LTX-2

LTX-2.3 is a major upgrade. It’s a production-ready multimodal engine - designed to be built on. Here’s what’s new 🧵 1/7

@haozhangml always 🫡

Thank you LTX for building such a great model and giving to the community!!

@Evinst3in ⏩⏩⏩

@ltx_model @haoailab this is speed!!

Love seeing teams build on LTX 🙌 Great work from @haoailab

Combine LTX 2.3 distilled with pose control LoRA & audio input and you get a fast, character animation model competitive with state of the art🔥 So I built a demo for it 👇 huggingface.co/spaces/linoyts…

@Arjannnl 👏👏👏

@ltx_model What do you think about my music video made by LTX 2.3 (music is also AI generated)

LTX-2.3 is a clear upgrade from LTX-2. The improvements translate into more stable motion and better detail retention in complex scenes and cleaner audio output. See for yourself ⬇️

@0rob0rus Hi @0rob0rus, happy to help! Mind sharing more details on your workflow on our Discord? discord.com/invite/ltxplat…

@NVIDIA_AI_PC @ltx_model @UnslothAI Any recommended settings for LTX 2.3 NVFP4 in ComfyUI? Testing it now: prompt adherence and motion are better, but the quality is really poor. It seems the community is struggling to find the sweet spot—anyone found a setup that works?

Local AI on NVIDIA RTX AI PCs and DGX Spark is leveling up fast at #NVIDIAGTC. Explore recently released Nemotron open models, @ltx_model LTX-2.3 NVFP4, NemoClaw, @UnslothAI Studio & more. 👇

LTX 2.3 is so good at lipsync out of the box

LTX-2.3 First-Last Frame ⬇️ youtu.be/4QBDpcIWCAc?si…

Meet LTX-2.3, a versatile AI model that's turning heads. It's a powerful image-to-video generator, but that's just the start. It can also create videos from text, audio, or even other videos. This multi-modal flexibility is what makes it so exciting for creators.

@jefharris @gucci 🔥🔥🔥

Printed silk jacquard chemisier dress by @gucci Made with the amazing new @ltx_model 2.3. #aifashion #aivideo #aistyle #gucci gucci.com/ca/en/pr/women…

Alisson Pereira's upcoming BFS (Best Face Swap) IC LoRA for LTX enables highly coherent face swaps — even in difficult conditions like distorted and angled views.

Open weights aren’t a side bet. They’re the foundation. Read @ZeevFarbman's full piece below 👇

Lightricks keeps cooking new stuff for LTX-2.3. hotfix just dropped, new spatial upscaler-x2 ver1.1 for long video gen huggingface.co/Lightricks/LTX…

My entry for the LTX Night of the Living Dead Community Cut, a community project where creators each reimagine a scene from the original film using LTX-2. I got scene 39: just a group making a plan in a room, seemingly boring at first... Made with @ltx_model LTX-2 and @NanoBanana.

@Arjannnl Glad you're enjoying the update 😎

@ltx_model Its so much better, thank you so much!

@limtzugar 👏👏👏

Found this lost footage in attic mess. Guess it can be part of Night of the Living Dead: The Community Cut. My cat is literary shaking... @NVIDIA_AI_PC @ltx_model @banodoco

@freefly189 🙏

@ltx_model Ltx2.3 is a great upgrade,thank ltx-team excellent work

@AIWarper 🔥🔥🔥

God this new LTX just bangs for lipsync. This is even without splitting the audio stems! Very fun model to play with... considering adding my workflow to the new Comfy creator hub.