ByteDance’s new model folds text-to-image + image editing into one engine, so you can generate and surgically edit in the same flow. Think: swap subjects, rewrite poster text while preserving font/kerning, relight scenes, or build consistent looks from multiple reference images. It’s also fast and goes up to 4K. If Nano-Banana wowed you, this one will BLOW. YOUR. MIND. 🤯 Why it’s cool: - One model for gen + edit (less tool-hopping). - Batch & multi-reference for style/character consistency. - Up to 4K output; faster than the last version. Designers: don’t fear it — master it and dominate. Now go make something and post it below. 💥🎬 Check it out HERE!

500 creators, one community. AIography is just getting started. 👀 We just hit another milestone — 500 members strong! 🙌A huge thank you to everyone who’s joined and contributed to making this such a special community. I know things have been a little quiet lately — that’s because I’ve been head-down, fully consumed with building Lumarka, our revolutionary new AI filmmaking/content creation app. 🎬🤖 But stay tuned! In the near future, I’ll be sharing some very cool things here, including exclusive sneak peeks of Lumarka before anyone else gets to see them. More soon — and thanks again for being part of this journey. 🚀

Big news for AI video creators: Wan 2.2 is here, and it’s a serious step forward. Here’s what makes it worth your time: - Dual Generation Modes – Text-to-Video, Image-to-Video, or a hybrid combo. One model does it all. - Cinematic Control – Prompts can now tap into lighting, composition, color, and real cinematic camera language. Think more “film look” out of the box. - Smoother Motion – Trained on way more data than v2.1, it handles body movement, facial expressions, and complex choreography with a lot more fluidity. - Smarter Architecture – A Mixture-of-Experts setup means the model uses different “brains” for rough vs. refined detail, so you get higher capacity without crushing your GPU. - Runs on Consumer Hardware – The lighter TI2V-5B version can crank out 720p/24fps clips on cards like the RTX 4090. - Benchmark Winner – On WAN-Bench 2.0, it outperforms even some big closed-source players like Sora and Hailuo 02. In short: better visuals, better motion, more filmmaker-friendly. I’ll be testing it out in some workflows for Lumarka soon, but I’d love to hear what you’re seeing. If you’ve run prompts through Wan 2.2 already, what’s impressed you, or what’s still frustrating? Drop your clips, tips, or thoughts below 👇

A little late to the game about the news on this but, yes, MJ now generates video. Check out this post from one of the leading AI filmmakers working today, Dave Clark.

Hey Party People, Happy Summer Solstice! Very little info available about Seedance, a new AI video gen model from China's behemoth ByteDance, (makers of TikTok). A lot of people are saying it's a monster, challenging, even surpassing, Google's Veo 3! 🤯 I mean, can this stuff be coming at us any faster?? Lot's of demos on their home page and you can check out some other info I found HERE. You can also try the model for yourself at Fal.ai If you make anything with it, or learn more details, be sure to share it with us!