Shizukana Koe - Runway Gen48 Finalist (Made with Runway's Gen3+lipsyncing, mj6, Cartesia AI)
Some examples of what fully AI generated videos can currently achieve for dialogue scenes with better lip-syncing and voice generation tools. (clips from a previous short film I made)
Finally an Update on improved training approaches and inferences for Boring Reality Images
Flux Dev (Q4-_0) + Boreal Lora + Photoshop to Color Grading
LoRA Training progress on improving scene complexity and realism in Flux-Dev
Quick Update on the latest capabilities of AI Image generations (Make of it what you will)
American Way of Life
A Reminder of what Photo-Realistic AI Image Generation Techniques have actually been able to achieve since at least last summer.
Rethinking LoRA approaches for normal photorealistic complex scenes with just SDXL base model. (more info in the comments)
Update on the Boring Reality approach for achieving better image lighting, layout, texture, and what not.
A small case study I did on Gen-2 to show how some of the img2Video approaches and other changes introduced last year may have pushed a lot of AI video generation quality in the wrong direction. It makes Sora's progress more understandable.
Feedback on Base Model Releases
Catfish dating app pics
All I want in 2024, is this in Stable Diffusion
Progress on more complicated scenes for Photo Realism with V6. (try not to look too closely)
Some normal images. Nothing to see here. Nothing to worry about.
Some ordinary, boring, and outdated photos to scroll past and ignore
90s and early 2000s Collection - From Nostalgic to Absurd
Photo Realistic approach using Realism Engine SDXL and Depth Controlnet. (ignore the hands for now)
Early insights on the improvements and limitations of Dalle-3 photo realism compared to Stable Diffusion and Midjourney
Do not let the Aliens subreddit see this
1980s Street Photography around New York
Prompt for multiple images for faster results if the resolution requirements are slowing you down. (Additional new tips and insights for SDXL in comments)
Boring America Photorealism
Quick comparison of a Midjourney+Gen2 video pipeline now that Gen2 is released.
