I made a composite artwork with the photos I captured during the Starship launch/catch. I'll share it tomorrow (email subscribers first!). Really proud of this, so hope you love it!
In the meantime, here's a teaser...
it's hard to believe people will just fly and walk on the moon and mars in the next decades #spacex hopefully inspires a new generation of scientists and artists
hot take i half agree with: we should retire the term "vibe coding"
it started as a description of one specific thing. you prompt an agent and accept the diffs without reading them
now it gets used for anyone who ships with claude code or codex. and those are really different workflows
gal vered put it well on x: he writes specs, runs agents in tmux sessions, reads every diff, and lets tests gate the merge. same tools, different failure modes
that last part is the point. if you don't read the diffs, your failure is a bug you never saw. if you do read them, your failure is a bad spec or a slow review
i don't think the useful split is "vibe coding vs real engineering". it's what happens between the agent's output and your main branch. who checks it, and what stops a bad change from merging
my bet: the next few years of ai coding get decided by the gates, not the prompts
what's your gate: tests, reading every diff, or hoping for the best?
when does vibecoding stop and ai assisted coding start? been working on a project for a year now, and showrunner for over 3 years, we are at 10.000 commits!
I vibecoded my own iMovie for timelapse π¬
We came to the point where it's faster to vibecode an entire dedicated tool rather than using an already established but general-purpose one
.@bstategames, the developer of @tarkov, told us about the challenges of building the new map with low lighting and discussed the tricky process of creating character animation. Read the interview to find out more and find a code that can get you a set of useful in-game items.
"Any directional light, like the sun or the moon, costs performance. At other locations, the player chooses the time of day for a raid, but Terminal offers no such choice, so every player walks straight into full night, whether or not they came with night vision or a flashlight. The location had to stay playable for all of them regardless, without eating the frame rate."
Read the full interview: https://t.co/gyCBDbkeEk
Deepak Kushwaha released a procedural Blender generator for creating organic-looking webs, ideal for horror environments.
Learn more and get it here: https://t.co/RFIc2xDf5P
HyperFrames has an official YouTube channel
every week we'll deep dive on what we just shipped in HyperFrames Studio and post tutorials for video editing, launch videos and more
first episode explains using a link to rebuild a video and switch it to your story and style
https://t.co/8f7WF0EXly
Food delivery company Meituan drops LongCat Video Avatar 1.6.
It can generate talking avatar videos >10 minutes.
Free & open source
https://t.co/ZyXqIdn8at
We're teaming up to accelerate open-source Al.
OpenRouter is now integrated into Hugging Face directly. Same token, same endpoint, plus live model stats on the Hub and an inference widget we redesigned together.
The best of open, in one place. https://t.co/fNbvtJGSJB
@bria_ai_ RMBG 3.0 is now in ComfyUI via Partner Nodes.
β Cleaner cutouts on hair, fur and motion blur
β True 8-bit alpha matte, not a hard mask
β Same image, same result, every time
β Trained only on licensed dataSelect RMBG3 in the Bria Remove Image Background node.
Seedance 2 release today! But seems completely broken:
Prompt: "World Cup, Dystopian Future in the year 2312, Germany vs. Spain"
WTF?
#Ai#Seedance2#fal