Creation of an automated system for generating visual content

Creation of an automated visual content generation system (images + videos with voiceover) with an admin panel and training of the AI model on a custom dataset

A full-fledged prototype / MVP of a system needs to be developed that will automatically generate images and short videos with voiceover based on specified scenarios. Integration of AI tools and training of a custom model on the available photo dataset is planned.

Tasks:

1. Image Generation

• Set up automatic photo creation based on specified themes, styles, and templates.

• Explore integration options: Midjourney, Leonardo AI, Stable Diffusion, or other tools.

• Mandatory: train or fine-tune the model (Stable Diffusion / Lora / custom pipeline) on the provided photo dataset to generate a personalized style or specific type of content.

• Integration via API or custom script.

2. Video Generation

• Implement the creation of short videos (up to 60 seconds) using AI services: Pika Labs, Runway, Kaiber, Synthesia, or alternatives.

• Automatic addition of:

  • text,

  • subtitles,

  • transitions,

  • animations and simple effects.

• The ability to use generated photos as a basis for videos.

3. Admin Panel

Create a simple and intuitive web interface with the following features:

• uploading scripts and texts for content generation;

• choosing templates and design styles;

• setting voiceover parameters;

• starting the generation of images and videos;

• viewing, saving, and downloading finished materials.

The interface should be minimalist, responsive, and fast.