Paper Light (紙の灯り · Kami no Akari)

Inspiration

“Paper Light” was born from an old idea sketched in a notebook, a small draft I always wanted to transform into something visual. I was inspired by memory, light, and the invisible bonds that remain even when people are no longer with us. I was also inspired by the excitement of exploring generative filmmaking and discovering how artificial intelligence can expand what we imagine.

What it does

The project blends AI-generated imagery, poetic visual storytelling, and human editing to create a short, emotional piece about memory and the lingering presence of love.

How I built it

I began by writing the story and drawing a simple storyboard on loose sheets of paper. Then I developed the graphic style and color palette using Midjourney. I crafted detailed prompts to generate images using Nano Banana (Google Gemini) and Midjourney, spending hours testing, refining, and rebuilding scenes through trial and error. I used Flow (Veo 3.1) and Midjourney for video creation, shaping camera movements, inserting elements, and giving life to each moment. Finally, I edited the video, adjusted timing, color, natural sound, and silence to give the film an intimate and emotional atmosphere.

Tools used: • Style and colorimetry: Midjourney • All images: Nano Banana (Google Gemini) and Midjourney • All videos: Flow (Veo 3.1 fast) and Midjourney • Image editing: Adobe Photoshop • Translations: ChatGPT • Video editing: InShot (mobile app) • Voices: Flow (Veo 3.1 fast) • Audio: – Veo 3.1 (sound effects & ambience) – free-stock-music.com (main track) – Pixabay (1 sound effect) – ElevenLabs (1 sound effect + 2 ambience tracks)

Challenges I ran into

One of the biggest challenges was maintaining visual consistency across all the AI-generated footage. Keeping the characters’ faces, their clothing, the color palette of the scenes, and the overall atmosphere aligned so everything felt like part of the same world was a constant battle. As a beginner in video creation, coming from a background focused solely on still image design, this was something I had to learn through trial and error.

And honestly, the biggest challenge "maintaining consistency" is still something I haven’t fully mastered and I know that with more experience I’ll be able to refine and control those details better. I’m aware that this entire process is part of learning something completely new. Every attempt teaches me something valuable.

Another ongoing challenge was the inherent unpredictability of AI. Nothing ever comes out perfect on the first try. Every step required iteration, adjustments, regenerations, and, above all, patience.

Accomplishments that I'm proud of

• Creating a warm and poetic visual language using only generative tools and manual editing. • Transforming a forgotten draft into a cohesive and emotional cinematic piece. • Discovering new creative skills through experimentation with AI tools.

What I learned

I learned that artificial intelligence doesn’t replace us, it expands what we can create. I learned to improvise, solve unexpected problems, and turn mistakes into creative opportunities. I also gained a deeper understanding of visual consistency, generative direction, and the narrative possibilities of AI.

What's next for Paper Light (紙の灯り · Kami no Akari)

I want to continue refining the project, create an extended version, and explore new techniques to enhance visual and emotional coherence. My goal is to keep using AI as a creative partner for larger, deeper, and more experimental projects.

Built With

• adobephotoshop
• chatgpt
• elevenlabs
• flow
• free-stock-music.com
• googlegemini
• inshot
• midjourney
• nanobanana
• pixabay
• veo3.1