LTX 2.3 - The New KING Of UNCENSORED AI VIDEO

Come here, big boy. I need you to help me relieve stress. The best not safe for work open source video model got a huge upgrade. Too late. It promises improved audio. Singing in French. Better prompt following and more consistency in image to video. In this video, I will show you how to generate videos from text, videos from images, and how to lip sync your audio files to the videos we create. What we need here is ComfyUI. If you have no idea what it is or how to install it, please watch this video here first, but don't forg but don't forget to come back afterwards. On my Patreon, you will find a free guide on what to download and in which folder to put it. Grab the workflows while you are there. Drag and drop workflow number one in ComfyUI and let's see what we got here. Let's begin here in the model section. These are all the model loaders you downloaded because we're using a couple of different models. The Jigguf model loader is the video model itself. It's the biggest and it should do all the heavy lifting for you. So, make sure you select the model you actually downloaded in this test scenario. I downloaded the Q3, it seems, and I click it. So, otherwise, you will run into some errors if you don't do this. If you click on an arrow here and it says undefined suddenly, then you put it in the wrong folder and you have to check back. So, if you put your models into the folders while ComfyUI was running, you have to press the R key on your keyboard. So, now here you can see node definitions updated. I press it again. Yeah, here you can see, and then it should appear if you put it in the right folder. So, the dual clip or text encoder is for transforming your video prompt into a series of numbers that the model here can understand and use