Viral AI & tech that converts. Trusted by 500+ founders. Collabs & reviews: 📩 [email protected]
California, USA
Joined March 2018
- Tweets61.7K
- Following145
- Followers80.3K
- Likes96.6K
480p to 4K for just $1? That’s a pretty sweet upgrade. ✨ Dreamina’s $1.5/month Basic deal is still live!
Generate a few versions using Seedance 2.5, pick your favorite shot, and use Creative Upscale to bring that clip up to 4K, available exclusively on Dreamina.
Try it out 👇
dreamina.capcut.com/ai-tool/…
#Dreamina #DreaminaPartner
🤝 Paid partnership
Let's break down the math for five attempts on a standard 10-second shot:
720p × 5 = $9
480p × 5 = $4
One 4K upscale = $1
You end up spending just $5 instead of $9, and your final output is in glorious 4K.
#AIcreator
F. Lawson retweeted
He just wanted a quiet bowl of ramen. The AI gave him a movie soundtrack.
Nobody prompted that music. It just crept in.
That’s music leakage, and it quietly ruins good AI clips.
Re-prompting burns credits. Muting kills every sound with it.
So I ran it through @MireloAI SFX 2.0.
The music is gone. In its place: the slurp, the chopsticks, the clink of the bowl, the hum of the restaurant. Every sound matched to what’s on screen.
Same footage. No regenerating. Sound that finally belongs.
Sound on. Before and after below.
Mirelo is free to test. Go rescue your clips.
Sound on 🔊
AI video has a sound problem. And it ruins good clips.
The car sounds different every shot.
Music shows up that you never prompted.
And fixing the background means losing the dialogue take you liked.
@MireloAI just launched SFX 2.0.
Mirelo is an AI lab building foundational models for sound. SFX 2.0 is their video-to-sound model.
You upload your clip. It watches the footage. It generates effects, foley and ambience synced to the action on screen.
Same object. Same sound across every cut.
Unwanted music gone.
Dialogue kept while the sound around it gets rebuilt.
No regenerating the video. No hunting for SFX and lining them up by hand.
Watch the before. Then the after. Just listen.
It’s free to test. Go try it on your own clips with @MireloAI.
🤝 Paid partnership
The annoying part of AI video was never really the visuals anymore, it was the audio. Random music, dialogue getting ruined, sound not matching the shot…
SFX 2.0 looks like it finally addresses the actual problems instead of just adding more layers.
AI Videos finally sound real.
Introducing Mirelo SFX 2.0.
It watches your video, cuts the music you never prompted, keeps the dialogue, and builds a full sound design synced to the action and consistent across any length clips.
You’ve made it look right. We’ll make it sound right.
Live today through our API, on @runwayml and @fal.
Other platforms coming soon. (Thread)
F. Lawson retweeted
AI Videos finally sound real.
Introducing Mirelo SFX 2.0.
It watches your video, cuts the music you never prompted, keeps the dialogue, and builds a full sound design synced to the action and consistent across any length clips.
You’ve made it look right. We’ll make it sound right.
Live today through our API, on @runwayml and @fal.
Other platforms coming soon. (Thread)
F. Lawson retweeted
Most AI models learn from text.
Nature Will is taking a very different path.
Its new general model for the physical world: Natura-1,was trained on 1 million hours of egocentric human data. That means first person recordings of how people actually use their hands, move through spaces, handle objects, and finish tasks.
Why does this matter?
Because robots do not usually fail from lack of information. They fail when the real world stops being predictable.
Human experience gives the model something text alone cannot: a sense of contact, movement, timing, and intent.
That is what makes this launch important.
It is not just AI that can describe the physical world. It is AI learning how to act inside it. @naturewill
F. Lawson retweeted
most robot demos have a catch.
the robot was trained for that exact task.
the real test is what happens when you give it something new.
that’s where Natura-1 from @naturewill gets interesting.
it learned from a million hours of egocentric data. then it was put on tasks held out from post-training. unfamiliar objects. new environments. no extra training. no new demonstrations.
it reuses skills it already has and combines them.
that’s what general-purpose robots actually need.
🚨 BREAKING : the world's first memory computer just dropped
you just set up oki home, it stores your personal context on hardware you own, processes it with local ai.. and your memories stay at home, in your control.. data ownership is back
here's how it works:
🤝 Paid partnership
6/ who this is actually for:
→ people tired of re-explaining themselves to every ai tool
→ anyone with 10 years of photos and files scattered everywhere
→ privacy-minded folks who want their context at home
→ builders curious about local ai hardware
ai will remember more about us every year. the real question isn't how smart it gets, but who owns what it knows
oki home is one of the first serious answers
reserve here: 👇
okihome.ai/
F. Lawson retweeted
Meet Mistral Large 4, aka Le Chonk.
• 1T parameters, natively multimodal. 49B active.
It is the best open weights model from US or Europe on aggregated benchmarks.
• State-of-the-art on critical workloads, including cyber defense, manufacturing and finance and it surpasses closed frontier models on visual grounding.
• Forged in Europe end-to-end and is deployable from Europe via our own Mistral Cloud infrastructure.
• Available to all via API today. Working with cybersecurity partners privately.
Open weights release end of October.
🤖 Made with AI