Opine 😎 Launches: AI Native Social Media App

Opine 😎 recently launched!

‍

^{"Opine is building a completely new social network"}

‍

^{TLDR; Opine is an AI native social media app. Users can create a character, make videos as this character, and share on the Opine platform!}

‍

https://www.youtube.com/watch?v=Q4D-Scw_hPE

‍

Founded by Manuj Mishra & Diogo Valdivieso‍

Hi 👋 everyone! Meet Manuj and Diogo - founders of Opine.

‍

Team

Manuj (CEO) was previously 2nd ML hire at unicorn defence startup, Rebellion. He ranked #1 in his Maths & CS Bachelor’s at Imperial College London and graduated from Oxford with a Master’s in Advanced CS. He published a first-author AI paper at UAI 2024.

Diogo (CTO) has built several mobile games. His top original game, Battle Arena, reached 1M users in 70 countries. He did his Bachelor’s in CS at EPFL and dropped out of his Master’s to build Opine.

‍

AI influencers already exist

On Twitch, there are a category of streamers called VTubers. They use depth cameras to overlay an anime character on their face and body. VTubers game and chat with their audience using this character overlay. They represent 0.5% of channels but bring in 5% of total viewership.

On Instagram and TikTok, AI influencers like Lil Miquela have millions of followers. They promote brands, react to memes, and make lifestyle videos.

However, until now, building an audience around an AI influencer has not been possible for most people.

‍

Opine is building a completely new social network

On Opine, users can create characters with a custom face, voice, and outfit. They can record any video of themselves and Opine transforms them into their custom character. That’s it! Now they can post and share the video on the Opine platform.

Opine features:

• A video feed with algorithmic recommendations.
• Likes, comments, and shares like TikTok or Instagram.
• Pseudonymous characters controlled by real people.

‍

The Founders Approach

The founders have built the technology to make it easy for anyone to be an AI influencer. Large diffusion transformers unlock the possibility of:

1. Controlling characters through movement and facial expression.
2. Maintaining a consistent character across separate videos

In the last 2 months, they’ve innovated their process to generate videos with visually consistent characters. Opine generates a synthetic dataset for each character on-the-fly which can be used to fine-tune any open-source diffusion model in under 20 mins.

Currently, top video generation models like Sora and Runway cannot achieve this level of consistency alone because they only allow video modifications through prompts.

As open-source models improve, the Opine product continues to improve. For example, this week, a new open-source model, Wan 2.1, was released. It’s now the best diffusion video model, topping the VBench leaderboard.

‍

Learn More

‍

^{🌐 Visit www.opine.cool to learn more.}

^‍

^{‍‍👣 Follow Opine on LinkedIn & X.}

‍