Recording from todays AI session

For who missed the session today you can see it here: https://fathom.video/share/7SyZTn2mhzXfoPWJj9Z868tY7hCjmWb7?timestamp=4214.9999

The powerpoint slides is attached here as well.

Below is the key takeways from the meeting:


Meeting Purpose
A beginner's guide to using AI for creative image and video generation.

Key Takeaways
Use LLMs for superior prompts: Use Claude or ChatGPT to structure your ideas into detailed, technical prompts, which significantly improves AI output quality and saves credits.
Start with a monthly subscription: Begin with a monthly plan for a platform like Hicksfield AI (~$30–$40) to test features and avoid being locked into a rapidly evolving tool.
Create character sheets for consistency: For video, generate a character sheet (multiple angles of a subject) and use it as a visual reference to prevent morphing and maintain identity across shots.
LumaLabs is a powerful free starting point: Its 3,000 free credits per account and voice-to-prompt feature (Whisperflow) make it ideal for learning without cost.
Topics
AI Platforms & Use Cases
Hicksfield AI & Freepik AI:
Recommendation: Start with one. They are all-in-one "AI suites" that include image and video generators.
Cost: ~$30–$40/month for unlimited image generation.
Hicksfield Sol: A unique feature for hyper-realistic, cinematic images.
LumaLabs.ai:
Recommendation: A powerful free starting point.
Credits: 3,000 free credits per account.
Key Feature: Whisperflow, which allows voice-to-prompt generation.
Video-Specific Platforms:
SeaDance 2.0: The current top platform for high-quality video.
Kling 3.0: A strong alternative, often integrated into suites like Hicksfield.
Cost-Saving Tip: If generating a lot of video, buy a separate Kling subscription. It's cheaper than using credits within a suite.
Image-Specific Platforms:
NanoBanana 2: High-quality, realistic images, especially for products.
Uni1 (via LumaLabs): Creates realistic, cinematic images with natural grain.
Midjourney: Generates unique, creative images; preferred by some agencies over NanoBanana.
Prompting with Large Language Models (LLMs)
Core Principle: Use LLMs (Claude, ChatGPT, Gemini) to translate your high-level ideas into the detailed, technical prompts that AI generators require.
Claude:
Recommendation: Currently the top-performing LLM.
Key Feature: Can upload PDFs (e.g., brand guides, film theory) to train it on specific knowledge, making it a powerful creative partner.
Prompting Workflow:
Idea: Describe your concept to the LLM in natural language.
Structure: Use a template to guide the LLM in creating a structured prompt with technical details (e.g., camera, lens, lighting).
Generate: Copy the LLM-generated prompt into the image/video generator.
Character Consistency for Video
Problem: AI video generation can cause characters to morph or distort between shots.
Solution: Character Sheets:
Process: Create a visual reference by generating multiple images of the same character from different angles (front, back, side).
Application: Use this character sheet as a visual reference when prompting for video to ensure the AI maintains a consistent identity.
Next Steps
Mamad:
Share the workshop recording, presentation, and prompt templates.
Participants:
Start with a free LumaLabs.ai account (3,000 credits) to experiment.
Consider a monthly Hicksfield AI subscription (~$30–$40) for unlimited image generation and access to video tools.
Use an LLM (Claude or ChatGPT) to structure prompts before generating images or videos.