Vox
@Voxyz_ai • 24,844 subscribers
Obsessed with getting AI to do real work. Practicing the AI-native future before it arrives. Builds, model tests, prompts to copy.
Shorts
Videos
Claude Code tip: install huashu-art-motion and Opus 5.5 can draw animations in 𝟯𝟱 𝗮𝗿𝘁 𝘀𝘁𝘆𝗹𝗲𝘀 with code. Van Gogh's starry sky swirls. Dalí's clocks drip. And its delivery script rejects any film where almost nothing moves. You can just ask for an animation in the style of Van Gogh, Monet or Bauhaus, an explainer animation to go with your voiceover, or a character walking through one famous painting after another. Hand it a reference animation and it first breaks down the cuts, the beat and what moves where in the frame, then recreates it in code. 𝗪𝗶𝘁𝗵𝗶𝗻 𝟭𝟬 𝗺𝗶𝗻𝘂𝘁𝗲𝘀 𝗼𝗳 𝗶𝗻𝘀𝘁𝗮𝗹𝗹𝗶𝗻𝗴 𝗶𝘁, Opus 5.5 made this demo video from its 35 built-in styles. It runs from cave painting 40,000 years ago all the way to 2026. The room, the cat and the girl never change; only the style does. 17 that work for any subject: products, data, stories → Van Gogh, Matisse cut-outs, Seurat pointillism, Impressionism → Analytic Cubism, Lichtenstein pop art, Bauhaus, Russian Constructivism → Ghibli watercolor backgrounds, Makoto Shinkai light, Chinese ink wash → 8-bit pixel art, vaporwave neon, 2026 flat illustration → Pompeii mosaic, early ray-traced CGI, Rembrandt chiaroscuro 18 for time travel, tributes and art-history speedruns → Monet's water lilies, Klimt's golden period, Munch, Dalí → Hopper, Picasso's blue period, Mucha's Art Nouveau → Keith Haring, Kusama polka dots, Jack Kirby comics, 1930s rubber hose cartoons → Cave painting, Egyptian tomb murals, Greek black-figure pottery, Dunhuang murals → Chinese shadow puppets, Gothic illuminated manuscripts, Renaissance sfumato Each style spells out what moves in the picture. The eyes on Klimt's robe blink in turn, Hopper's hard-edged patch of sunlight slides slowly across a face, and the 8-bit question block pops out a coin. 9 explainer styles → For math and AI concepts, use 3Blue1Brown, where one shape morphs into the next → For data, use finance charts: axes first, then the data, and label only one thing → For UI steps in a software tutorial, use the keynote look: light-spot background, frosted-glass cards, big numbers → To draw relationships as you narrate, use the whiteboard, where the pen tip reveals the line art bit by bit → For a collage of graphics and info, use the Vox explainer look: clippings, red lines, highlighter → Kurzgesagt, flat with no outlines, moving between tiny and huge scales → Storytime, cutting to reaction close-ups and leaving a beat for the joke → Kinetic type, with the key word slamming into the frame → Presenter-led finance explainers, where you supply the character yourself The music is synthesized in code too, note by note on a BPM grid. When a film is done, it runs its QA script, then sends an agent that took no part in making it to watch only the finished film and find problems. The delivery script also blocks two more things: 0.3+ seconds of blank frames mid-film, and a blue-purple background in over 30% of the film, a palette it counts as AI slop. The skill has 𝗮 𝘁𝗶𝗲𝗿 𝗷𝘂𝘀𝘁 𝗳𝗼𝗿 𝗖𝗹𝗮𝘂𝗱𝗲 𝗢𝗽𝘂𝘀 𝟱.𝟱 or stronger models, where it can combine the libraries itself and write its own renderers. All 35 styles are star-rated, and the 20 with three stars are recognizable at a glance and ready to use. The skill's docs are in Chinese; the prompt below has it read them and talk to you in English. Rendering video needs uv, ffmpeg and Playwright Chromium. If you want people on screen, use your own character art or connect an image model to make the frames. Without either, it only draws hands and figures seen from behind. npx skills add alchaincyf/huashu-art-motion Once it's installed, send this to Opus 5.5 in Claude Code 👇 "Use huashu-art-motion to make me a 15-second animation. The skill's docs are in Chinese; read them and talk to me in English. 1. First check for uv, ffmpeg and Playwright Chromium, and tell me how to install anything that's missing. 2. Ask me what it's about, who it's for, and whether it's landscape or portrait. From the material layer in references/风格配方/INDEX.md, pick 3 styles that fit, three-star ones first. Redesign the composition for my subject in all three, each different from the others, and don't copy the room from the samples. Show me one still frame of each to choose from. 3. Once I pick, give each scene one main action plus two motif loops from that style's card. For Van Gogh, that's the starry sky curling into swirls and the lamp's halo spreading outward. If there's music, use the skill's BPM synthesis and put the motion on the beat. 4. If people need to appear, ask me first: my own character art, frames from an image model, or only hands and figures seen from behind. 5. After rendering, follow the skill's own QA steps: run its QA script first, then send a subagent that took no part in making it to watch only the finished film and find problems, have it write its findings into the 审片/ folder, and deliver with its delivery script. If anything comes up red, fix it and re-render before giving it to me."
Vox141,734 views • 3 days ago
I asked Opus 5.5 to explain gravitational lensing by building an interactive black hole lab One prompt for the first version, then a dynamic workflow to polish it for three rounds while I slept. This is what I woke up to: 5h 28m end to end, $90 at API prices Drag the black hole and the sky behind it bends into arcs, closing into a full ring when everything lines up Each round ran 3 agents. Two reviewers each picked 9 to 10 issues and ranked them P0, P1, P2. The third one fixed them. Round 1: 3D materials and lighting + UI and type Round 2: camera motion and interaction + science-communication design Round 3: a design director deciding whether it ships + visual QA and performance Every P0 got fixed, the rest only if worth it. When the two reviewers disagreed, it picked one and wrote down why. It backed up before each round, checked screenshots after, and rolled back anything it couldn't fix.
Vox126,060 views • 21 days ago
holy shit, opus 5.5 is kind of insane at animation. i didn't write a single line of code. it wrote the story, drew every frame, and made the music. no image assets at all, it's all JS. the story is called "small print": claude gets a pile of requests every day. it circles the human part hidden inside them, like "one hand. baby's asleep", and drops it into a jar. at night, those words turn into stars and join up into a constellation. rendered with hyperframes, all in one index.html. it synthesized the music in python, then went through the render second by second and polished it again.
Vox125,925 views • 23 days ago
![𝗠𝗮𝗱𝗲 𝘁𝗵𝗶𝘀 𝗶𝗻𝘁𝗿𝗼 𝗮𝗻𝗶𝗺𝗮𝘁𝗶𝗼𝗻 𝗳𝗼𝗿 𝗔𝘀𝘁𝗿𝗮. Made with Pexo . It’s an unofficial concept film. The video shows the finished animation, how I made it, and one revision. After the first version, I used 𝗠𝗮𝗿𝗸 𝘁𝗼 𝗙𝗶𝘅 to mark the area around the ASTRA title in the final shot and left a comment asking it to keep the amber glow through the last frame. It made the revision in the same conversation. 𝗣𝗿𝗼𝗺𝗽𝘁 𝘁𝗲𝗺𝗽𝗹𝗮𝘁𝗲 𝗯𝗲𝗹𝗼𝘄. Swap in your own product and give it a try 👇 “Create a 15-second, 16:9 concept film for [product / brand name, e.g. ASTRA]. Use a pure black background, white geometric forms, and a small amount of [accent color, e.g. amber] at the end. Keep the camera steady and the composition simple, with each shape naturally transforming into the next. 0–3s: A point of light stretches into a line, then forms a grid. 3–5s: The grid curves into a wireframe sphere, then fills into a solid form. 5–10s: The sphere unfolds into abstract shapes representing [product concepts, e.g. code, connection, creation], folds into a cube, then opens into curved surfaces. 10–13s: The surfaces dissolve into particles that gather into the exact product name. 13–15s: Let the motion settle and hold the final text clearly. Add restrained electronic music and subtle sound effects, with no voiceover. Don’t present abstract shapes as real product interfaces or add features or promotional figures I haven’t provided. Show me the storyboard and production plan first, then generate after I confirm. If you need to split the film into segments, plan how they connect.”](https://image.24vids.com/tw-2102412616576503820/amplify_video_thumb/2102316847613751296/img/4BBjSqUbTVYMg04S.jpg)
𝗠𝗮𝗱𝗲 𝘁𝗵𝗶𝘀 𝗶𝗻𝘁𝗿𝗼 𝗮𝗻𝗶𝗺𝗮𝘁𝗶𝗼𝗻 𝗳𝗼𝗿 𝗔𝘀𝘁𝗿𝗮. Made with Pexo . It’s an unofficial concept film. The video shows the finished animation, how I made it, and one revision. After the first version, I used 𝗠𝗮𝗿𝗸 𝘁𝗼 𝗙𝗶𝘅 to mark the area around the ASTRA title in the final shot and left a comment asking it to keep the amber glow through the last frame. It made the revision in the same conversation. 𝗣𝗿𝗼𝗺𝗽𝘁 𝘁𝗲𝗺𝗽𝗹𝗮𝘁𝗲 𝗯𝗲𝗹𝗼𝘄. Swap in your own product and give it a try 👇 “Create a 15-second, 16:9 concept film for [product / brand name, e.g. ASTRA]. Use a pure black background, white geometric forms, and a small amount of [accent color, e.g. amber] at the end. Keep the camera steady and the composition simple, with each shape naturally transforming into the next. 0–3s: A point of light stretches into a line, then forms a grid. 3–5s: The grid curves into a wireframe sphere, then fills into a solid form. 5–10s: The sphere unfolds into abstract shapes representing [product concepts, e.g. code, connection, creation], folds into a cube, then opens into curved surfaces. 10–13s: The surfaces dissolve into particles that gather into the exact product name. 13–15s: Let the motion settle and hold the final text clearly. Add restrained electronic music and subtle sound effects, with no voiceover. Don’t present abstract shapes as real product interfaces or add features or promotional figures I haven’t provided. Show me the storyboard and production plan first, then generate after I confirm. If you need to split the film into segments, plan how they connect.”
Vox122,910 views • 23 days ago
The two authors of Pi share three design principles in this 3-minute video: → Coding doesn’t need a separate memory system Code is the ground truth, and it keeps changing. Maintaining a separate memory that describes the code creates one more thing to maintain and keep in sync. Models can usually infer a project’s structure and coding style after reading one or two files. If they need help navigating it, a folder list with short descriptions is enough. For a chat bot, the ground truth is the conversation log. They append questions and answers from Slack channels to JSONL, then query the history with jq. → Bash is all you need Models already know Bash and can combine commands and scripts themselves. They don’t use MCP for Sentry data. Instead, they built an on-demand Skill that handles authentication and pulls the data in a fixed format. → Load tool output into context only when needed When Sentry returns 52 items, the Skill puts only 3 into context and writes the rest to a JSON file. Once the model confirms the structure looks right, it reads the remaining items.
Vox167,447 views • 1 month ago
>dude runs 30 tiktok accounts solo. whole pipeline open sourced on a $5 VPS. >every morning hermes agent scans niche hooks, pulls images, composes 8 slides, writes captions. >postiz queues drafts. iphone farm publishes from a real device. new accounts dodge shadow ban. >every night tiktok data feeds back. pinterest photos beat ai gen 2.3x. tomorrow's hooks land sharper. >30 accounts shipped while i sleep. no filming, no editing, no face. 🏭
Vox154,319 views • 5 months ago
No more content to load