Interesting find of the week: Sougwen Chung, a Chinese-Canadian artist pioneering human-machine collaboration in art. Her "Assembly Lines" project features robotic art assistants that sync with her brainwaves to create paintings together.
Flux updates:
GPU compatibility: Successful generation on AMD GPU (RX 6600 XT), overcoming compatibility issues using Zluda.
CFG improvements: Support for negative prompting and values >1 without image degradation, based on PuLID team's work.
Consistent character frames: Technique using Flux and ControlNet for generating multiple consistent frames.
LoRA and DoRA training: Insights on training models using OneTrainer with Flux.1 architecture, including detailed configuration settings.
ComfyUI Flux pipeline: Clean and organized workflow for Stable Diffusion image generation using Flux.
Seamless outpainting: New workflow for precise background and human feature outpainting using Flux models in ComfyUI.
Lions Gate x Runway: Lionsgate partners with AI firm Runway to develop exclusive AI models based on its film and TV library, focusing on integrating AI into pre- and post-production workflows.
EA x AI: Electronic Arts positions AI as core to its business strategy, with over 100 AI projects in development across efficiency, expansion, and transformation areas.
Put This On Your Radar:
Tripo 3D (Version 2.0): Text-to-3D model generation tool releases version 2.0 with significantly improved mesh quality.
CogStudio: Advanced web interface for AI video generation based on the CogVideo model.
OmniGen: New unified multimodal AI model combining text and image generation capabilities.
Differential diffusion technique for AnimateDiff: Technique for creating more stable backgrounds in AI-generated videos.
Pony and non-pony AI model merging technique: New method for merging specialized AI models to expand capabilities.
Image and sound generation workflow: Workflow for generating both images and corresponding sound effects from a single prompt using Stable Diffusion and Stable Audio.
CogVideoX-5B: Open-source image-to-video model weights released for generating short video clips from input images.
CogVideoX-Fun: Open-source text/image/video-to-video model by Alibaba PAI with enhanced video generation capabilities.
ComfyUI workflow for replacing video backgrounds with Flux model: Workflow demonstrating how to replace backgrounds in videos using the Flux model.
Multi-face swap workflow for ComfyUI: Workflow for swapping multiple faces in a single image with customizable options.
Audio reactive particle simulator in ComfyUI: Workflow demonstrating an audio-reactive particle simulation system for creating visually dynamic content.
KLING 1.5: Update to KLING with motion control and general improvements.
Flux LoRA showcase: New FLUX LoRA models including Miniature People, Omegle Webcam, Gesture Drawing, Jigsaw, and SameFace Fix.
📰 Full newsletter with relevant links, context, and visuals.