About
A canteen for the creatively hungry.
The AI Canteen is where cutting-edge AI meets editorial craft — a place to taste what's possible and leave knowing how it was made.
Chioma Onwumere
Founder & Creative Director
The Philosophy
The AI Canteen began with a simple conviction: that generative tools, in the right hands, could produce work with real soul — cinematic, editorial, and unmistakably intentional.
Founded by Chioma Onwumere, the studio blends the instincts of a director with the fluency of a prompt engineer. The result is a body of work — films, photography, and characters — that feels handcrafted, even when a machine held the brush.
But a canteen is meant to be shared. Alongside the work, we publish the prompts, the frameworks, and the hard-won lessons — so you can pull up a chair and make something remarkable of your own.
Come hungry, leave smarter. That isn’t a tagline — it’s the whole idea.
Craft over volume
We would rather ship one unforgettable frame than a hundred forgettable ones. Every piece is curated, not churned.
AI as a brush
The tools are extraordinary, but taste is still the differentiator. We direct the machine the way a filmmaker directs a set.
Teach as we go
Everything we learn becomes a prompt, a tip, or a resource. The canteen is a table, not a gate.
20+
Personal projects shipped
0→13K
TikTok followers in 3 months
13
Tools in the creative stack
9
Photography disciplines
Creative Stack
The tools behind the work
A working toolkit refined across 20+ personal projects — with the exact job each tool does best.
AI Assistants
Thinking partners for scripts, research, and prompt refinement — each with the model it runs on today.
ChatGPT
GPT-5.5The studio’s default writing partner. GPT-5.5 reasons carefully, follows a brief closely, and handles scripts, structure, and prompt refinement — with native image generation, voice mode, and custom GPTs for workflows you run again and again.
- Scriptwriting & story beats
- Prompt engineering & refinement
- Concept brainstorming
- Tightening & rewriting copy
Best for: Scripts & prompt craftVisit
Claude
Claude Opus 4.8Anthropic’s assistant, prized for natural, human writing and its faithfulness to a detailed brief. The Claude 5 family — led by Opus 4.8 — excels at long-form structure, nuanced tone, and careful editing, for copy that reads written, not generated.
- Nuanced scriptwriting & tone
- Long-form structure & editing
- Following complex briefs
- Polishing copy until it sings
Best for: Writing & toneVisit
Grok
Grok 4.3Built into X with a live pulse on what’s trending right now. Fast, current, and candid — ideal for catching the cultural moment, drafting scroll-stopping hooks, and pressure-testing ideas, now with native video input and a 1M-token context.
- Trend-aware ideation
- Real-time cultural references
- Punchy hooks & captions
- Fast first drafts
Best for: Trend-aware ideationVisit
Gemini
Gemini 3.5Google’s multimodal flagship with a huge 1M-token context and native understanding of images, video, and audio. Best for deep research, analysing reference material, and Search-grounded fact-checking — and it plugs straight into Google’s creative stack.
- Long-context research
- Analysing image & video references
- Search-grounded fact-checking
- Multimodal reasoning
Best for: Research & analysisVisit
Image
Where a still frame becomes a signature look.
GPT Image
GPT Image 2OpenAI’s reasoning-powered image model. It plans a shot before it draws it — near-perfect in-image text, reliable composition, and up to 4K output.
- Text-in-image & logos
- Accurate, on-brief composition
- Fast, editable iterations
Best for: Accurate compositionVisit
Midjourney
V8.1The mood-maker. Painterly, art-directed aesthetics with gorgeous light — now faster, with HD 2K renders and sharper small-detail retention.
- Cinematic mood & lighting
- Painterly, art-directed looks
- HD 2K hero frames
Best for: Cinematic beautyVisit
Flux
FLUX.2Black Forest Labs’ photoreal workhorse. Best-in-class realism for people, products, and editorial covers, with dependable multi-reference character consistency.
- Photoreal humans & products
- Consistent character work
- Editorial covers with text
Best for: Realism & consistencyVisit
Video
Motion, camera, and performance — generated.
Kling
Kling 3.0Kuaishou’s cinematic model with native 4K, up to 15-second clips, and an AI Director that composes multiple shots in one generation while holding continuity.
- Human movement & physics
- Native 4K, up to 15s
- Multi-shot AI Director
Best for: Human movementVisit
Runway
Gen-4.5The director’s toolkit. Top-rated fidelity with consistent characters and locations across scenes, native audio, and precise creative control.
- Director-grade control
- Consistent characters & scenes
- Native audio tracks
Best for: Creative controlVisit
Veo
Veo 3.1Google DeepMind’s premium model. High-fidelity, coherent cinematic shots with synchronised dialogue and sound, up to 4K.
- Premium cinematic shots
- Synced dialogue & sound
- Up to 4K delivery
Best for: Premium film looksVisit
Seedance
Seedance 2.0ByteDance’s expressive motion model. Strong on stylised movement and performance, and able to blend many image, video, and audio references in a single pass.
- Expressive, stylised motion
- Multi-reference generation
- Dynamic performance
Best for: Dynamic performanceVisit
Audio
Voice and sound that carry the story.
ElevenLabs
Eleven v3The most expressive voice model. Natural voiceover, narration, and character voices in 70+ languages, with emotion, direction, and multi-speaker dialogue via inline audio tags.
- Natural voiceover & narration
- Emotion via audio tags
- Multi-speaker dialogue
Best for: Voice & narrationVisit
Editing
Where everything is assembled and finished.
CapCut
Fast, capable editing where it all comes together — assembly, captions, sound, and export tuned for every platform.
- Editing & assembly
- Captions & subtitles
- Per-platform delivery
Best for: Edit & publishVisit