So excited to share that I’m now part of the @FAL team that I’ve admired for a long time! If you have questions about video or image models, aren’t sure which one would work best for you, or for anything else along those lines, my DMs are always open!

Made this with Opus 5.5 in a single prompt. I think we're getting surprisingly close to the kind of quality you'd expect from professional work.

It's live 🎉 Type a topic and Storycast makes a narrated explainer video, like the ones you watch on YouTube. Opus 5.5 writes and edits, H3 Max animates. What I love most is how Opus 5.5 writes. The stories have real warmth, the facts are right, and every scene flows into the next one. It never feels like slop. I ran around 250 generations and spent about $1.8K to get the pipeline right, so you don't have to. It's all open source. website: storycast-fawn.vercel.app repo: github.com/ilkerzg/storycast

playing around with @ailker WIP H3 Max spatial reframing tool; loving the unintentional artifacts that create these painterly textures

3D-to-video is live! I’ve shared a few examples of turning Blender previz into photoreal footage with H3 Max before. It’s really good at this, and now you can try it directly on fal

I really love how Opus 5.5 writes and edits a story. I made all of these in a very short time, and they feel like the characters from the storybooks I read as a kid. Website and github repo coming today, if all goes well.

WOW this was all from just one prompt. “Use fal to create a 2.30 min video for me.”

Opus 5.5 made this video entirely on its own. I’m really impressed by its vision capabilities, and I already have so many ideas for combining them with fal (especially H3 Max) Stay tuned.

Generative media is blowing up. That’s the gist of Vol.2 of @fal’s State of Generative Media report, which just came out today. So much to dig into here ahead of their Generative Media Conference later this week, here’s a few findings that stood out to me: - The AI-native companies drove the first wave, but enterprise is now the fastest-growing segment on fal (grew more than 4x in 6 months) - 3D and spatial is the fastest growing modality at 163%. Images and video improved quickly, but 3D was hard to crack as it’s very complex, compute-heavy and unforgiving. By generating the underlying material data (the full set of PBR maps a renderer actually needs) rather than a flat picture of a surface, there have been massive improvements here (fal’s homegrown Patina model is a good example). - Embedded generation is also growing faster than standalone generation. Most people will never bother picking a model on their own. Object removal, virtual try on, extra product angles etc, will all be generated running inside tools people already use like Canva and Figma. - It’s more about what combo of small, fast, task-trained models you can chain together, vs. one single greatest model to rule them all. (@ailker has a good metaphor here: You wouldn’t expect a single person to have a Michelin star, an Oscar, and a World Cup trophy.) - As @OdinLovis writes, that is why there are around 1400 models on fal. Even the far less powerful ones can do something no other model can. They integrate roughly a model a day (!!)

So much is happening in generative media that it’s hard to keep up. We put together a report to help make sense of it all. Check it out!

H3 MAX is basically a Swiss Army knife. Been testing spatial reframing with it, and even this early, the results are looking really promising.

A year ago, LLMs could barely make a 3D sword. Now you can make that game you’ve had in your head with GPT-6-Astra and fal. I made this game in a very short time. If you want to play it or fork it and build on it: Play: cabsolutely.vercel.app/ Code: github.com/ilkerzg/cabsolute…

fal Marketing is HIRING! Since I joined a year ago, we’ve 4x’d the team, grown to 2.5M+ users, and shipped more products and launches than I can count. If you care deeply about where media and creativity are going, this is the place for you. We’re a scrappy team that’s constantly experimenting, iterating, and building. We’re hiring for: - Growth marketers - Solutions marketers - Community growth - Creative producers - Creative technologists And if you don’t fit one of the roles above reach out anyway. If you’re excited about what we’re doing we’ll find a place for you.

Who in generative media (image/video/audio/3D/robotics) fields needs a shit ton of hoppers right now? not in a few months/weeks etc. literally now. Hit me up in DMs.

another one LOOK AT THIS CONSISTENCY!

I think H3 MAX can absolutely work as a rendering engine with the right prompting. I’ll share a post today showing how.

I might’ve found a way to make yesterday’s Blender to Real video ~5,000x faster (and better imo) lol. I put the scene together with GPT-6 Astra and PATINA in about 30 minutes. But it had around 7 million polygons, so rendering took a while. I split it into chunks across 8 GPUs and rendered it in about 30min. My estimate for an M4 Pro was around 40 hours. Obviously, I could’ve optimized the settings, but that’s for the version I posted yesterday. Since then, I’ve been trying different prompts to get H3 Max to follow the Blender animation while making the buildings and materials actually look real. I think I finally got it. The 30s video took about 30s to generate with H3 Max. Compared to that estimated 41h render, that’s roughly 5000x faster :) The part I’m most excited about is how much control this gives me. GPT-6 Astra can build a detailed 3D prototype in minutes, and I can use Blender to set up the camera and movement however I want, then have H3 Max follow it.

Who misses driving a taxi in San Francisco?

I made this with GPT-6 Astra and @fal PATINA GPT-6 Astra might be the best model I've seen for 3D intelligence and it becomes much more powerful with fal PATINA.