by thundervelvet on 4/8/25, 12:25 PM with 0 comments
I love creating pictures and video on socials using things like ChatGPT and Mid-journey and convert it to video on Replicate and Fal.
But I realized it's super time consuming
So I created a AgentHeroes, a repository to train models, generate pictures, video and schedule it on social media.
Not sure if it's something anybody needs so happy for feedback.
Of course a star would be awesome too
*Here is what you can do:*
* Connect different services like Fal, Replicate, ChatGPT, Runway, etc. * Train images based on models you upload or using models that create characters. * Generate images from all the models or use the trained model. * Generate video from the generated image * Schedule it on social media (currently I added only X, but it's modular) * Build agents that can be used with an API or scheduler (soon MCP): * Check reddit posts * Generate a character based on that post * Make it a video * Schedule it on social media
*Everything is fully open-source AGPL-3 :)*
*Some notes:*
Backend is fully custom, no AI was used but the frontend is fully vibe code haha, it took me two weeks to develop it instead of of a few months.
There is a full-working docker so you can easily deploy the project.
*Future Feature:*
* Connect ComfyUI workflow * Use local LLMs * Add MCPs * Add more models * Add more social medias to schedule to
And of course, let me know what else is missing :)