Opens profile photo
Follow
Justin Means
@justincmeans
Creating and
Technology CompanyUtah, USAjustinmeans.comBorn May 10Joined April 2014

Justin Means’s Tweets

Create secret phrases known only offline with your family and friends, which can be used for validation. Treat it like your private key, or your totem from Inception. Or risk being cloned and scammed of everything…
Image
Quote Tweet
AI of you will call mom / dad asking for money. It’s an emergency, “you” say, transfer the money immediately. This will be happening soon, at scale, and not many talking about it.
Nearly ready to kick off, phew Education will never be the same
Quote Tweet
Drop me a message if you’d like to come work on the real world version of the Young Lady’s Illustrated Primer with us. It will be open source, vertically integrated and help close the digital divide. Actively hiring, more details in the new year. twitter.com/pwang/status/1…
13
228
Power On: Apple is planning to debut the Reality Pro mixed-reality headset this spring, ahead of WWDC, and ship it later in the year. The push to launch — along with the economy — will lead to an otherwise muted 2023 for Apple.
27
714
Show this thread
Stable Diffusion with coherent text? Or something music related? 👀
Quote Tweet
If I were a swan, I'd be rainbow-colored. If I were a train, I'd be intergalactic. If I were a model, I'd be open-source. @AVShonenkov, @_bra_ket, @_gugutse_, @susiaiv, @StabilityAI #deepfloyd #deepfloydif
Image
Image
Image
Image
2
I thought we were done with big surprises for a while, but Bing offering serious competition to Google would be even more surprising than a 100-year pandemic or a European war.
160
3,385
Photoshop used to take minutes to apply a filter. Diffusion algorithms feel like the early days of CPU renderers. Wonder if by 2024 we’ll still use diffusion or something entirely new. Excited for things like this and Distilled Diffusion in 2023
Quote Tweet
Muse: Text-To-Image Generation via Masked Generative Transformers abs: arxiv.org/abs/2301.00704 project page: muse-model.github.io a t2i Transformer model that achieves sota image generation performance, significantly more efficient than diffusion or autoregressive models
Show this thread
Embedded video
1:40
25.8K views
1
Interpolation between two prompts (with negatives too) running well. Next up is implementing DPM-Solver++ scheduler to decrease step count, and expand to multi-prompt key frame interpolation.
Quote Tweet
Interpolating between prompts in the latent space of Stable Diffusion entirely on-edge with the M1 Mac and CoreML. Product announcement soon! 🆒 ⁦@Outtakesllc⁩
2