Creator seungho__yeo ( IG ) built an entire anime-inspired 2D motion sequence - storyboard to final render in one hour. Inside Comfy.
seungho__yeo created this entirely in a single workflow pipeline:
→ Poster art
→ Storyboard frames
→ Transitions + animation
→ Edit + final render
What usually takes multiple apps was built in one tool.
The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.
The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance.
Access to all other Claude models is not affected.
We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible.
Read our full statement: https://t.co/bwn0sximKZ
Iv been granted access to Bytedances edge computing image editing model DreamLite! (yes you have to apply for it)
I coded up a quick UI for both training as well as inference testing.
immediate thoughts on this tiny little beast, it needs work. It 100% does the thing, super fast, but given its tiny size, it needs to be fine tuned to get more quality out of it.
The future is already here and nobody is talking about it. Some dev (me) just hacked their Ray-Ban Display glasses to run GBA games, and they’re playing Pokémon Emerald with a transparency passthrough. Step outside and you’re catching Zigzagoons in actual grass. #metarayban
We’re FINALLY seeing the trend of pixel space diffusion models hit the scene. But why is pixel space more desired than the traditional VAE? It’s because compression.
We usually have to compress our images through a VAE (variable auto encoder) to bring the image to a lower dimensional representation before doing transformations on it, then decode it back into pixel space.
This results in slight alterations in the pixels being sent in even if they arnt actually being touched by diffusion (Think masking).
Now obviously compressing something makes it faster, and more efficient. Which is why we went this route in the first place. But with pixel space, we can make pixel perfect manipulations without actually having to touch the surrounding pixels!
This is still a hot topic of research, and you can tell by the 1k resolution lock on this model that pixel space is still insanely computationally expensive. But it’s nice to see labs really start to throw resources at the problem!
I bought the Meta Ray Ban Display glasses and hacked in my own chatbot I call “Atlus” - it can take and review photos and use the browser to look up stuff and the voice is powered by Eleven Labs. The developer SDK is pretty permissive so far and only in v0.7- this is the future. #elevenlabs #meta #rayban #hacking
Introducing Echo! Voice-to-text that runs on your machine, not someone else's.
- 100% local inference
- LLM-enhanced, fully prompt-able
-one payment, yours forever
- free updates, always
the cloud is just someone else's computer. Echo runs on yours.
Introducing Echo! Voice-to-text that runs on your machine, not someone else's.
- 100% local inference
- LLM-enhanced, fully prompt-able
-one payment, yours forever
- free updates, always
the cloud is just someone else's computer. Echo runs on yours.
Iv been developing this for the last while and its completely changed how i interact with computers.
Echo is a small piece of software on both macOS and Windows that lives in the tray icons, allowing you to transcribe and even utilize a local instance of Gemma 4 to enhance your transcriptions!
Its a one time buy and you own it forever. It has a built in updater so i will be supporting the software until its perfect!
This is a daily driver for me and all of my claude coding sessions. I hope it brings you guys as much usefulness as it has me!