Amor Fati. / Co-founder @imgn_ai
Dubai
Joined February 2021
- Tweets863
- Following267
- Followers186
- Likes9.8K
J. retweeted
It's time to start getting this into your hands!
We'll be rolling out imgnAI Director in Beta today - a full Adobe Premier style suite that runs natively in your browser.
This is our most ambitious build to date, allowing you to create seamless cinematic experiences right in your browser - and fully supporting local uploads, and local exports completely for free.
Please remember that this system is in Beta - and while we've put an enormous amount of time into polishing this up as best we can, you should still expect to run into bugs and other small annoyances as we iron out all the kinks.
Director is mobile-ready, and optimized for low-end devices with in-browser preview preparation/early rendering.
We hope that Director will set a new standard for free browser tooling, alongside making long-form AI content far more ubiquitous, and easier to create across our space.
Along with Director, we also have a host of general new updates going live - from ui/ux improvements to new models, inclusive of an illustrious-based Anima finetune, and the latest offerings from OpenAI's GPT-Image 2.5. And of course, even more improvements to our in-house H3 Video model!
These updates will begin rolling out this evening. Note that we expect services to be intermittent for up to 1 hour as we begin to roll these out - we'll keep you updated as the rollout completes!
J. retweeted
At @imgn_ai we're putting in some dev effort to help the hypersnap migration take flight. We've been working on getting the quorum web client secure, up to scratch, and ready for everyone to use. We're hyped with the TRUE decentralised vision for farcaster, and excited to be building alongside @cass_on_mars fo' dis! 😻
Personally, I have been working on a CLI & TUI client which is 100% pure hypersnap (I don't use any farcaster APIs), to provide a truly decentralised client for nerds, shells and agents.
If you know me, you'll know that I firmly believe that strong CLI/TUI tooling is a great way to extend, automate, script, and interface with the ecosystem! This means you can hook your account up to an agent, to ask them for notifications, post, and more!
J. retweeted
Casual reminder that @imgn_ai is based, cheaper than competitors, and the official ai partner of The Emonad Foundation.
If you’re burning money because you’re getting violated on fees that’s sad.
You could just stop doing that tbh.
Imagine that.
Emonad
J. retweeted
Preview of something we've been working very hard on these past few weeks.
A full Adobe Premiere style suite right in your browser.
Oh, and it's completely free - you can even upload content you made locally, or on other sites.
Coming very soon, mobile-friendly too!
J. retweeted
There is a crypto centric AI project that delivered true 4k before Midjourney.
It's 70% cheaper than higgsfield.
Delivers every major AI model at wholesale + margin without scalping premiums.
And I may have annoyed @jessepollak and @Nibel_eth with my base fudding that they are paying attention to the only AI project on base that isn't ripping people off and is delivering on what crypto should do - cheap, anonymous AI access - no borders, jurisdiction oversight or scalping.
If Base stick to their vision of supporting builders it'll be a quick journey for @imgn_ai to become a market leader.
If base don't stick to their vision then it'll take a little longer.
I like a good underdog story and this is it. Imgn_ai season. Soon.
J. retweeted
Imitation is often called the most sincere form of flattery
..but maybe not when you need to imitate a Wassie and a couple of cats working from a basement
More major releases coming next week ✌️
Congrats to Higgsfield on launching Katana, their video editing tool
You can also edit videos using our IMGN H3 model, released.. over a month ago
At around 30% the price of Higgsfield, and up to 2K resolutions
..on imgnAI Katana: kat.imgnai.com
J. retweeted
Recently, we unveiled our enhanced MiniMax H3 harness which allowed for infinitely extendable (seamless!) video.
Since then, we've begun working on the next logical step forward - an Adobe Premiere style suite, running natively in your browser!
Supports FREE local use/exports, and premium server-side rendering with access to our full model library.
Coming soon - watch this space!
J. retweeted
> Jesse finally buys some base:0x9bba915f036158582c20b51113b925f243a1a1a1 after we built on Base for 3 years
> He immediately blames it on a scammer
bros I fucking hate this place
J. retweeted
The Emonad Foundation is pleased to announce @imgn_ai as our official generative AI partner.
To celebrate this very bullish partnership we are running a content creation contest.
Make Emonad content using the 7 reference images in the reply below.
Drop your images or videos in the comments.
Tag @imgn_ai in your submission, like, and rt this post.
The top 3 submissions can make an account on imgnAI and will recieve free credits to create on their stack.
They have the latest image, video, and language models. They also have their own in-house lineup of models.
For builders and devs they have their Katana API. One endpoint for image, video, and text. Pay with imgnAI credits, or skip the account and pay per generation.
They also added x402 payments on @monad. Generate with USDC on Monad. No account. No API key. Wallet pays, model runs. Built for humans AND ai agents.
Privacy is baked in. Prompts can run end-to-end encrypted. Anonymized models so jobs aren’t tied to an identity. x402 privacy means you can generate without an account.
The videos in this post were all made with Minimax H3 on IMGN.
Have fun.
The ticker is emonad:native btw.
emonad
J. retweeted
We're working to solve one of the largest challenges AI creators face daily: how do you make seamless long-form content? Superficially, it seems simple: "brav just use first+last frame"... but the nuances go deeper.
- How do you preserve the detail fidelity across videos? Changing positions? Changing props, etc?
- How do you maintain consistent scenes across gens where new prompts are sent to the models every time? When you're dealing with 20 clips or more, this quickly becomes a massive challenge.
- What about seamless audio?
- What about adding sound effects manually, transition effects, and the likes which are commonplace in video editing software but missing in AI?
- How do you balance refined control over the process, while giving LLMs creative freedom to "go wild" within set bounds to remain consistent across videos?
TL;DR: we're close to solving this at @imgn_ai, and it's coming soon. This video is a very early demo of our upcoming tooling: throwing together some of the concepts I mentioned above.
The Cats Aren't Alright - The Cataclysms
Starring our own cats Shadow (vocals), Heathcliff (lead guitar), Pixie (bass) and Teg (drums). 😻
Tech demo of a longer video made using an upcoming tool release with our IMGN H3 video model! This is composed of over 15-20 video gens, all using the same character reference profiles: witness the consistency in characters across all the shots!
Lip syncing, video-to-audio beat matching, character consistency & fidelity (check the beauty mark on Pixie's face, and the rose in her bass & teeth throughout), seamless transitions across videos... cheap as chips. Welcome to IMGN H3.
J. retweeted
I don't know much about these AI deployment products, but it just irks me greatly when there is a competing product (@imgn_ai ) that is cheaper, offers the same models and has a much better token structure than this and is trading at like 1-2m mktcap.
J. retweeted
Brand new and LIVE on imgnAI: IMGN H3! 🧵👇
IMGN H3 is MiniMax H3 supercharged - optimized for maximum quality, running under our custom harness.
- Prompt like it's Seedance 2.0 - no LLM required.
- 4 Quality presets, with more to come!
- 👉Starting from 4c per second!! 👈
J. retweeted
H3 arrives very uncensored out of the box
We have lived to see man-made horrors beyond our comprehension
imgnai.com may never be the same again 👌
J. retweeted
We're hosting an upcoming workshop with @inconetwork to show you how generative AI can be used to improve your game dev pipeline!
We'll also be dishing out some free credits to let you see how imgnAI and our Katana API can fit right in with your workflow! 👀
🏝️ INTRODUCING THE INCO X @MEGAPOT SUMMER GAME JAM 🏝️
🛠️ workshops from @base, Inco, @megapot, @imgn_ai, @sendarcadefun, @kamigotchiworld, and more
💻 then, a two-week hackathon, with prizes for the best games
what will you build?
J. retweeted
🏝️ INTRODUCING THE INCO X @MEGAPOT SUMMER GAME JAM 🏝️
🛠️ workshops from @base, Inco, @megapot, @imgn_ai, @sendarcadefun, @kamigotchiworld, and more
💻 then, a two-week hackathon, with prizes for the best games
what will you build?