Click here to try online first:

Workflow: LTX2.3 - Fully Automated Prompt - Image-to-Video - Modular Optimization Edition  

Experience link: https://www.runninghub.ai/post/2031218459471777794/?inviteCode=rh-v1401

Workflow: Universal Image-Pro - Text-to-Image - Single, Double, Triple, Quadruple Images - Image Editing  

Experience link: https://www.runninghub.ai/post/2026244873988345857/?inviteCode=rh-v1401

Workflow: Fully Automated Prompt - LTX2.3 - First-and-Last-Frame Video Generation  

Experience link: https://www.runninghub.ai/post/2035325472770367490/?inviteCode=rh-v1401

Workflow: LTX2.3 - Fully Automated Prompt - Text-to-Video  

Experience link: https://www.runninghub.ai/post/2031218445026594817/?inviteCode=rh-v1401

Workflow: LTX2.3 Creation Assistant - Automated Prompt Collection - Text, Image, First-and-Last-Frame Video Generation  

Experience link: https://www.runninghub.ai/post/2033139139712917506/?inviteCode=rh-v1401

Workflow: IndexTTS - Dual Voice Cloning - 8 Emotion Controls  

Experience link: https://www.runninghub.ai/post/2036366094109249538/?inviteCode=rh-v1401

Workflow: Multi-Voice Cloning - FishAudio-S2 TTS - Multi-Language - Supports up to 10 Voices  

Experience link: https://www.runninghub.ai/post/2033180796315508737/?inviteCode=rh-v1401

本期带来 LTX2.3 图生视频完整工作流，五个核心设置让你快速上手：上传图片、简洁提示词（自动扩写）、25帧、时长按秒设置、尺寸自适应。提示词前置词决定模型理解力，CFG推荐4–7，步数20–40，二次采样保持4步即可。

亮点：✅ 去字幕用开源神器 VSR，90秒搞定20秒视频 ✅ 放大用 Flash VSR 直达4K ✅ 模型演技自然，细节丰富，中文准确率超90%。

痛点：音频一致性仍是挑战，现有 TTS 难有真实环境感，保留原声更沉浸。从剧本、分镜底图到抽卡剪辑，五步走通全流程，助你高效产出电影感 AI 短片。

English Introduction

This eposode  presents a complete LTX2.3 image-to-video workflow. Five core settings get you started quickly: upload an image, write a concise prompt (auto-expansion supported), set 25 fps, define duration in seconds, and adjust resolution as needed. The pre‑prompt is crucial for model comprehension. Recommended CFG: 4–7, steps: 20–40; keep the second sampler at 4 steps to save time.

Highlights:

✅ Open‑source VSR removes subtitles in ~90s for a 20s video.

✅ Flash VSR upscales to 4K.

✅ Natural acting, rich details, 90%+ Chinese accuracy.

Challenge: Audio consistency remains difficult—existing TTS tools lack real‑world ambiance, so keeping original audio often yields better immersion.

From script and storyboard to image generation, video sampling, editing, and upscaling, the five‑step pipeline helps you efficiently create cinematic AI shorts.