Updated: May 11, 2026
character
This workflow is designed for cinematic omnidirectional video outpainting with LTX 2.3, using ImagePad KJ as the key frame-expansion method. Its main purpose is to take an existing video or source frame sequence, expand the visual canvas outward, and generate a wider cinematic video while preserving the original center content, motion rhythm, lighting direction, and scene continuity as much as possible.
Compared with a basic image-to-video workflow, this version is more focused on controlled video expansion. The source video is loaded through VHS_LoadVideo, and its frame information, frame rate, audio, and video metadata are passed into the LTX generation and export pipeline. The workflow then uses ImagePadKJ to add controlled padding around the source frames. In this setup, the padding values are designed to expand the frame outward, creating additional visual space for LTX 2.3 to generate new surrounding content instead of simply stretching or cropping the original video.
The main generation backbone uses LTX-2.3 22B Dev-Dare merged distilled components, LTXAV text encoding, LTX video VAE, and LTX audio VAE. The workflow also includes the LTX 2.3 distilled LoRA route and the LTX IC LoRA outpaint model, which is important for guiding the model toward outpainting behavior. Instead of treating the padded area as random empty space, the IC LoRA guide helps the model understand that the new frame area should visually extend the original shot.
The workflow uses LTXVImgToVideoConditionOnly and LTXAddVideoICLoRAGuide to connect the padded source frame to the video latent process. This gives the generated video a stronger visual anchor. The original frame remains the center of the shot, while the model expands the surrounding environment with new cinematic detail. This is useful for converting narrow footage into a wider shot, creating 16:9 or cinematic-style layouts, extending fantasy scenes, expanding character environments, or preparing stronger video covers and social media assets.
The sampling section uses SamplerCustomAdvanced with manual sigma values, Euler ancestral sampling, CFG guidance, and fixed noise control. This gives the workflow a more specialized LTX sampling structure rather than a simple default sampler. The generated video latent is then separated back into video and audio latents, cropped through LTXVCropGuides, decoded with tiled VAE decoding, and exported through VHS_VideoCombine as an H.264 MP4 file.
The workflow also includes practical finishing tools. Color Correct nodes help adjust darker scenes or improve the visual balance of the outpainted result. Switch image and ImageConcanate nodes allow comparison between original and expanded output, making it easier to inspect whether the new frame area blends naturally with the source footage.
In short, this is an LTX 2.3 cinematic video outpainting workflow built around ImagePad KJ, IC LoRA outpaint guidance, video-frame conditioning, and final MP4 export. If you want to see how ImagePad KJ expands the source frames, how LTX 2.3 fills the new visual area, and how the final cinematic outpainted video is produced, watch the full tutorial from the YouTube link above.
⚙️ Try the Workflow Online
👉 Workflow: https://www.runninghub.ai/post/2053497184217247746/?inviteCode=rh-v1111
Open the link above to run the workflow directly online and view the generation results in real time.
If the results meet your expectations, you can also deploy it locally for further customization.
🎁 Fan Benefits: Register now to get 1000 points, plus 100 daily login points — enjoy 4090-level performance and 48 GB of powerful compute!
📺 Bilibili Updates (Mainland China & Asia-Pacific)
If you are in Mainland China or the Asia-Pacific region, you can watch the video below for workflow demos and a detailed creative breakdown.
📺 Bilibili Video: https://www.bilibili.com/video/BV1Qa5s6tESM/
I will continue updating model resources on Quark Drive:
👉 https://pan.quark.cn/s/20c6f6f8d87b
These resources are mainly prepared for local users, making creation and learning more convenient.
⚙️ 在线体验工作流
👉 工作流： https://www.runninghub.ai/post/2053497184217247746/?inviteCode=rh-v1111
打开上方链接即可直接运行该工作流，实时查看生成效果。
如果觉得效果理想，你也可以在本地进行自定义部署。
🎁 粉丝福利： 注册即送 1000 积分，每日登录 100 积分，畅玩 4090 体验 48 G 超级性能！
📺 Bilibili 更新（中国大陆及南亚太地区）
如果你在中国大陆或南亚太地区，可以通过下方视频查看该工作流的实测效果与构思讲解。
📺 B站视频： https://www.bilibili.com/video/BV1Qa5s6tESM/
我会在 夸克网盘 持续更新模型资源：
👉 https://pan.quark.cn/s/20c6f6f8d87b
这些资源主要面向本地用户，方便进行创作与学习。