This is a text-to-video workflow using the Smooth Mix Wan2.2_T2V_14B model. It performs really well with both high-noise motion and low-noise details. I've tested and modified the original author's workflow to get even better dynamic performance. Another important point is that it supports NSFW video generation.For important details, please see the testing and usage instructions below.💻I've already set up the complete ➡️ <a target="_blank" rel="ugc" href="https://www.runninghub.ai/post/1980191945823784961/?inviteCode=rh-v1182">workflow</a> for you. This will let you quickly try it out online and avoid the hassle of not having enough VRAM or dealing with local plugin conflicts.🎁Bonus: If you're signing up for the first time, you can get 1,000 free RH Coins by using my link and the invite code ➡️ rh-v1182. Plus, you'll get another 100 RH Coins for logging in daily.🚀Workflow Testing and Usage Instructions:1. First, a big thank you to DigitalPastel for releasing this model. You can find their Smooth Mix model at this <a target="_blank" rel="ugc" href="https://civarchive.com/models/1995784/smooth-mix-wan-22-i2vt2v-14b">link</a>.2. This is an FP8 quantized version of the model, so when you're using KJ's model loader node, you'll need to set the quantization to the fp8 option.3. This V2.0 model is a big improvement over V1.0 in terms of motion. Compared to DigitalPastel's original workflow, mine uses a 6+4 sampling setup: 6 steps for the high-noise pass and 4 for the low-noise pass. For the low-noise model, you could also try using the official Wan low-noise model. In my tests, that combo gives you richer backgrounds, but the character details might not be as good as with the Smooth Mix low-noise model.4. For other operational details, please see the notes inside the workflow.

Files with SHA256: e298da765127a1a7fc7519ca0a71cc4dd78cdf04678491f02b31d237c8f7901c