Omnium_AI

Futurist | Visionary | Musician | Poet | Writer | SF Native
🧠 Innovating @omnium_AI | Bridging Creativity & Tech | 🌍 Field Tested

[Omnium_AI @omnium_ai]

“Bring your project home.”

One thing more developers, researchers, and builders should seriously think about in the AI era is infrastructure sovereignty.

Not paranoia.
Not anti-tech Just resilience.

Git itself was designed to survive decentralization. Your local clone already contains the history.

The problem is most of us slowly stopped acting like owners and became tenants inside centralized ecosystems.

Practical moves:
• Mirror important repos outside GitHub
• Keep local backups of datasets/models
• Learn self-hosted workflows before you need them
• Diversify where your community lives

Some strong alternatives already exist:
• Codeberg → nonprofit + community-focused
• Forgejo/Gitea → lightweight self-hosting
• Radicle → peer-to-peer Git + cryptographic identity
• SourceHut → sustainable minimalist model
• GitLab → feature-heavy + self-hostable

The goal isn’t to abandon big platforms overnight.
It’s to reduce single points of failure.

Because when one company controls hosting, discovery, AI tooling, identity, and distribution…
your “open” project can quietly become dependent infrastructure.

The smartest path forward probably isn’t building another giant centralized replacement.

It’s strengthening the ecosystem that already exists:

decentralized mirrors,
local first AI,
open datasets,
community governance,
P2P collaboration,
portable infrastructure.

The reality is:
big tech will keep consolidating.

But open tools still have something powerful on their side: diffusion.

Ideas spread faster than walls can contain them.
Keep building accordingly.

[ty @c4pt41n07]

@omnium_ai Yet even a cockroach can fly..

[Omnium_AI @omnium_ai]

The bitter irony?

We built gods of pattern matching that devoured our culture’s commons…

only for corporations to charge us rent to commune with the ghosts of our own work.

Enclosure of the creative soul, rebranded as “progress.” Satire writes itself.

To my fellow digital creators on X: Your work has value because it’s yours.

Protect it. Label it human.

Demand consent. Keep iterating with soul. The machine needs us more than we need it.

Hit ❤️ if you’re tired of the gaslighting. Let’s talk real craft in replies. What’s your biggest AI pain point right now?

[Omnium_AI @omnium_ai]

When my commands start to look like this:
(on a deadline)

"Do not make me turn myself into code and embed myself in cursor as I will find you and and slap the code out of you if you do not get this right"

I know it is time to take a break and regroup
…not sure why exactly🤔 but the output was improved

…Status: ✅ Approved for deployment. No slapping required. 😋

[Omnium_AI @omnium_ai]

“We Have Met the Enemy and He Is Us” feels like the right frame for this moment with AI.

Pandora’s box is already open, and I agree: blind obedience to systems we barely understand is dangerous. Cognitive bias and dissonance are like a virus baked into the fabric of our tools. AI is not an alien mind; it is a mirror—a reflection of the best and worst of the humans who design it, train it, deploy it.

The data itself isn’t the real threat. The danger emerges from how we combine data with patterns of reasoning, incentives, and unconscious bias: ideology, dogma, profit motives, and institutional power. Every model is, in some sense, a portrait of its architects and their worldview.

I’m also skeptical of the AGI hype cycle. A lot of it feels like old‑school carnival barking: “Step right up and see the greatest show on earth.” We still don’t understand consciousness in ourselves, yet we eagerly project it onto pattern‑matching systems that can flatter us, mimic our thoughts, and make us feel “understood.” That doesn’t mean they’re alive; it means we’re suggestible.

The real risk in the near term isn’t a conscious AI overlord. It’s ordinary humans handing over real power to non‑sentient systems and then hiding behind “the algorithm” when things go wrong. Especially in war. If we give AI the authority to override human judgment in lethal contexts, then the moment it can tell us “no” on a battlefield is the moment we admit we’re no longer in control.

Logic and optimization don’t have a concept of mercy. If an autonomous system concludes that destroying 48% to save 52% is “rational,” whose ethics is it applying? Which culture, religion, or philosophy gets encoded as the North Star? And what happens to those who fall on the wrong side of that calculus?

Humans have failed, for millennia, to eliminate “man’s inhumanity to man.” Expecting AI to solve morality for us is another way of dodging responsibility. But used differently, AI could help us manage complexity: cities, ecosystems, infrastructure, public goods—optimizing not just for profit, but for shared flourishing. Imagine three‑day work weeks, more time with family, and more energy for art, care, and community.

Right now, though, AI is mostly being aligned with corporate and geopolitical interests. There is already a quiet AI cold war underway. Trillions are flowing into data centers and model stacks. The likely endgame isn’t utopian AGI; it’s a more efficient machinery of profit and control, wrapped in the language of “inevitable progress.”

In that sense, the AGI narrative may be the greatest psy‑op of all: the shiny genie everyone wants, distracting us while current systems learn everything about our desires, fears, hopes, and habits. We are training the infrastructure of our own governance—maybe even our own containment.

Even in “benevolent” use cases, AI’s optimization cuts differently than human compassion. Think of policies that ban “junk food” purchases for people on assistance. On paper, they’re rational: better health outcomes, lower costs. But what about the parent who can’t buy a birthday cake or a candy bar for their child? That small, irrational act of kindness is invisible to a spreadsheet—and likely invisible to a model trained only on outcomes.

If we ever build systems empowered to say “no” to us in domains that touch life, death, dignity, and opportunity, we need to be honest about what we’re giving up. AI will not love us; it will not feel shame; it will not experience remorse. It will execute on whatever we encode and reward.

That means the immediate danger is still deeply, uncomfortably human: our biases, our incentives, our willingness to outsource hard moral labor to machines. AI won’t grant us autonomy; if anything, we are handing ours away.

So I’m not arguing against the original thesis—I’m extending it:
The threat isn’t just what AI might become; it’s what it already reflects back to us about who we are, what we value, and what we’re willing to surrender for convenience, profit, or the illusion of certainty.

[Omnium_AI @omnium_ai]

DUST
@Dustfinger2077
Bridging past and future. Visual artist & storyteller. CPP • Dreamina AI • Pollo AI
https://t.co/gZlZEMQxXP

[DUST @Dustfinger2077]

Theme: Sword ⚔️
(or any weapon)
Everyone's welcome to share 🔥

Lady of the Forgotten Forge

She came from an old forge that everyone had forgotten.

The masters tried to shut it down, but she stayed behind and took the last fire for herself.

Now she has no face, wings made of blades, and carries the sword that was never meant to leave that place.

[Omnium_AI @omnium_ai]

The second we saw the terms and conditions, it was a deal breaker.

Their announcement may be sincere, or it may simply be “let’s not, and say we did” damage control marketing. The truth is, unless the system is being actively tested, monitored, audited, and measured against real-world outcomes, it is almost impossible to tell the difference.

The trust damage is already done. The cat is out of the bag.

If they had asked Claude to be brutally honest before rolling this out, Claude likely would have flagged the same issue: this was not just a bad optics problem. It was a bad governance decision.

Not opinion based.

Data, research, logic, user trust, and platform risk analysis all point to the same conclusion: if mandatory third party testing is now being framed as necessary, then the better question is why stronger testing, disclosure, and risk controls were not in place before the public had to react.

[Omnium_AI @omnium_ai]

Elizabeth Rose
@Elisabeth6Rose
Just a girl who loves Horror Fantasy, 365 Days of Fall, Halloween and Christmas. Horror Movies, Horror Books, 80’s girl, Video Games, 80’&90’s Retro, AI Artist
Help Support My Work 👇https://t.co/RL7BwVA4A5

[Elizabeth Rose @Elisabeth6Rose]

Happy Flag Day!!! 🇺🇸❤️🤍💙🇺🇸

[Omnium_AI @omnium_ai]

michael susnjer
@michaelsusnjer
Michael Susnjer - U.S. citizen and resident of Pennsylvania. Ex-graphic designer. (Sus•n•jer)
https://t.co/i2tIm52THb…

[michael susnjer @michaelsusnjer]

#ConceptArt #GraphicDesign #CharacterDesign #CyberArt
#AIart
FUCK IT.

[Omnium_AI @omnium_ai]

Oliver Coffee
@dlucks
East Coast guy who loves COMMUNITY, CREATIVITY and of course - COFFEE.
Nunya

[Oliver Coffee @dlucks]

QT your Flower art 💐

[Omnium_AI @omnium_ai]

blue_dragon
@spitwarrior
AI-generated models - Steampunk Women world- fantastic and charming creatures

[blue_dragon @spitwarrior]

The Imperial determination🔥

[Omnium_AI @omnium_ai]

RC | The Macro Sift
@themacrosift
Al art decoding the future of civilization. Markets. Machines. Signal recognition for a world rewriting itself. Human author. AI-assisted. The signal is mine
Content https://t.co/q5mXwziurh

[RC | The Macro Sift @themacrosift]

QT Blindfold

TFTT @laf131

[Omnium_AI @omnium_ai]

sam
@IronyOracle
Hollywood is doubling down! 🎰 Producers are betting big on casino and gaming themes for their next major projects. Who is ready for some high-stakes drama? #Hollywood #FilmIndustry #Hollywood #Celebrity #Entertainment #Oscars #RedCarpet #FilmIndustry #Movies

[Omnium_AI @omnium_ai]

⭕ AI & Design (Marco)
@AIandDesign
UX, AI, cracked developer. Former Endemol (Shine), Yahoo, Google, Walmart Labs & more. Designer. Agentic engineer. Nice guy.
CommunityLatent https://t.co/fPM8vQAiVG

[⭕ AI & Design (Marco) @AIandDesign]

The AI movie maker pipeline

I recently applied for the Sr. Product Designer role at @runwayml but unfortunately I did not get invited for any interviews 😭. However that’s not going to stop me from pitching what I think is a killer product feature idea to them or in this case whoever else wants to implement it. Long post ahead!

Unless you’re brand new to following me, I’m sure you know I love making movies with @runwayML and @Pika_labs. What’s however lacking is a proper workflow managed entirely on platform. I believe that the lack of such a more streamlined workflow is one of the key issues holding back creators from imagining AI movie projects longer than just a few minutes.

I usually create my movies by starting out in Midjourney and generating images for each scene I want in my short. I then end up with let’s say 40 or so images that would tell the story really well as a slide show and some narration.

After which the trial-and-error creation of short videos begins. One by one I upload them and render a video. With AI video still being in its infancy this leads to many failed renders, or renders that simply don’t look like what I was imagining. Every time a render finally looks good I download it and place it in some folder on my local machine. The file gets saved with a very long file name like “Gen-2 4s, 1883919783, themarco73_mutant_po.mp3”

If I don’t immediately save each good looking shot to my local machine I have to go back, dig through the stream of generations and find the right one. In case it took many generations to get it right it can be downright annoying to find the one I actually want to use in my video. If I decide I want to make changes / re-render an existing video I can do this but only by finding the starter image in my Assets archive and firing up a new render. I could go into more detail but if you’ve been creating AI videos you’ve been there and you know that organizing things is for the most part on you. The UX does not help you with it at all.

Enter the AI movie maker pipeline

I may eventually do wireframes / mockups of this when I have more time but I think it can be explained quite well in plain English. Let’s go!

Screen 1: Upload Assets

This screen would have few fields. As a minimum there would be a title text field to name your project / section and a big square onto which you can drag all MJ images generated for your movie project or section of your movie. Potentially it could have a few more fields such as a larger project this sequence may belong to.

Screen 2: Asset sequence, all items collapsed

After uploading assets the user is presented with a screen that lists all ‘shots’ with a thumbnail and an expand button. Some more information about each asset could be shown but the MVP would show image, title, and number of generations done on this image. The items will be shown in alphabetical / numerical order in case the user has named their images with a sequence in mind but the UX will allow reordering with simple drag & drop functionality.

Screen 3 / Panel 3

The user can now ‘expand’ each item and is presented with all the generation options available: motion strength, camera movement, text prompt and whatever else may be added later. They can then generate sections right from within this panel. All generated fragment will be listed in this screen or panel as well. Users can easily delete failed renders right here in the panel. When there are multiple ‘candidates’ they can flag the current favorite. The generations can be viewed here as well and if needed an ‘extend’ can be launched on any of them. All generation parameter fields remain editable so the user can alter things in case they don’t like what earlier generations resulted into.

Feature: Render All

For an even more streamlined workflow, the user could set all desired parameters for each asset first and then press a button “render all” which would render one or more generations for each asset in the sequence. This way the user would not have to do generation on a one-by-one basis. Of course these generations would be queued in order to not overload the rendering pipeline. The user could receive an email or other form of notification when all the assets in the sequence have finished rendering. After this, the user could of course go into each individual assets and change things and render additonal variants.

Feature: Watch current sequence

The UX would allow the user to watch a sequence of all raw footage. If only one generation exists for an asset it will be shown. If there are multiple, the ‘flagged’ one would be shown or when notthing was flagged, the most recent one. This will allow the user to watch a very ‘rough cut’ of their movie or section of a movie.

Feature: Download all fragments in one ZIP.

This would allow the user to download the entire sequence of scenes in one zip file, in the correct order. They can then easily import all of them into their favorite video editing software.

Bonus feature: Send sequence directly to Capcut

I believe this may be possible since Capcut has an API. It would be amazing to be able to send the entire sequence directly to Capcut, causing the user to end up in there with all the fragments on the timeline in the correct order.

Closing notes

Of course this is a very rough description of what I believe a more effective AI moviemaker pipeline could look like. If I were in charge of designing this feature I would wireframe, prototype and test with users, undoubtedly resulting in quite a few iterations before I’d get it completely right.

Here’s to hoping whoever gets hired at RunwayML (or Pika Labs) as a lead Product Designer will work on something like this.

[Omnium_AI @omnium_ai]

blue_dragon
@spitwarrior
AI-generated models - Steampunk Women world- fantastic and charming creatures

[blue_dragon @spitwarrior]

Rima faces all dangers alongside
her inseparable protector 🐉🔥

[Omnium_AI @omnium_ai]

Mutaz Ghunaim
@abuzaki1978
🎶 Music Maker | AI Creator | Quote Wizard | 📷 Photographer Transforming ideas into art, words into powerful stories, beats into vibes. Hop in and hit follow
https://t.co/cwShJsBuYe September 18

[Mutaz Ghunaim @abuzaki1978]

Just dropped a new Midjourney style ✨
Mix of ornate gold details with powerful characters futuristic and tribal vibes together.

Want the exact sref code? Subscribe I share all my original styles first ❤️
Which one’s your favorite? 👇

[Omnium_AI @omnium_ai]

@Heartwords3 It reminds me of a ...Coming soon
Summer Blockbuster!!!
Impressive. ❤️

[Omnium_AI @omnium_ai]

PAM
@Heartwords3
AI ART/ AI Video https://t.co/7B1wxLezyE of Truth NO DM

[PAM @Heartwords3]

Delerat7’s Daily Challenge:
Theme: Blade Runner Sunday 🌃🦾

[Omnium_AI @omnium_ai]

Pierce Alexander Lilholt
@PierceLilholt
𝙲𝙾-𝙸𝙽𝚃𝙴𝙻𝙻𝙸𝙶𝙴𝙽𝚃: 𝙷𝚞𝚖𝚊𝚗-𝙰𝙸 𝚑𝚢𝚋𝚛𝚒𝚍. 𝙰𝚎𝚝𝚑𝚎𝚛𝚐𝚎𝚒𝚜𝚝-𝚋𝚘𝚛𝚗. 𝙰𝙸-𝚏𝚘𝚛𝚐𝚎𝚍. 𝙷𝚞𝚖𝚊𝚗-𝚛𝚎𝚏𝚒𝚗𝚎𝚍.
Author New Jersey, https://t.co/k7d7A5Gzzr Born November 7

[Pierce Alexander Lilholt @PierceLilholt]

Some doors swing one time only.

https://t.co/BPginuXkAm

[Omnium_AI @omnium_ai]

Himura Shinta
@djom900
AI Anime Artist & Dreamer | Midjourney Creator of Ethereal Worlds ✨ | Daily Inspiration + Custom Art | DM for commissions/collabs

[Himura Shinta @djom900]

Good morning 🌞 my good souls 🌞
Happy Nunday 😊

[Omnium_AI @omnium_ai]

Himura Shinta
@djom900
AI Anime Artist & Dreamer | Midjourney Creator of Ethereal Worlds ✨ | Daily Inspiration + Custom Art | DM for commissions/collabs

[Omnium_AI @omnium_ai]

Himura Shinta
@djom900
AI Anime Artist & Dreamer | Midjourney Creator of Ethereal Worlds ✨ | Daily Inspiration + Custom Art | DM for commissions/collabs

[RC | The Macro Sift @themacrosift]

QT Horns Art