Edit 31-08-2025: 🚀 Boost Your ComfyUI Workflow by up to 50%! One Click Triton - Sage Attention Install Script🚀 (Patreon)

Hello my incredible Patrons, I didn't even realize how much I needed to share this with you until I found myself suddenly inspired, typing a

Edit 31 August 2025: Friends, I’ve rebuilt everything from scratch for you! 🎉
With the recent CUDA + Sage changes, the old scripts were no longer reliable—so I created brand-new install scripts that run smoothly and silently. They’ll even leave a handy comfy_install.log on your disk so we can easily track down any issues.

👉 If something goes wrong, just send me the log in a single message and I’ll help you out.
💡 NVIDIA 50XX users: remember to run the InstallComfyUI script first, then the Nvidia50XXUpdate right after.

Thank you all for your support—you inspire me to keep making these tools better every day! 💖

Original Post:---------------------------------------------------------------------------------->

Hello my incredible Patrons,

I didn't even realize how much I needed to share this with you until I found myself suddenly inspired, typing away at lightning speed!

🔥 WAN 2.2 has dropped, and let me tell you—this is a total game-changer for Local AI Video Generation! It's insanely consistent, handles camera movements beautifully, and significantly enhances the overall output quality.

But here's the catch: it's painfully slow—even on my absolute beast of an RTX 5090. Naturally, I couldn't just let this stand, so I've been tirelessly tweaking and experimenting these past few days. The result? A major breakthrough! 🚨

Introducing my ComfyUI Install Script featuring Sage Attention & Triton 🚀. With this setup, you'll witness generation speeds boosted by an impressive 20–50%, tailored specifically to your GPU and workflow! It will install a new version of ComfyUI Manual on you machine.
See the requirements in the PDF.

✨ Next Video Update: ✨

I know I promised a deep dive into Local LoRA Training and Flux Concept Training. And trust me, the script is shaping up nicely. But honestly, I've hit a bit of a creative wall. LoRA Training is highly specialized, and when I teach you something, it's crucial for me to nail every detail perfectly—without burning myself out.

So, I've decided to first bring you a dedicated, in-depth video on WAN 2.2. It's straightforward, powerful, and exactly what we need right now to shake things up. For those who voted and were excited about LoRA training, I promise—I'm going to make it worth your wait. But first, I need a short breather from epochs, batch sizes, and training steps to recharge my creative batteries.

Thank you so much for your patience and ongoing support. You're amazing, and I'm committed to delivering content you'll absolutely love.

Stay tuned, stay excited, and let's push Local AI Video Generation to the next level together!

Warmest regards,

Chris

P.s: For the installer to work you need to have 2 things installed:

• Python 3.12 – Required for running ComfyUI and installing Sage Attention. We use Python 3.12 specifically to ensure compatibility with PyTorch 2.8 and the latest ComfyUI builds.

• Git – Needed to download ComfyUI and related repositories directly from GitHub using the git clone command.