x320 и 320x576, после refine — 1152x640 и 640x1152, длина 5 секунд.
Лицензия: LTX-2.x Community License Agreement; для SoL-Refiner лицензия не заявлена, проверяйте условия LTX-2.x.
Discuss Telegram 5 27

Will it run?

Only with RAM offload

Even 4-bit is larger than your VRAM. Some layers can stay in system RAM (llama.cpp, ComfyUI), but it will be several times slower.

  • BF16weights 75 GB · ready files89 GB
  • FP8 / INT8recommendedweights 38 GB47 GB
  • 4-bitweights 22 GB29 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: Won't fit any home GPU — renting is cheaper.

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

This is a LoRA: counted together with its base model LTX-2.5.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # lvladikov/LTX-2.5-Stage1-Distill-2step-LoRA
    # ===== Tokens: paste yours between the quotes =====
    export HF_TOKEN=""        # huggingface.co/settings/tokens, for gated models
    # ==========================================
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download Lightricks/LTX-2.5 --local-dir ~/models/LTX-2.5
    hf download lvladikov/LTX-2.5-Stage1-Distill-2step-LoRA --local-dir ~/models/LTX-2.5-Stage1-Distill-2step-LoRA
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to LTX-2.5
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • SageAttention 3v2.2.0from October 28, 2025

    even faster on RTX 50 / Blackwell

    FP4 attention for Blackwell. Faster than SageAttention 2, but the authors suggest v2 when accuracy matters.

    How to enable: Build from the sageattention3_blackwell folder of the repo; in ComfyUI via KJNodes.

    Earlier generations (2)
    • SageAttention 2++ — Quantised (INT8/FP8) attention with no visible quality loss. The main speed-up for image and video generation.
    • SageAttention 1 — First version: INT8 for Q·K. Still works, but 2++ is faster and more accurate.
  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (3)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
    • xFormers — The older way to speed up attention, works almost everywhere. Sage or Flash are faster where available.
  • cache-dit (DBCache)v1.5.6from November 20, 2026new release

    1.5–3× at nearly the same quality

    Reuses block outputs between diffusion steps. Supports most DiT models.

    How to enable: pip install cache-dit; one line cache_dit.enable_cache(pipe) for diffusers.

    Earlier generations (2)
    • MagCache — Skips steps by magnitude of change. More accurate than TeaCache, no calibration.
    • TeaCache — The first popular diffusion step cache. Works, newer methods are more accurate.
  • LightX2V step distill0.6.1from November 9, 2026new release

    5–10× fewer steps

    Distilled LoRAs and an engine: video in 4–8 steps instead of 30–50.

    How to enable: Load their step-distill LoRA, CFG 1.0, 4–8 steps.

  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

Model family

Requires the base model

Lightricks/LTX-2.5

This model is a LoRA / adapter. Download the base model first, then load this one on top.

New posts about versions of LTX-2.5 (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 12All on Hugging Face
Fine-tunes 12All on Hugging Face
LoRAs and adapters 12All on Hugging Face
Merges 3All on Hugging Face

About the model

lvladikov/LTX-2.5-Stage1-Distill-2step-LoRA
Author
lvladikov
Category
Video
Released
Posted

On NeyroLearn

Hugging Face

275

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments