ComfyUI Gen AI Course (Adult)

ComfyUI Gen AI Course (Adult) 

Open for personal, corporate and tertiary signups. Corporate Group Bookings available. Useful for inhouse creative teams.
Ages 21-Above (2:00pm-6:00pm)
(Class size of approx. 5-7 students)
(Nearest MRT is Tai Seng (Exit B), Building has parking available.)

The 2-day (6hrs) Generative AI course offers a comprehensive lesson plan for adults, designers and working professionals to learn about the power of ComfyUI with Flux Dev Model. Being the only course provider in Singapore to teach ComfyUI, our toolset workflow includes  “Dual Clip Loader” node, which is used to for text encoder and embeddings, including options for both clip_l.safetensors and t5xxl fp 16 safetensors. The t5xxl fp 16 safetensors in this context refers to the larger, more powerful text encoder used in Stable Diffusion 3 and related models.

Flux Dev is a powerful 12-billion parameter text-to-image model developed by Black Forest Labs. It is designed for both research and development purposes, offering high-quality image generation capabilities, including the ability to preserve character consistency, perform local editing, and reference styles from images.

As an open-source software hosted on GitHub, its capabilities continue to enhance through contributions from users around the globe. Currently this tool is opensourced (get it while you still can), as compared to Midjourney or Leonardo which requires a monthly subscription of ~30USD.

Consistency of Faces with the aid of LoRA

In the lesson, students will receive guidance from their instructor, on the steps of installing the software, loading nodes, which  can be intimidating to 1st time users. The advantage of ComfyUI over Midjourney, Leonardo and other web based Apps, is that you can create high quality, realistic faces, control their poses & facial expressions, inpainting (cleaning & regenerate certain areas) at the privacy of your own local machine, without guardrails. You can go a step further by training the software with novel faces and images, or merging the appearance with a specific celebrity or style, using LoRa (Low-Rank Adaptation)

Whether you are a complete beginner or simply looking to enhance your current profession with AI skills, this professional course can provide a solid foundation for future artistic endeavors in the field of photography, CGI influencer content creation, interior and architectural design, creative illustrations or 3D Avatars. The curriculum covers essential AI JSON workflows, providing a comprehensive understanding of the technology that increasingly shapes industries, including your own. From the basics to advanced applications, you’ll gain insights that can elevate your creative projects, especially in the realms of TV commercials, motion design, and 3D animation.

Interior designers can input a reference image and output different types of design styles.

Lesson 1  – 4hr

• Installing ComfyUI
• Using the Interface
• Loading Useful Nodes
• Understanding Sampling Methods, CFG, Batch Count, Seed
• What are Checkpoints Safetensors
• Prompt Techniques
• Using our Workflow (Json Files)
• Using Checkpoints
• Using LoRa (Art Styles, Character)
• Using Inpaint / Outpaint (Removing or Modifying an Image, Extending an Image)
• Using ControlNet (Using References to guide the AI)
• Using Canny, Depth
• Image to Image, Interior Design Workflow

Lesson 2 – 2hr

• Share Post-Lesson 1 results and roadblocks
• Debugging Workflow Errors, Fixing Poor Image Results
• Github & Huggingface Resources for Checkpoint Models, Loras

(Requirements: You need a Powerful Laptop/PC with a Nvidia GPU to run ComfyUI, minimal 4070, best 5090, with ~300-700GB of free  SSD space – depending on how deep you wish to go)

Questions? Enquire about this course now via Whatsapp.

Outpainting – Extends the image by predicting what should appear beyond the original borders.

Inpainting – Removes or add certain objects in the scene

Inpainting – Add a World War II tank in the scene