### 1. Workflow Function

Generate long videos using the LTX 0.9.1 model, with generation continuing from the last frame of the video.

### 2. Workflow Details

a. Since the LTX model is not particularly friendly to certain resolutions, the workflow includes using the fluxfill model for expanding the image.
b. A slight blur on the background has been added to enhance the perception of motion, making it more natural and pleasant for the subject in motion.
c. Testing revealed that adding a slight Gaussian blur and a light motion blur allows for better movement of the subject and smoother camera motion.
d. After incorporating the image expansion and background blurring, the overall motion effect has significantly improved.
e. The workflow now includes loop control, which uses the last frame of each video as input to generate the next segment of the video in subsequent loops.

### 3. Usage Instructions

a. Upload an image and click on the run queue to generate a video approximately 5 seconds long.
b. To adjust the resolution, refer to the preset values in the **LTXV Model Configurator** node; ensure that the width and height settings are consistent with this node.
c. To create longer videos, increase the number of loop iterations by changing the number 3 in the **For Loop Start** node to a higher value.
d. If you follow me, you can also access some of my creations. I've just made a page using V0, available at [https://app.facemimic-ai.com](https://app.facemimic-ai.com). Feel free to try it out. 


1. 工作流功能

通过LTX 0.9.1 模型生成长视频，通过最后一帧延续的方式进行生成。

2. 工作流内容

a. 由于LTX模型对一些分辨率不是特别友好，所以添加了使用fluxfill模型进行了扩图。
b. 由于背景的适当模糊，可以让主体产生更强的运动，更友好的运动，所以添加了背景的轻微模糊处理。
c. 测试发现添加轻微的高斯模糊，让主体更好的运动，轻微的motion blur可以让镜头更好的运动。
d. 添加了扩图和背景模糊后，整体的运动效果更好。
f. 工作流添加了循环控制，每次循环使用视频的最后一帧当中输入，继续生成下一个视频。

3. 使用说明

a. 输入一张图片，直接点击运行队列。会产生一个5秒左右的视频。
b. 如果想调节分辨率，请参考🅛🅣🅧 LTXV Model Configurator这个节点的preset值，分辨率的宽和高和这个节点设置要一致。
c. 如果要生成更长的视频，请把循环次数调大即可，在For Loop Start节点把数字3直接改大即可。
d. 如果你关注我，你也可以访问我制作的一些东西，我刚用V0制作了个页面，https://app.facemimic-ai.com，欢迎您尝试使用。