Creaturessa retweeted
One of the most anticipated and requested features in nearly 8 years of building Runway is finally here. Sorry it took so long.
16-bit HDR in ProRes and EXR sequences is here FOR ANY VIDEO MODEL IN RUNWAY.
Pretty excited to get this one out, and this is just the beginning. The next few weeks are going to be packed.
Seedance 2.5
The more complex the scene you're trying to create, the more issues you're likely to run into. You fix one thing, think "ok, now it works," generate it again and suddenly something else is wrong.
Since I usually share my prompts, I try to make sure they work as reliably as possible before posting them. But for your own projects, instead of generating the same scene over and over, you can try fixing the problematic parts with the video edit feature or use a video editor to combine the best parts from different generations.
You can run the exact same prompt 10 times and get 10 different results. If you've looked at any of the open-source AI film projects, you've probably noticed that they often generate the same scene many, many times before getting what they need. Of course, those projects usually have time pressure and much bigger budgets, so that also plays a part.
So don't get discouraged when your prompt doesn't give you exactly what you wanted. Sometimes it's simply the randomness of the model.
For this video, I started with an English prompt. The first result was actually pretty good, but there were one or two small issues. I tried a few more generations while tweaking the prompt each time, but as it became more detailed and crowded, I eventually switched the prompt to Chinese.
And now I'm still not sure whether I prefer the first version or this one. Sometimes you just shouldn't push it too far.
Created on @astorie_ai
You can check the prompt and Midjourney style below.
🤝 Paid partnership
Creaturessa retweeted
Dreamina Creative Partner Program — a new chapter begins 🎬
To every AI filmmaker who puts craft into every frame and intention into every story — this is for you.
Since CPP launched, we've supported 2,000+ creators across 40+ countries. Their work has gone viral, been featured in festivals, and pushed the boundaries of what AI filmmaking can be.
Now we're back with bigger rewards, higher ceilings, and a deeper commitment to creators who treat AI video as a storytelling medium, not just a generation tool.
And we're putting real weight behind it: you can up to 500,000 credits/month in the program.
Ready to join? check the details in the images 👇
Come make something the world hasn't seen yet.
#DreaminaCPP #DreaminaAI #AIFilmmaking
Playing with @dreamina_ai Seedance 2.5, 30 second gen. Experimenting with an absurd, surreal continuous take in a single generation. What do you think?
@midjourney → Dreamina Seedance 2.5 → @topazlabs Astra → @LumaLabsAI (for music)
The Seedance 2.5 guides everyone has been gatekeeping.
Check both before writing your prompts. The way they define reference roles, structure actions into stages, handle continuity and especially the syntax used for audio and text is pretty interesting.
🔗 User Guide: bytedance.larkoffice.com/wik…
🔗 Prompt Guide: bytedance.larkoffice.com/doc…
Creaturessa retweeted
IT'S FINALLY HERE!
🔥 Freepik is now Magnific! 🔥
They loved the brand so much they decided to rebrand their/our name/domain.
Let me tell you the story of how two guys created the fastest growing bootstrapped startup EVER in history and the love story with Freepik 🧵👇
Creaturessa retweeted
1/7: 📢 Big release!
Introducing Astra 2 🚀 Our next-gen creative model with a whole new level of controls, including PROMPTS!
Try it NOW in Astra app
See what it's best for in this short video and 🧵👇
🔥@topazlabs launch ASTRA 2 🔥 and I was lucky enough to beta test it. I was interested in testing with very low res footage and used 832x464 video outputs from Midjourney. The results are impressive!
Testing @topazlabs Starlight Precise 2.5 on one of my beloved 🐿️🤖characters and it's pretty good! I started from 1280x720 Seedance 2.0 output using @dreamina_ai and upscaled it with Topaz Labs Starlight Precise 2.5 .
Try Starlight Precise 2.5 at astra.app.
An experiment using @LumaLabsAI Creative Agents. Luma Agents advance multiple directions simultaneously, and sometimes it gets confusing, because the process is fast! I started from some frames I generated in @midjourney
This is nowhere perfect
I am absolutely thrilled to have received the Silver Alchemy Award, 2nd place, at the [esc] Awards 2026. ALCHEMIST AWARD: this Creator is an experimentalist, explorer, mad scientist, conceptualist and likely a non conformist.
Thank you. 🙏🏻🖤
Creaturessa retweeted
Introducing Pika AI Selves: AI you birth, raise, and set loose to be a living extension of you. They’re rich, multi-faceted beings with persistent memory, and maybe even a peanut allergy. It’s up to you!
Have them send pictures to your group chat. Make a video game about your fish. Call your mom while you do anything but call your mom. The possibilities are as myriad as the stars ✨
Get on the list to give birth to yours at pika dot me
Celebrating my latest CPP collab with @invideoOfficial with some underwater mutations using @Kling_ai 3.0 Omni + @BytedanceTalk Video Upscaler. I used three frames that I generated with @midjourney 6.1, some (not all) upscaled with @Magnific_AI + prompt.
Creaturessa retweeted
⚡ Magnific on the big screen!
I CAN FINALLY TALK ABOUT THIS!
The VFX team of Here (directed by Robert Zemeckis and starring Robin Wright & Tom Hanks) used Magnific for their FX 🤯
To break it all down (+more), I interviewed VFX supervisor Kevin Baillie! 🧵👇
First quick test with @Kling_ai 3.0 with unusual subjects and it’s impressive. I’ve only scratched the surface here, but surely I’ll dive deeper because this update is incredibly powerful.
Amazing news! XERO - Flame in The Wire was just selected by BAIFF - Burano Artificial Intelligence Film Festival. paolarocchetti.com/xero-flam…
Did a pass with the new @Magnific Skin Enhancer on a 2023 low res @Midjourney v5.2 output, and animated it with @Kling 2.6 first and end frame. I’m so excited at the idea of being able to use a library of old outputs that I had to leave “behind” because they were too low res.















