Description
A Gift for the Ten-Year-Old You is an AI short film told in the visual language of Japanese coming-of-age cinema. After a fight, 27-year-old Saki leaves home alone and, on an old street, runs into ten-year-old Yuma: the childhood self of the partner who has just argued with her and walked out. At first she simply follows this familiar stranger in disbelief, then slowly realizes that this is one of the most important afternoons of his life. He has always believed his mother would be back before six, and in the end he watches that train pull away.
Saki cannot change what has already happened, and she cannot bring his mother back for him. The gift she can give him is not an object. It is staying with him through that afternoon, when he is loneliest, and telling him what no one told him as a child: "Your mother didn't leave because you did something wrong." Seventeen years later, the boy who was left waiting has grown into a man who walks away from conflict, and Saki, who has always been tired of waiting, sees for the first time where that comes from. Understanding is not forgiveness, and the ending offers no easy reconciliation. The two simply return to the same home and, at six in the evening, say "okaeri" (welcome home) and "tadaima" (I'm home) at the same moment.
Summer streets, a local railway, a station and a riverbank build the remembered world of 2009, set against a modern apartment in contrasts of warm and cool, space and rhythm. The sound design leaves room between ambient sound, pauses in dialogue and the piano score, so cicadas, wind, railway warning bells, footsteps and breathing all take part in the storytelling.
TapNow Creative OS helped with shot planning, scene breakdowns and managing the generation workflow, and Tripo H3 was used to build 3D layouts of some scenes as a reference for shot design. Most of the moving images were generated with Seedance, with image tools such as Midjourney 8.2, GPT Image 2.5 and Seedream 5.0 Pro used to set up character, scene and visual references, and unified settings for characters, scenes, period and voices kept the shots continuous. To counter the spatial jumps, broken character states and inconsistent performances common in AI video, the author repeatedly rebuilt the characters' relationships and the story's cause and effect through 3D previz, locked reference assets, "in-between state" shots, sound bridges, match cuts and editing.
For the author, the gift the film really wants to give is a belated confirmation: some people leave not because you weren't good enough, and those who were left behind can still learn to come home. It is for ten-year-old Yuma: an answer, and some company, 17 years late.
How Tripo was used
Tripo served as a previsualization tool for this film. The author used Tripo H3 to block out some key scenes as simplified 3D spaces, a kind of "spatial sketch" within the AI filmmaking workflow.
- Spatial checks: confirming the layout of rooms, streets and other settings, along with where characters stand, which way shots face and where the camera sits.
- Feeding generation: those spatial relationships were carried into shot planning in TapNow Creative OS and video generation in Seedance, reducing the spatial jumps, camera confusion and position breaks common in multi-shot generation.
- Seeing it early: the 3D spaces don't replace the final images. Their value is making spatial relationships that 2D reference images struggle to convey visible early on, and linking them to the generation and editing steps that follow.
