There has been an update on the diffusers dependency and it has broken the standalone MagicQuill installation ... for now.
In the meanwhile, if you have a working version DO NOT uninstall it. Hopefully we will have this resolved soon.
NOTE - the ComfyUI MagicQuill Installation still works
2024-12-25 00:40:57 +0000 UTC View Post
In this live session, we’ll take a video, find the perfect scene for a faceswap, experiment with different settings, and output the final result together! It’s a relaxed, interactive process—perfect for learning, sharing tips, or just hanging out. Let’s make some magic happen! 🎥✨
2024-12-23 09:51:33 +0000 UTC View Post
LTX has released its 0.9.1 update and it delivers even more than expected. Not only is the speed blazing but the animations are better than ever. In this tutorial we focus on the image to video aspect of the workflow - and this is where LTX shines! Truly mind-boggling movement from still images.
Integrating perfectly with ComfyUI. welook at the fan workflow called "SaltNPepper" which gives you even finer control over your animations. The image not moving much? Add a little salt. Need so...
2024-12-23 08:36:03 +0000 UTC View Post
Create stunning marketing visuals in seconds with the new VTON Full Wardrobe Edition! We’re excited about how easy it is to create model photos for our product on social media sites! Choose between upper, lower, and full body dressing. Now offering you complete freedom to try on outfits, without limits. With a simple point-and-click interface, you can easily add clothing to any image in seconds, while the software automatically adjusts for perfect fit and posture.
2024-12-22 12:21:57 +0000 UTC View Post(This podcast was made with Google Notebook LLM from original content by: Diffusion Digest)
Be sure to support them and sign up for the digital digest at: https://diffusiondigest.beehiiv.com/
https://aibrews.substack.com/
This was made by the Viking legend Feddakalkun - Drømmer skapes i sinnet.
The video was made from a single still image (using the "SaltNPepper" workflow) .
Other installations he used:
LivePortrait - https://www.patreon.com/posts/liveportrait-w-107862591
Rope - https://www.patreon.com/posts/i...
In this livestream we're going to talk briefly about the important of getting your system setup properly. The hard truth about AI technology is that it is still in the realm of Computer Science and is not quite "magic" yet. That being said, there is some technical expertise that is required to utilize the technology successfully. In this video we'll help you cover the basics.
Free system checker can be downloaded here: 2024-12-20 07:02:15 +0000 UTC View Post
Google's Notebook LLM just got even better with a powerful new update that allows us to speak directly with our own recorded podcasts! In this video, we join a live podcast, seamlessly interjecting into the conversation to discuss how people are connecting with AI on emotional levels. After sharing our thoughts, and going back and forth with the AI personalities, the podcast continues as usual, showcasing the AI’s ability to interact in real-time and enhance the discussion.
The big ra...
2024-12-20 02:17:16 +0000 UTC View Post
Good news! This new release revolutionizes the Hunyuan Video Generator with GGUF models designed for low VRAM systems! Whether you're working with 12GB of VRAM or higher, you can now unlock creative possibilities with unprecedented ease. in this installation tutorial we cover installing comfyUI, downloading the models, adding needed nodes, and modifying the workflow.
✨ Key Features:
🎁 Low VRAM Friendly: Supports GPUs with as little as 12GB VRAM!...
This new release brings GGUF models for low VRAM systems, supporting Q3-Q8 and full BF16 precision! It's never been easier to set up, thanks to an extra model installer that lets you quickly add or remove versions until you find the perfect fit for your system.
✨ Features:
Low VRAM Friendly: Works seamlessly on GPUs with as little as low as 8GB VRAM.<...
2024-12-18 11:00:53 +0000 UTC View Post
Look at these beautiful AI-generated babies! Who would think something so lifelike could be created on just 16-24GB VRAM using Hunyuan’s FP8/BF16 models?
Each scene took just 1:30 to 5:00 minutes to process, with the video smoothly transitioning from FP8 to FP16 models as the frame count increases. It’s incredible to see how far AI-generated video has come—prepare to be amazed! 🤖✨
And you can do it too!
🔧 ComfyUI & WSL (Sage Attention) Install available her...
2024-12-17 00:05:04 +0000 UTC View Post
In this step-by-step tutorial, I show you how to install the ComfyUI version of MagicQuill: the revolutionary open-source image editing software that’s changing everything. With just a few simple steps, you’ll be ready to dive into the next evolution and ease of inpainting technology.
MagicQuill makes photo editing effortless, and getting started is just as easy! This guide walks you through the installation process, so you can unlock its cutting-edge tools and start transforming yo...
2024-12-16 22:06:05 +0000 UTC View Post
This scene was modified / faceswapped in mere minutes using our one-click Rope-Next installer.
And you can do it too! At home!
Find Rope Next here:
https://www.patreon.com/posts/introducing-rope-110571649
********************************
Discord: https://discord.gg/jRYeUbBt6P
LinkedIn: 2024-12-14 11:00:00 +0000 UTC View Post
In this first look at Google AI Studio, I just barely scratch the surface of its powerful, multimodal capabilities. This tool isn't built for chatting—it’s designed to run functions, making it a versatile resource for creators and developers.
Watch as I demonstrate how it handles audio input (talking to the model), live video responses, and even video uploading and processing. There are countless other applications I didn’t even touch on, but you can explore them for yourself. Plu...
2024-12-14 10:08:32 +0000 UTC View PostYou can now gift my Patreon membership to friends, family, and anyone else you want to invite into our space!
Your gift helps grow our community and support my work, while giving you and your friend one more thing to talk about. Plus, you don't need to deal with shipping or wrapping.
Just go to https://www.patreon.com/cognibuild/gift to give a gift. 2024-12-12 22:32:37 +0000 UTC View Post
🔥 Ready to unlock the full potential of Hunyuan? In this tutorial, I show you how to install and set up Hunyuan in WSL for seamless access to the Sage Attention mechanism, which doubles generation capability! Whereas installing Sage in Windows requires jumping through a lot of hoops, by using WSL, with a single command, you avoid the hassle of making drastic changes to your base Windows environment, ensuring a clean and simple setup without disrupting other programs.
With quantized F...
2024-12-12 21:20:20 +0000 UTC View PostThis took 1 1/2 minutes on my 4090
2024-12-12 07:09:05 +0000 UTC View Post
A simple drag-and-drop converter that automatically transforms your videos to GIFs and GIFs to videos. Perfect for content creators, Discord/Reddit users, and anyone who needs quick, media conversions without dealing with complex settings. Just drop your file and let the converter do the rest!
*Ffmpeg is required to be pre-installed
Free for all members
In this video, I compare OpenAI's Sora with Hunyuan bf16, and the results might surprise you! Despite being a quantized version of a much larger model, Hunyuan manages to come impressively close in performance. The Sora videos are directly from the OpenAi website. The Hunyuan videos were made on my local machine w/ a RTX 4090.
Hunyuan is specifically designed for smaller, single GPUs, running on just 16-24GB VRAM thanks to FP8 and FP16 optimizations. However, it’s important to note th...
2024-12-10 03:11:28 +0000 UTC View Post
Ready to dive into the BEST open-source AI video generation available? I show you step-by-step how to install and set up ComfyUI for Hunyuan, the most realistic open-source video generator yet! With quantized FP8 and FP16 models, you can create stunning videos on GPUs with just 16-24GB VRAM. Not only is the quality amazing but it can be done in minutes! This is the video generator you've been waiting for!
The full Hunyuan model typically needs 40-65gb of VRAM however with the quantized ...
2024-12-09 11:31:48 +0000 UTC View PostGuys, this is what we've been waiting for! Hunyuan is the most realistic open-source video generator yet! Originally designed for 40GB GPUs, it now features quantized FP8 and FP16 models, making it accessible for systems with just 16-24GB VRAM. These optimized models deliver lifelike, high-quality video generation in minutes. 🤖✨
The installer below must be placed just outside your ComfyUI installation. Simply double-click and it will began to install the models.
IMPO...
🎅 Santa like you’ve never seen him before! Using the Hunyuan AI video generator, I created this hilarious and lifelike clip of Santa’s big night out—all on just 16-24GB VRAM with FP8/BF16 models.
Each scene took between 1:30 and 5:00 minutes to process, depending on the number of frames. As the video progresses, the frame count increases, transitioning from FP8 to FP16 models. See the future of AI-generated videos in action! 🤖✨
ComfyUI Install available here: 2024-12-09 08:49:43 +0000 UTC View Post
https://www.youtube.com/watch?v=EP6EyeyVfus
2024-12-09 04:55:43 +0000 UTC View Post
Unlock the power of Sana in this step-by-step guide! In this tutorial, I’ll show you exactly how to set up and run the Sana framework on WSL to generate high-resolution, 4K-quality images—right on your laptop! 🚀
With Sana’s advanced text-to-image capabilities, you can create stunning visuals in seconds,, even with a 9GB GPU. We’ll walk through the entire process: from setting up your Subsystem for Linux to installation to generating images that rival the quality of massive di...
2024-12-08 22:46:40 +0000 UTC View Post
Say Hello to Sana – the text-to-image tool that brings creativity to everyone, no matter your hardware! With versions tailored to different GPUs, Sana delivers stunning visuals up to 4096×4096 resolution, faster and more efficiently than ever.
FREE FOR ALL SUPPORTING MEMBERS
🌟 Sana-1.6M: Perfect for GPUs with 12GB of VRAM, delivering high-quality images quickly and smoothly.
🌟
This audio podcast will be available every Sunday . Check in every week!
(This podcast was made with Google Notebook LLM from original content by: Diffusion Digest)
Be sure to support them and sign up for the digital digest at: https://diffusiondigest.beehiiv.com/
https://aibrews.substack.com/
Share
Generating unlimited diverse training environments for future general agents
Today we introduce Genie 2, a foundation world model capable of generating an endless variety of action-controllable, playable 3D environments for training and evaluating embodied agents. Based o...
2024-12-08 07:28:38 +0000 UTC View Post
OpenAI’s latest artificial intelligence (AI) system dropped in September with a bold promise. The company behind the chatbot ChatGPT showcased o1 — its latest suite of large language models (LLMs) — as having a “new level of AI capability”. OpenAI, which is based in San Francisco, California, claims that o1 works in a way that is 2024-12-05 04:18:09 +0000 UTC View Post
