WierszoNutki: Kotek (Kitten)

A lonely kitten's simple wish unfolds into a cinematic dream, exploring a world where rivers flow with milk and sadness melts into sweet slumber 🐱🥛🌙

Comment

Inspiration

The "Wierszonutki" project was born from a personal need: to provide my young son with valuable, high-quality content, which is often difficult to find on Polish YouTube. This motivation grew into a larger mission, and to date, I have created over 30 animated musical adaptations of classic Polish children's poetry.

My journey into Kotek (Kitten) began with Julian Tuwim's beloved poem from around 1955. My goal was to create a musical adaptation that would serve as a calming, lullaby-like experience for children, while still offering deep emotional resonance. I aimed to explore the seamless transition between the kitten's waking reality and its dream world, where a simple wish for milk transforms into a fantastical journey.

A core artistic principle was to defy the trend of hyper-fast cuts common in children's media. Instead, I focused on creating long, immersive takes, allowing young viewers the space to absorb the atmosphere and connect emotionally with the kitten's journey.

What it does

The film presents the emotional journey of a small kitten, moving from a state of loneliness and longing to a world of fantastical dreams, and finally to a state of contentment and peace. It's a visual narrative that follows a classic three-act structure: the initial problem (an empty bowl), the transformative journey (a dream of a milky river), and the resolution (the wish fulfilled). It aims to be a gentle, soothing visual poem that both children and adults can connect with.

How I built it

The creative process began by transforming Tuwim's poem into a modern, dreamy song using Suno AI. To find the perfect tone that was both ethereal and engaging, I generated approximately 250 variations before crafting the final one. This musical foundation became the blueprint for the entire animation.

The core visual idea revolved around creating long, uninterrupted shots. To achieve this, I employed the 'frame-chaining' technique - manually connecting the chosen frame of one generated clip (or generated image) to the start/end frame of the next. While this is a known method in the AI community, its effective application requires significant iteration and artistic judgment to create truly seamless sequences that serve the story. This allowed me to push beyond the typical 5-10 second limits of AI video models.

While Hailuo-02 served as the base model for its excellent frame-blending capabilities, I extensively used the fal.ai platform to experiment with dozens of other models. Kling 2.1 Pro and Seedance 1.0 Pro were particularly vital for challenging scenes that required specific transitions, actions or aesthetics. Additionally, Google's Nano Banana proved invaluable for generating the final, target keyframes, allowing me to precisely achieve the intended composition to conclude a scene. This multi-model approach allowed me to select the best tool for each specific narrative beat.

My entire creative workflow, from script concepts to scene-by-scene analysis, was managed within Google Gemini 2.5 Pro. Its multimodal ability to "see and hear" my video drafts made it an invaluable AI creative partner, helping me iterate and refine the story at every stage.

Finally, subtle SFX were crafted using ElevenLabs to add a layer of realism and warmth - the kitten's meows and purrs, the gentle lapping of milk, and the sound of the bowl being filled. These sounds had to be understated to enhance immersion without breaking the dream-like atmosphere.

Challenges we ran into

  1. Animating the act of drinking: This simple action proved incredibly difficult for most AI models, which often produced artifacts or unnatural movements. After testing numerous models on fal.ai, I discovered that Seedance 1.0 Pro handled this action best, though it came with a trade-off in character consistency, requiring a careful balance in the final edit.

  2. Seamlessly Blending Scenes: The frame-chaining technique often introduced slight color shifts or tempo changes at the connection point. However, I learned to turn this limitation into a narrative tool. By strategically placing these "cuts" at moments of musical or emotional transition (e.g., when the music swells into the "dream river" sequence), the technical flaw became an asset that served the story, enhancing the shift from reality to dream.

  3. Maintaining Asset Quality: Generating long sequences through frame-chaining led to a gradual loss of image quality. I learned the critical importance of starting with 4K keyframes and carefully selecting sharp, detailed frames to begin new segments. This required a constant balance between the desire for long takes and the need to maintain a high, cinematic quality standard.

Accomplishments that we're proud of

  • Reimagining a beloved Polish classic poem, known to generations of parents and grandparents, and introducing it to a new generation in a modern, gentle, and universally understandable format.
  • Creating a calming, cinematic experience that stands in contrast to the overstimulating, fast-paced content typically aimed at children.
  • Skillfully applying advanced AI techniques to create a seamless, cinematic narrative that pushes beyond the typical limitations of AI video, resulting in a cohesive and emotionally resonant film.

What I learned

  • Story Cohesion Over Visual Spectacle: I learned that especially in children's content, a clear and emotionally resonant narrative is more powerful than visual fireworks. I discarded dozens of technically spectacular but narratively inconsistent scenes in favor of simpler shots that better served the story's emotional core.
  • The Art of Imperfection: While striving for technical perfection, I learned that the ultimate goal is immersion. A minor technical flaw that doesn't break the viewer's experience is acceptable. The most important thing is to maintain the emotional connection, not to scrutinize every single pixel.

What's next for Wierszonutki: Kotek (Kitten)

The Wierszonutki project will continue its mission of adapting classic poetry for children into high-quality, thoughtful animated content. I'm currently exploring new aesthetics, such as a papercraft style and stop-motion-inspired animation, paired with swing and jazz music. The core philosophy remains the same: to captivate young minds with rich storytelling and gentle visuals, proving that children's content can be both engaging and calming.

Note:

  • The poem is in the public domain, so it can be freely adapted and creatively reimagined in this film.
  • Turn on CC subtitles to follow the full text of the poem.

Built With

  • elevenlabs-sfx
  • fal-ai
  • gemini-2.5-pro
  • hailuo-02
  • kling-2.1-master
  • kling-2.1-pro
  • midjourney-v7
  • nano-banana
  • seedance-2.1-lite
  • seedance-2.1-pro
  • suno-v4.5+
Share this project:

Updates