Project Overview: "2047 Toddler Gym"

Project Overview: "2047 Toddler Gym" ​

The Concept: Screen time is rising, physical activity is falling, and children’s health is suffering. This video imagines the year 2047, where the damage from excessive mobile use is so widespread that specialized gyms for 2-to-3-year-olds now exist. To compensate for screen-induced health problems, tiny kids are shown dead-seriously bench pressing, squatting, and running on treadmills.

The Goal: While the visual is satirical, the warning is real. This was made purely to raise awareness and make parents think twice before handing a phone to a small child.

Build the Story Structure: Map out every scene sequentially, ensuring each carries one specific, unambiguous moment.

Refine with an LLM: Drafted and optimized the prompt using a separate LLM to make the language machine-readable. Human judgment was applied at every stage to ensure the tone and structure were correct.

Generate in Melius: Submitted the refined, tested prompt as a single input to generate all scenes at once.

Review and Quality Check: Conducted a manual review of each clip for consistency, expression, and timing. AI generates the content, but humans must judge if it works for the audience.

Experience with Melius AI Overall, the experience was positive. Melius is genuinely useful for short video projects, offering a logical interface and fast turnaround times. However, there are a few limitations.

What Works Well: Quick generation of short clips.

Excellent for projects with fewer than 10 scenes.

Areas for Improvement: The 10-Clip Limit: This is a hard ceiling that breaks the workflow for more complex projects.

Voice Generation: The chatbot unreliably follows voice commands, often ignoring or only partially applying instructions.

Platform Stability: The platform slows down during longer sessions. Tools occasionally freeze without throwing an error, requiring a page refresh to check progress.

Final Thoughts: Melius has immense potential and a solid core idea. If the team is open to it, I have specific, detailed suggestions to improve stability, voice control, and clip limits that I would be happy to share via a Notion page.

This is the canvas link of the project https://app.melius.com/projects/6d624463-df9f-44d7-a7f0-63d07c90e0a1/canvas/df5f6265-62c3-4594-ad49-001a28f52188

This is the link of linkedin post https://www.linkedin.com/posts/araeen_screentime-digitalwellbeing-parenting-activity-7462735226264334336-F2mq?utm_source=social_share_send&utm_medium=android_app&rcm=ACoAABlykYkB7dywhqhS1a0pzqPXSWZ3hJGLdTI&utm_campaign=copy_link

This is link of X post https://x.com/i/status/2056972956883144929

I have also unloaded screen recording of the entire workflow but since it is long videos it is taking time to upload and there is no time left for submission. Hence if possible, I'll add link of workflow later.

I have also unloaded screen recording of the entire workflow but since it is long videos it is taking time to upload and there is no time left for submission. Hence if possible, I'll add link of workflow later.