LTX 2.3 - essa IA gera vídeo com áudio em português, de graça

LTX-2.3 is an open-source model with 22 billion parameters that generates video and audio together in a single inference. In this video, I test its two main capabilities: text-to-video and image-to-video, all via ComfyUI with ready-made workflows. The distilled version allows generation in a few steps, and the fused version (full + LoRA) is ready to use. We go through spatial upscaling, temporal upscaling to increase FPS, 8-bit quantization, and also use Qwen 3.5 2B as a local prompt enhancer. We tested simple prompts, anime, Pixar style, vertical video, and bilingual generation (Portuguese + English). Strengths: improved prompt adherence, camera movements, and lip sync. Weaknesses: famous characters are not generated, and very complex prompts can still fail. 📌 If you want to keep up with what's going through my head, follow me on Substack: https://progbr.substack.com/ 👉 And if you're still unfamiliar with LLMs, check out this video where I explain everything you need to know to take your first step into this world:    • O Que São LLM’s? (Large Language Models)   _______ 🔹 Want to learn how to create apps from scratch to publication with me? So check out my course: https://programadorbr.com/flutter?src... _______ 0:00 Intro 0:34 Documentation 2:49 Text to Video 4:11 Image to Video 6:26 Office Scene 7:46 Goku 9:48 Vertical Video _______ 👀 If you're curious, take a look at some of my published apps: ⏵ Teleprompter: https://apps.apple.com/app/apple-stor... ⏵ Pomodoro: https://apps.apple.com/app/apple-stor... ⏵ Infinite Tic-Tac-Toe iOS: https://apps.apple.com/app/apple-stor... Android: https://play.google.com/store/apps/de... _______ ✴️ Also follow my other channels: ⏵ Channel about the Apple universe:    / @iprogbr   ⏵ Programming channel:    / @programadorbr   ⏵ Instagram profile:   / progbr