Palu retweeted
Oh boy, it's time to get out the baseball bat again and really pound the balls of someone with a worthless opinion in the modding scene.
A.I decompilation is the best thing that ever happened to gaming. It has, overnight, eliminated the predatory stranglehold that game developers have tried to inflict on the marketplace for decades. It has revitalized dead modding scenes and overcome issues that had stumped private projects for years. This is unambiguously a good thing.
Yet modders like this man are mad and raging incoherently about it. Why is that? Simple answer: it takes away their power.
I'm in a much better position to say this than most, because before I was a vtuber fan and got a real life job, I was a modder. I worked on various simple homebrew projects and contributed to a number of mods that I was interested in. My only coding experience comes from banging my head against NWscript long enough to make sense of it purely via observational learning.
During my time in the modding scene, I was able to observe first hand just what paranoid, tinpot dictators most modders actually were. They were selfish, vain, miserable, mercurial, prone to spontaneous bouts of megalomania, and often abandoned their projects at the first sign of critique or negative feedback.
To this day I quietly mourn the many projects I watched be sabotaged by entitled modders who would rather destroy everything they had worked on than pass it to another person who might change their precious, sacred vision of it. There are wonderful stories and creative writing projects I sunk days of my life into that are dead because one entitled little code-gremlin threw a hissy fit and went on a rampage.
That's why I say you should hate modders as much as you should hate journalists. There are some that are good people, but most are tyrants and egotistical psychopaths who only care about modding because it is a way for them to exert power over a community and get attention when everything else in their life is worthless.
The democratization of modding tools by A.I is unambiguously a good thing. I hope it continues and gets bigger. Every time I see a post like this, a little part of me is happy, because it means one of the vile rats who ruined my enjoyment of the modding scene for the games I loved back in the late 00's and early 10's is suffering the completely deserved fate of utter irrelevancy.
Palu retweeted
niggas will wait for years upwards to a decade for a mod that AI can create in a week lmfaoooo. like this isn't cool, this isn't woke...you literally just look like a fucking retard saying this to normal peopleπππππ
Palu retweeted
So lemme get this straight: Brawlback made their code open source, and now smash players are mad that somebody used it?
Are you people slow? Like are you genuinely just stupid? Can you put the square block in the square hole?
Palu retweeted
OPUS 5.5 IS ALL YOU NEED FOR MOTION DESIGN
this launch video cost $0 and was made entirely with code
everyone else on X says the same shit, then sends you to a website or an MCP
this is completely free and uses 0 external tools
take the prompt and try it in your Claude β
Ask me for: a one-word brand name for the wordmark (a verb works best), 9 to 12 high-res photos, a royalty-free track around 120 BPM with a drop and a quiet breakdown (Mixkit, free for commercial use), and a free stock clip of a plain wall with moving plant shadows (Pexels).
An Apple-keynote-style launch film, 2D only, in one continuous take. Each scene grows out of the last: no fades, blurs or cuts. Objects transform instead. Text rises from a mask line, icons spring up from zero, bars draw across the screen, pages push forward, and a black shape fills the frame before shrinking into the next scene. Use a warm off-white canvas, black UI and iOS 26 liquid glass over the photos. Archivo (wdth 125, weight 800) for the wordmark, Geist for the UI. A cursor triggers every change with actual clicks, drags and long-presses. Zoom the camera screen-studio style so each moment fills the square, and scale the cursor with it.
Banned: crossfades, blur-ins, brightness "developing", 3D flips, particles, glows, holds longer than 1s, anything that feels like a template.
120 BPM, 54 beats, with something happening on every beat.
Open: the wordmark compresses into its own period like an accordion. The dot expands into a black pill, and a label rises inside it. Click: six iris blades close over the label, then snap open onto a photo. The circle turns into a square and shrinks. A grid unfolds behind it like a paper map (center, plus, corners), rearranges into a bento, and a click zooms into one tile, landing right on the drop.
Glass: a glass word appears letter by letter, then melts into a droplet that stretches into a glass toolbar. The adjust icon turns it into a slider. Dragging changes the photo lighting from day to golden hour (two aligned shots), while the held knob becomes a glass lens. It rises into a glass orb, the next photo opens inside it as a circle, and the orb grows into a lock screen with glass clock digits, a date and a home bar that stretches into a glass music player.
Stage: the lock screen pulls back to reveal a phone, with the bezel growing from the screen edge. The Dynamic Island stretches like liquid, pinches off, flies across and expands into a Mac window that rolls up like a blind. Long-press the wallpaper and drag it onto a Safari tab. The page pushes in; when you drop the image, it becomes the hero of a landing page. Scroll: the hero transforms into a framed print on a product card, with the mat and molding growing from the photo's edge. Choose a frame color (it paints across) and a size, then the nav button flies down and becomes "Order print".
Order: one black shape keeps changing: Ordered β β Printing % β On its way (a van on a route) β Delivered β.
Wall: the delivered circle expands until black fills the frame, holds for one beat, then shrinks into the framed print hanging in the real wall footage. The same iris opens onto the print and closes again. The frame fills the screen, contracts into the pill β the dot β the letters spring back out on the returning beat. Last frame = first frame.
1. One HTML file, square 1440x1440. Compute every style from time inside an async seek(t): no CSS transitions, timers or state carried between frames.
2. Use closed-form step responses for springs. When a value has multiple targets, add one spring per change so it remains a pure function of time.
3. Liquid glass: each glass element keeps its own clone of the scene behind it. Filter that clone with an SVG feImage displacement map (a rounded-rect distance field) through three feDisplacementMaps at slightly different scales for chromatic edges, then add a rim light. For glass letters, use a canvas distance field per glyph to create the map, mask and highlights.
4. Goo: apply blur and an alpha threshold, then composite the source on top so the glass stays sharp inside.
5. Iris: place 6 blades around a hexagonal aperture. Each blade uses its two vertices, both edge extensions and the SHORT arc between them.
6. Wordmark squeeze: move every letter toward the dot by the same factor and adjust its drawn width to match (narrow the wdth axis, scale the rest), keeping the letters in contact.
7. Footage: re-encode it all-intra (ffmpeg -g 1), load it as a blob URL and await "seeked" before drawing each frame.
8. Sound: use a downloaded SFX for every event (Mixkit), never synthesized, and position each by its measured peak. Start the song on a downbeat: land the zoom on the drop, place the wall in the breakdown and bring back the wordmark with the beat. Loudnorm to -14 LUFS.
9. Render with Playwright: blend 4 subframes per frame using ffmpeg tmix at 60fps. Check one frame per beat, then look for single-frame pops (frame-difference spikes 3x higher than their neighbours).
backdrop-filter: url() misreads displacement maps in Chromium, so clone the scene instead. A flood needs to extend beyond the corners and take about 0.3s; otherwise, half the screen changes in a single frame. A child set to visibility: visible can show through a hidden parent, so use inherit. Text swapping inside a morphing shape needs its own mask. python http.server can't range-seek video, so use the blob URL.
Ask me for the inputs, then show me the beat map and 4 stills (open, glass, stage, wall) before writing the full film.
Palu retweeted
"visualize the biggest scam in humanity, no words allowed"
opus 5.5 came back with this β
Someone turned the design language of 2,000+ of the worldβs best products into DESIGN.md files that Codex / Claude Code can read directly.
Each file includes:
β Colors and tokens
β Typefaces, sizes, and weights
β Spacing and page layouts
β Component styling
β Clear rules for what to use and what to avoid
Linear, Notion, ElevenLabs, and Lamborghini are all included.
Pick one that fits your product, drop it into the project, and have your agent build a new UI or align the existing one with those rules.
The library is currently free to use:
styles.refero.design/
Opus 5.5 is already strong at frontend. Give it ππ΅π²ππ² π΄ ππΆππ²π and it feels like cheating: 2,000+ design styles from real product sites, 12,000+ components and templates, and 153 motion effects that come with prompts. You can feed all of it straight to Opus.
Sorted by ππ΅π²πΏπ² ππΌπ π΄π²π ππππ°πΈ π
No idea what style to go for
β Refero Styles: each product site's colors, typography and spacing, written up as a DESIGN.md for AI to read. Pick one, drop it into your project and have Opus follow it
β awesome-design-md: a GitHub collection of DESIGN.md files for 74 brands, with 118k stars. Open source under MIT
Components look rough
β 21st.dev: React and Tailwind components and templates. Connect its MCP and Claude Code can search it on its own. Copying and installing has a free usage limit
β Component Gallery: look up any component and see how 95 design systems handle it
Motion feels flat
β Kinetics: spring-physics animations. For each one you can copy the CSS, the React, or a ready-made AI prompt
Need a demo video
β whatships: 2,000+ product launch videos. Pick one in your category, send it over, and have Opus tile its frames into one image and match it
β HyperFrames: Claude Code writes the video in HTML, and HyperFrames renders it to MP4
Done, but something still feels off
β Impeccable: a set of design commands you install in Claude Code. bolder, distill and polish turn "make it look better" into specific changes
Send this to Claude Code so it remembers the list π
"Add a section called Frontend references to ~/.claude/CLAUDE.md. Use it only when building a new page, when I say something looks bad, or when I name one of these sites. For small changes, just do the work:
- Style: pick a DESIGN.md that fits the product from styles.refero.design or VoltAgent/awesome-design-md on GitHub. Put it in the project root and add an @ import for it in the project's CLAUDE.md, so from then on everything follows its colors, typography and spacing.
- Components: check 21st.dev first, and call its MCP directly if it's installed. It has a free usage limit, so tell me what you're looking for before you call it. Then check component.gallery to see how mature design systems handle the same component.
- Motion: get a ready-made prompt or React code from kinetics.colorion.co.
- Demo videos: I'll pick reference videos on whatships.com and send them to you. Tile the frames into one image to see the pacing and transitions, then build it with HyperFrames (hyperframes.dev).
- If it still feels off when it's done: run it through polish and distill from Impeccable (impeccable.style).
The project's existing design system and components come first. Outside references only fill in what hasn't been decided yet. If an MCP, skill or command-line tool you need isn't installed, ask me whether to install it, and don't imitate it yourself. If you can't read a page's actual content, stop and ask me to paste it in. Don't fill anything in from memory. Every time you use an outside reference, tell me which one and what you changed. Show me what you'll add first, and don't write it until I confirm."
Someone turned the design language of 2,000+ of the worldβs best products into DESIGN.md files that Codex / Claude Code can read directly.
Each file includes:
β Colors and tokens
β Typefaces, sizes, and weights
β Spacing and page layouts
β Component styling
β Clear rules for what to use and what to avoid
Linear, Notion, ElevenLabs, and Lamborghini are all included.
Pick one that fits your product, drop it into the project, and have your agent build a new UI or align the existing one with those rules.
The library is currently free to use:
styles.refero.design/
TASK: frame-locked 1:1 remake of REF=[path/to/reference.mp4]. Swap ONLY: brand nameβ[NAME], logoβ[path/to/logo.png|svg], paletteβ[hex list or "derive from logo"], platform UIβ[e.g. LinkedInβX], faces/screensβ[source]. All else identical: layout, sizes, positions, timing, easing, camera, cuts, blur, cursor path, typing cadence, audio beat grid.
ACCEPTANCE: REF and remake stacked + frame-locked β same pose every frame. Cuts 0 frames off. Position/size error β€1% of frame. Typing char-count, cursor, camera on the same frames. Differences allowed only where the swap forces them (word widths); report each.
PHASE 0 β ANALYSIS (no building yet)
- ffprobe fps/res/duration. Extract ALL frames 0-based to ref/full/fNNNN.jpg + audio to ref/audio.wav. 1fps contact sheets for overview.
- Detect cuts via per-frame mean-abs-diff spikes; confirm visually. Write SPEC.md: shot table (id, f0βf1, content, transition in/out), per-shot detail, component inventory, verbatim on-screen text in order, colour tokens sampled from pixels, font sizes from cap height, cursor paths (tip xy per frame), camera keyframes.
- MEASURE with numpy on ref frames (ink bounding boxes, flood fills, template match), never eyeball. Store per-frame sample arrays for every major move; the spec's prose is a guide, ref/full is truth.
- Audio: STT with word timestamps (narration table: line, start, end, text). Music BPM + beat phase (onset autocorrelation), drop time, loudness arc per section, hard-stop time. SFX hit times from onset/spectral analysis. Voice pitch/wpm.
PHASE 1 β ENGINE (you, before any agents)
- Single HTML page, stage at REF native resolution. window.seek(t) renders frame F=t*fps as a PURE function of F: no timers, no Date, no Math.random (seeded hash), no CSS transitions/animations. Shots register SHOT({id,f0,f1,render(lf,F)}) returning HTML; seek routes F to its shot. window.ready=true only after document.fonts loaded + all images decoded.
- core.js shared helpers: easing set, kf(F,keys,ease), samples(F,f0,arr) interpolating measured arrays, camera(inner,scale,tx,ty,origin,blur), directional motion blur (SVG feGaussianBlur), cursor (glyph MEASURED from ref, press-scale curve measured), ripple, text reveal, brand tokens, logo component (mask-image of logo so any fill/gradient works), avatar/persona pickers.
- Palette swap as ONE deterministic filter applied to each rendered HTML string (hex/rgb/rgba re-hued by band, lightness preserved) so hard-coded colours in shots can't leak the old brand. Bitmaps untouched.
- render.mjs (Playwright Chromium, deviceScaleFactor 1): modes stills | compare (ref left | ours right + labelled sheet) | full . Per-agent OUT dirs so parallel runs don't collide. Print page errors.
- sync.py: REF over remake stacked, frame-locked mp4. encode.py: PNG frames β h264 at REF fps, mux audio.
PHASE 2 β PARALLEL BUILD
- Split shots into 4 contiguous groups; one agent each; each writes ONLY shots/.js (IIFE, helpers prefixed _), never edits core.js (request changes from you). Give each: BRIEF.md (acceptance bar, swap rules, file rules, verify loop), SPEC.md sections, core.js API.
- Verify loop per shot: compare first/last frame, every keyframe, 2 frames into each transition; iterate until within tolerance. β€15 frames per render call; one render at a time per agent.
- Report table: shot | frames | MATCHES/CLOSE/ROUGH | residual diff | frames that would drift when stacked | spec errors found vs ref.
- 5th agent = audio: royalty-free commercial-OK track (record URL+licence), time-stretch to REF BPM, cut on beats so drop/breaks/silences land on REF times; SFX synthesized (numpy) on REF hit times; mix script with VO slot table (REF start times, text swapped) that fits each line (β€8% stretch), ducks music under voice, loudness-normalizes (~-14 LUFS, TP β€ -1). Never reuse REF music or voice.
PHASE 3 β INTEGRATE
- Unify shared glyphs/components across groups (cursor, DM window, logo) β one definition, everyone uses it.
- Full render all frames (2β3 parallel chunks max), encode with mix, run sync.py, build 1-per-second ref|ours contact sheet + sheets at every group seam (last 2 / first 2 frames). Inspect. Fix drift. Re-render. Scan all frames for leftover old-brand colour.
- VO: generate per line (any TTS), trim silence, place on slots, re-mix.
PITFALLS (seen in practice)
- Measure text widths only AFTER fonts load (measureText before load caches fallback widths β words collide). Lazy-init any width tables.
- Word-length changes from the brand swap shift centred layouts; keep shared elements on ref positions, absorb the delta in the swapped word, report it.
- Randomised bursts/particles won't match stacked; drive the visible ones from measured tracks.
- Ref whips may be crisp (no blur); check before adding motion blur.
- Headless Chromium may need to run outside any OS sandbox; use swiftshader/ANGLE if WebGL is involved.
- Never claim a shot matches without viewing ref|ours side by side for it.
DELIVER: remake.mp4 (with audio), sync-check.mp4, SPEC.md, source, and a list of every remaining difference vs REF with frame numbers.
Palu retweeted
Even Opus 5.5 can't help you without cool references.
modulify.ai/templates β Templates
wedoflow.com β 1,300+ website pages
landdding.com β Visual references
motionin.design β Motion Sections
Palu retweeted
everyone's sharing motion graphic videos that Opus 5.5 made, and it's genuinely insane
everyone says they created it with "one prompt", but my one prompt video looked mid
so i went through a bunch of these videos to see how they were actually made, and found the workflow that works
here's how to generate pro level motion graphic videos w/ opus:
1. get reference videos to direct from -> whatships.com
pick 1-2 videos whose style you want and tell opus to match them. naming a style works way better than describing one
without a reference, opus falls back to its default look: centered text, gradient background, everything fading in
that's why so many of these videos look the same. a reference gives it the pacing, the type and the transitions to copy
2. install @HyperFrames_ or @Remotion so opus can build the video
both let opus write every scene as code and render it straight to mp4. no video editor
without one, opus can only describe a video or hand you a rough html page you have to screen record
with it, every frame is exact, and when you ask for a change it edits one line and re-renders instead of starting over
3. install @21st_dev for high quality components in the video
real buttons, cards and UI components made by design engineers, instead of whatever opus invents on the spot
without it, opus draws your product UI from scratch and it looks off. wrong spacing, placeholder boxes, fake-looking buttons
anyone who's used good software can feel it in a second, and the whole video reads as cheap
4. steps 1-3 were context + setup. now dump all of it into opus
your brand (logo, colors, fonts), screenshots of your real product, the reference video, and a quick braindump of how you see the video
then ask for 3 storyboard variants
without this, opus guesses your colors, your font and what your product even does. the video could be for any startup
with it, it could only be yours. and 3 variants means you pick a direction instead of fixing the first idea it had
5. pick the storyboard you like
ask for one still frame per scene before anything moves. fixing a storyboard is way cheaper than fixing a render
without this step, you only find out scene 4 is wrong after the whole thing is animated, and every fix means re-rendering. a still frame takes seconds to change
6. let claude cook
then give notes like a director: "slow every zoom to 0.7x", "hard cut here", "push in on the button"
without notes, the first render is usually 80% there, and that last 20% is what makes it look pro. vague notes like "make it better" get random changes. camera words get exactly the change you want
everyone has the same model. the context you give it is what makes it look pro
let it cooookk
Palu retweeted
I asked Claude Opus 5.5 to make an animation comparing Astra to Opus 5.5.
Claude mogged so hard.
Set up my AI agent to report Pokemon scalpers to the IRS for tax fraud.
I get 30% of all proceeds collected by the IRS for my report, hereβs how it works:
- Scrapes facebook marketplace for sealed ETBs and negotiates a cash deal with sellers
- It has to be cash, as the scalper is extremely likely to not be paying taxes on this income
- Automatically records name on account, location, and all other information
- Fills IRS Form 211, Application for Award for Original Information
- Files and monitors email for requests for more info & payouts.
I expected to collect over $500,000 this year⦠and pay taxes on every cent made.
I'm sorry, when did Death Battles get THIS good??? The animation is not only so clean but the effects are like...pretty innovative too. The sketchy line art overlay clash part is insanity
Palu retweeted
βWHY ARE YOU IN V-TRIGGER???β
BrianF vs iDom FT100
Currently at 65-84
twitch.tv/brian_f
