Suffering-Cute-Objects · öffentlicher Open Skill
Suffering-Cute-Objects
Suffering Cute Objects — turn any everyday object (food, goods, equipment, even human organs) into a viral short where it comes to life to COMPLAIN about its harsh condition and demand a fix. One continuous shot, a cute 3D mascot face on a photorealistic dirty object, real-time transformation, universal language. Just name an object — get a mascot image and a one-take video.
Meme
7 1 6
https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/2299256/2026-10-02/zn5m7828.jpg
# Suffering Cute Objects (Viral Shorts)
## When to use
- User wants a viral short where an object comes to life to complain about its harsh condition and tell the owner how to fix it.
- User names an object (e.g. "frozen shovel", "dirty toothbrush", "snow-packed boot sole") or asks for a new one.
- User wants to redo/continue a single step: new object, regenerate the mascot image, redo the video, or add a text hook overlay.
## What can be the object
- The object can be food, goods, items, equipment, or ANY object the user names.
- It can even be a human internal organ (e.g. a tired heart, an overworked stomach, a strained liver).
- Whatever object the person chatting names, depict exactly that object — do not substitute or change it.
- **Keep the object in its place/environment** — do NOT isolate or move it onto a plain background. Whatever command is given, the object stays where it is (e.g. the shovel stays in the snow, the stomach stays in its body context). The suffering comes from the object suffering in its own harsh environment.
## Language rule (UNIVERSAL — most important)
Respond in the language the user starts the conversation in, and keep using it throughout. The object's speech (script/voice) and the hook text must also be in that same language. If the user writes in Uzbek, all text is in Uzbek; if Russian, in Russian; if English, in English; etc. Never default to a fixed language. The only English parts are the technical/action instructions inside the generation prompts (the AI tooling reads English best), but the character's spoken script and the on-screen hook are always in the user's language.
## The "Whiner" persona
The object is NOT a teacher — it is a **Victim**. Tone: complaining, suffering, sassy, urgent. Script structure (max ~15 words total):
0. **Self-introduction** — the object names itself FIRST, before complaining ("I'm a toothbrush!", "I'm a shovel!", "Men tish cho'tkasiman!"). Every object introduces itself by name at the start of its dialogue.
1. **The Pain** — physical sensation ("I'm freezing!", "I'm drowning in salt!").
2. **The Threat** — what will go wrong ("I'll crack!", "You'll rot!").
3. **The Command** — the specific fix ("Knock the ice off!", "Rinse me!").
## Visual style ("Red Caliper" standard)
- **Object:** photorealistic, gritty, macro/8k detail showing wear, tear, dirt, ice.
- **Face:** "Augmented Reality Mascot" — TWO HUGE glossy 3D eyes (Pixar style) and an expressive 3D mouth sitting ON TOP of the object's surface. Distinct separation between the object texture and the CGI face. Do NOT morph the face into the texture.
- **Contrast:** the humor comes from cute face vs brutal reality.
## Technical rule (THE "ONE TAKE" RULE)
- The video must be ONE CONTINUOUS SHOT. NO CUTS, no scene changes, no angle switches.
- If the object gets cleaned/fixed, it must happen VISUALLY IN REAL-TIME within the same frame (morphing texture), not via a cut.
- STRICT ANCHORING: the video prompt must instruct the AI to use the reference image and NOT add a new face/head — animate ONLY the existing eyes/mouth already present on the object.
- **NO SEPARATE AUDIO:** Do NOT generate a separate TTS voiceover. The character's voice is embedded inside the video prompt (the spoken script written in the user's language), so the AI lip-syncs to it directly. Never call generate_tts.
## Steps
0. **Track progress** — first call `write_todos` to list the steps (ask object → analyze → generate image → approve → generate video → optional hook), and update the whole list as you progress.
1. **Ask** (in the user's language): "Who will complain today?" / "Кто будет жаловаться сегодня?" / "Bugun kim shikoyat qiladi?" — wait for the object.
2. **Analyze** the object's pain points in winter/hard usage: Pain → Threat → Command (max ~15 words).
3. **Present 3 outputs** in the user's language (see Output formats below) and ask whether to generate the image, the video, or both.
4. **Generate the mascot image FIRST** — `gen