Published AI filmmaking methodology and produced five films my kids can watch

Achievements:

- Authored and published the CAVAS + CHAPS Methodology Specification v1.0 (August 2026): CAVAS (Cinematic Audio and Visual Architecture System) for cinematic architecture, and CHAPS (Cultural & Human Authenticity Prompting System) for dignified, stereotype-resistant representation.

- Produced five productions across five distinct registers: hybrid live-action emotion, Hollywood trailer grammar, stylized anime worlds, documented cultural authenticity, and conversational avatar realism, all under one architecture.

- Hybrid filmmaking milestone: composed and performed the piano score for "Phoenix" myself, with AI character sheets holding avatar consistency across a denial → anger → grief arc. Three weeks later, my son taught himself the melody by ear from watching me play. The human choice multiplied itself, and you can't imagine how I felt as a father to see this happen.

- Selected for Creative Partner Programs by ImagineArt and OpenArt. Two AI generation companies independently validating the craft. Competitors to Higgsfield AI.

- Foundation pre-film (my 7-minute AI film) submitted to the Toronto Cinema Shift Festival; premiere summer 2026.

Tools:

Image: GPT Image 2.0, Midjourney, Flux, Nano Banana, ImagineArt, OpenArt.

Video: Higgsfield, Kling, Veo, ImagineArt, OpenArt

Voice & Audio: ElevenLabs, HeyGen, my own piano, room-tone engineering.

Music: Udio.

Assembly & finishing: Adobe Premiere Pro (the human cut is where the film is made).

Direction & architecture: Claude, ChatGPT, Gemini, Perplexity.

Knowledge and Context: Cursor CLI and Obsidian based LLM 2nd Brain, Airtable

Methodology:

Last cycle, I showed you Extended Continuity, gatekeepers speaking Arabic and Swahili in sequences that broke AI's 5-second barrier. This cycle, that discipline matured into a full published architecture.

CAVAS designs the scene as cinema: narrative intent, emotional architecture mapped over time, camera intention, visual and audio continuity, pacing and silence, multi-tool orchestration, and mandatory human debug. Core rule: the camera commits when the story needs it to commit.

CHAPS designs the scene as human reality, through eight principles including Context Before Aesthetic, Specificity Prevents Stereotypes, Representation Requires Restraint, and AI Must Not Flatten Humanity.

Every finished scene must answer two questions:

Does this feel directed — or generated? Does this feel observed — or stereotyped?

Five clips, each proving one load-bearing principle:

1. Phoenix (Grief Never Goes Away, But This Helped) → Silence Is Narrative + hybrid craft. My piano as the score; the scream faded to silence on a freeze-frame; slower moments allowed to build.

2. That Will Never Work → The Camera Commits. Full Hollywood trailer grammar built to a real industry brief in a four-day window.

3. Foundation: Heroes → Stylized worlds. The architecture holding beyond photorealism, in anime language.

4. Jamaican Dominoes → CHAPS Test Case 01. Cultural verification before generation; specificity over stereotype; symbolic precision (the domino pips) handled as craft, not left to the generator.

5. Chess Conversation → Human Debug at the audio layer. An ElevenLabs voice swap muted the world; I rebuilt continuity with room tone from my own footage. The fingerprint is in what you hear.

Learnings:

- That Will Never Work didn't place in its contest. I judged it by process, and the trailer grammar it forged shows up in everything after. That's the discipline.

- I've been through hardship the past few years mainly from losses in the family. So I used expression through generative media content creation as an outlet to share my stories and feelings more. But more importantly, my goal is to create pieces that my children (9 and 7yrs old) can enjoy and be inspired to find their creative and building paths as well.

- AI gets the work to 70–80%. The remaining 20%, my live piano composition, the room tone, the cut, the cultural verification, is where the human artistry lives with AI's technologies. A lot of pre-work research and planning was also required before just rushing forward to do experimental image and video generation. And end-to-end strategy must be in place with clarity on what you are trying to solve, achieve or convey to the audience.

- The standard hasn't changed since my first win; the architecture to govern and foster it has: AI should disappear behind the artistry.

Relevant URLs:

- Phoenix (Denial/Anger/Grief): https://youtu.be/2Vjnj-7T5Dc

- That Will Never Work trailer: https://youtu.be/DkeGKxN7FUY

- Foundation: Heroes (anime): https://www.youtube.com/shorts/G4rmauyGtDk

- Jamaican Dominoes (CHAPS Test Case 01): https://www.youtube.com/watch?v=zT82bDgZdUA

- Chess Conversation: https://www.youtube.com/shorts/BJ2kuf94rWs

- LinkedIn: https://www.linkedin.com/in/karim-gallimore-1933272/