GiveMePrompts

// SELECTED WORK

THE WORK, UNEDITED.

Every piece below is labelled for what it is. Concept pieces are concept. Each one lists what is strong and what is still weak, because a hostile review would find the weak spots anyway.

Motion

Films drawn in code. Cut to the audio, checked frame by frame.

Concept film. Fictional story.

55 s / 1080p / brushstroke film, concept

Origin: a brand as a character

A paint-blob hero drops in, learns to type, faceplants, picks up four skills and walks on stage to ask for a name. Every stroke is drawn in code and cut to the beat.

Watch it render (fast mode)Case study on Contra

What is strong

  • The story reads with the sound off: prompt, drop, born, overreach, faceplant, getting up, four new skills, the rush, arrival.
  • Cuts, badges and the logo hit land on the beats of the track.
  • One colour per chapter, so each scene is recognisable from a thumbnail.

What is still weak

  • The hero is closer to a sticker than to oil paint: strokes overlap, they do not blend wet-in-wet.
  • The opening five seconds are the quietest stretch, and 24 to 29 s is sparse on purpose.
  • Small text lines are tiny on a phone.

Built with JavaScript canvas and WebGL, FFmpeg

Concept film. Fictional product. Sample data.

30 s, 1080p, concept

Plumbline: concept SaaS launch film

A made-up product, a real launch film. Cards land on measured snare hits and every number on screen is labelled sample data.

Watch it renderCase study on Contra

What is strong

  • One idea carries it: a plumb bob opens the film, the word "drift" slides off the baseline, and the bob lands true on the end card.
  • Motion is locked to the audio. Cards land on measured snare hits and the strongest hit racks focus onto "Which one is true?".
  • Real motion blur, built from 4 to 32 sub-frames per frame and blended in linear light.
  • Checked on the exported file: 900 frames, no frozen runs, no black dips, one contact line on screen.

What is still weak

  • The interface text is small. On a phone the app reads as texture and the captions do the explaining.
  • It is CSS 3D, not a true 3D render, so it looks flatter than a studio-rendered launch film.
  • The soundtrack thins out after about 27 seconds and there is no voiceover.

Built with HTML, CSS and SVG motion, FFmpeg, Python

Concept film. Original alphabets.

30 s, custom letterforms

Kinetic alphabet

Four custom alphabets, every letter drawn in code with no font files. Each scene changes the alphabet and the colour together.

Watch it renderCase study on Contra

What is strong

  • Four custom alphabets, one per scene, all drawn in code with no font files: stencil bars and rings, a pressure-sensitive pen alphabet, Bezier wireframe, pixel cells, and the closing lockup.
  • The idea is on screen the whole time. The readouts are live numbers from the render, so "every frame is code" is shown, not claimed.
  • Checked on the exported file: 900 frames, no frozen runs, no black dips before the closing fade.

What is still weak

  • The small readout text is tiny on a phone.
  • The music is a cut from an existing instrumental, not scored to this edit beat for beat.
  • It was reviewed from sampled frames, not frame by frame. A full second pass is still planned.

Built with JavaScript canvas and WebGL, FFmpeg

Interactive

Pages that react. Open them and play.

Real-time simulation

12 s clip / live simulation

Technicolor ink in black water

A fluid solver running in the browser: ink dropped into black water, mixing the way pigment does. The same engine runs live in the Lab, where you can stir it.

What is strong

  • A real stable-fluids solver on the GPU, not a looped video.
  • Hue-preserving shading keeps the colours rich instead of washing to white.

What is still weak

  • The 12 second clip is a capture; the live version in the Lab is the real thing.
  • Heavy on older graphics hardware.

Built with WebGL2, JavaScript

Brand systems

Identity built from geometry, then put to work.

Concept brand. Fictional.

30 s / 1080p / fictional brand, concept

Quoin: a brand identity in motion

A fictional identity built from a grid: the mark constructed from circles, a four-colour palette, a custom wordmark and pattern, then a card, a tote and an app screen.

Watch it renderCase study on Contra

What is strong

  • The mark is constructed on screen from circles and tangents, so the logic is visible.
  • One system carries from the mark to the card, the tote and the phone.
  • Checked on the exported file: no frozen runs and a complete end card.

What is still weak

  • It is a concept brand with no real client behind it.
  • Small interface text on the phone scene is tiny on a mobile screen.
  • The final wipe timing was still being tuned when the build was cut.

Built with JavaScript, FFmpeg, Python

3 concept brands / 14 boards / 1 working app

Brand kits gallery

Terza, Moonwort and Kilnwood: logo boards, colour and type guides, packaging, and a fintech app you can tap through. Every board drawn in code.

Open the gallery

WANT ONE LIKE THIS?

Open the LabServices and pricesadmin@givemeprompts.com