WAN Video 2.2 is here—This Uncensored Model is a game changer. 🚀

Vanvideo got an upgrade to version 2.2 and it's everything we ever dreamed of. Produces higher quality video, has a better understanding of lighting, camera, and it is still 100% uncensored. The best about it is that you can run it on your local machine and that it has Laura support. By the end of the video, you will know how to generate videos from text, videos from images, how to use loras, and all this with only four steps. As a bonus tip, I'm going to show you later how you increase the length of your video even further without the video falling apart. We need Conf UI. So, go to this URL and click here on download and select the version you need. Follow the installer. It's pretty easy. If you have an Nvidia graphics card, however, and you want a real speed boost, then go to my channel and watch this video where I show how to install CompuI including Sage Attention. After you've done this, head over to my Patreon and grab the workflows on the bottom. You will also find a concise guide that maps VRAMm sizes to recommended downloads and the folder placement. There's now a high noise and a low noise model. Make sure you grab one of each. This is due to the mixture of experts architecture that van 2.2 uses. While the high noise expert takes care of the early generation steps, the low-noise expert does the finishing touches. After you downloaded and installed everything, start Comi and drag my workflow in. It should look like this. The workflow itself is pretty straightforward. We start with the model loaders here on the left side. So you can test if you put every model in the right folder by clicking this arrow here. If it says undefined, then this is wrong. Then you put your model in the wrong folder and Confi can't find it. If you followed my last video like I told you to do, then you can set this here to auto. This will speed up your generation about 30 to 50% depending on your GPU and depending on the workflow you are using. Another important part are these Laura power loaders here. So you can use them to use your own Loras. actually one 2.1 Loras are working. Um, some better than others and yeah, some even crash the workflow, but yeah, you can try for yourself.