Tendências em Mídia e Interação (CIN0055)

Tendências em Mídia e Interação (Trends in Media and Interaction, CIN0055) is a 60-hour undergraduate course in CIn/UFPE's new curriculum, taught by Giordano Cabral since 2026.1, with Jefferson Marcelo da Paixão Santos as teaching assistant. It succeeds Introdução à Multimídia (Introduction to Multimedia), which he taught for 21 editions, and keeps the same structure: students identify the trends and the professor provides the framework to analyze them. The field spans multimedia, XR, computer graphics, audio, games, animation, and interaction; rather than covering it in full, each student picks a topic and goes deep into it.

Two editions: 2026.1, from March to July, with around 41 students; and 2026.2, currently underway, with 14 students, remote by default. Each edition has its own website, with submissions, galleries, and future maps made available to the class on presentation day; the current edition is at www.cin.ufpe.br/~grec/tmi/26-2/.

In 2026.2, three assignments weighted 2, 4, and 4: sweeping tens of thousands of tools with an AI and narrowing down to one, inside the Hiper Deep Research system, which logs the time spent on each item; building your own futurization skill and producing a future map of your topic, confronted in class with a map made beforehand by the professor, blind, from the same briefing; and an executable experiment, accessible to people outside the class, tested in class for 25 minutes by classmates. A criterion common to all three: the technology must not be mature. What is assessed is the process, and the pedagogical contract is written for the students' AIs, in an llms.txt file that states what they must and must not do. In 2026.1, the design was still an adapted version of Introdução à Multimídia: seminars by area, one individual experiment per student, and one group project.

• https://farol-jogo.vercel.app/

2026.2 — in progress: nineteen topics, one map per student, one experiment per student

A class of 14 students, remote by default.

Public links (mirror on the CIn server, published on September 17, 2026):

• hub: https://www.cin.ufpe.br/~grec/tmi/26-2/
• the 19 topics: https://www.cin.ufpe.br/~grec/tmi/26-2/temas/
• galleries (inspiration and development): https://www.cin.ufpe.br/~grec/tmi/26-2/galeria/
• Hiper Deep Research: https://www.cin.ufpe.br/~grec/tmi/26-2/hdr/
• examples: https://www.cin.ufpe.br/~grec/tmi/26-2/exemplos/
• activity 01 instructions: https://www.cin.ufpe.br/~grec/tmi/26-2/atividade/
• materials (course plan, activity 03, document format, movement 3): https://www.cin.ufpe.br/~grec/tmi/26-2/materiais/
• movement 3: https://www.cin.ufpe.br/~grec/tmi/26-2/experimento/
• futures (maps by topic, made available on each presentation date): https://www.cin.ufpe.br/~grec/tmi/26-2/futuros/
• original site on Vercel: https://tendencias-midia-interacao.vercel.app/

2026.1 — seminars by field and one experiment per student

First edition, from March to July 2026, with around 41 students — the count comes from the random allocation of individual presentations: 14, 14, and 13 per day.

Individual experiments (April–May):

• PlAIne — AI-generated travel itineraries presented as narratives, grouped by geographic proximity and climate. Author: Anne Collier.
• EchoFeed AI — a simulator of personalized feeds and algorithmic bubbles: the same topic framed for two profiles.
• Flash Mind — generates structured flashcards from medical PDFs.
• O Olhar Normativo XR — a live demonstration of algorithmic bias in computer vision and augmented reality.
• HobbyHub — cross-recommendations among books, films, and music by "vibe" (Lovable, Gemini, Spotify, Google Books, and TMDB APIs).
• Dungeon Forge — a dungeon generator with a tactical map for tabletop RPGs (Gemini, Canvas API).
• Gerador Dinâmico de Conteúdo — the prompt defines the behavior of an animated scene in real time, rather than generating media.
• PGPresença Geo — automated spatial verification and privacy in academic administration.
• Você já foi reprovado por um robô sem saber? — generative AI and computer vision for image interpretation.
• Also submitted: Mechanistic Interpretability (IOI, activation patching), Lore Forge, a maze, and a public tender analyzer. Final projects (June–July):

• Mock Interview AI — technical and behavioral interview simulator with transcription (Whisper) and assessment by GPT-4o, built with Next.js on Vercel. Team: Anne Pereira Collier de Mendonça, Antonio Henrique Souza Lima, João Luis Gomes Agra, João Victor Ribeiro Costa de Omena, Maria Clara Albuquerque Moura, Renato Moreira Serrano de Andrade.
• Smart DJ — from a camera-based idea to AI music curation: users describe the occasion and the app creates a journey with arrival, warm-up, peak and closing stages, playing music through Spotify. Link: https://github.com/virolivs/smart_dj.
• SoundScape AI — the city narrated by AI in real time: combines weather, traffic, events and points of interest into spoken narration, designed with accessibility in mind.
• LLMs as Autonomous Agents in Games — a survival game with a procedurally generated world for observing selfish or altruistic behavior in LLM-controlled agents.
• Meownarch — a game in which a cat monarch commands LLM-controlled dinosaurs with distinct personalities.
• FitFlow — personalized workout generation using RAG over a scientific knowledge base, with safety guardrails.
• AI Narrative Game — a text RPG in which AI acts as the game master, generating scenes, choices, images and atmosphere (Next.js, Vercel AI SDK).
• Microexpression analysis in sales meetings — a video pipeline using MediaPipe and Whisper. Team: João Henrique Portela, Matheus Borges, Rodrigo Rossiter, Eduardo Vinnícius.
• Kineo-AI, Civilization (the Games & Engines group's GDD, in Notion), DJ Interativo and a game in Unreal Engine.