// curriculum
Three courses.
One professional ladder.
A complete path from your first node graph to supervising AI-assisted pipelines at studio scale. Each course builds on the last. Each one stands alone.
The curriculum is structured but not rigid. Exams happen when the cohort is ready, not on a fixed week. You'll receive an industry-recognized certificate of completion at the end. The real goal is fluency, not a piece of paper.
This is a ComfyUI course for VFX professionals. Structured, cohort-based AI compositing training built for working compositors and VFX artists. Not a hobbyist tutorial channel. Not a prompt-engineering crash course. Production training for people who ship shots.
Most ComfyUI material teaches you to run someone else's workflow. That breaks the moment a model changes or a supervisor hands you a real plate. This programme teaches the underlying model deeply enough that when a new release drops, you can read its components, wire a pipeline, and apply it to a VFX task on your own.
The workflows are the same ones deployed on real feature films: Nuke and DaVinci Resolve handoff, EXR pipelines, Blender integration, temporal consistency for shot-level video, and studio-scale automation.
Built for industry
Vetted by the masters.
Not an academic exercise. Not a collection of cool internet workflows. This curriculum exists to solve two specific problems.
01 · Mapped to real job postings
Every module comes from real job postings at top-tier VFX studios hiring for AI-assisted roles. If a studio isn't asking for it, we don't teach it.
02 · Supervisor approved
Veteran VFX Supervisors at major studios have reviewed the full track. Every workflow holds up to real production pressure.
Foundation
Course goal
Build the mental model so deeply that when any new model or workflow drops, you can identify its components, wire a basic pipeline, and apply it to a VFX task without hand-holding.
Term 1
The Machine
Understand the primitives
ComfyUI Orientation + Cinematic Language
Node graph, queue, execution order. Plus lighting, lens language, framing, and colour theory as it applies to generative output.
Diffusion Model + VAE
DiffusionModelLoader, VAELoader, VAEEncode/Decode. Latent space explained.
CLIP & Text Conditioning + Foundational Prompting
CLIPLoader, CLIPTextEncode, KSampler settings. Prompt discipline, positive/negative structure, habits to set early.
Full txt2img + Image to Image + Inpainting & Outpainting
Wiring all components, denoise dial, mask nodes, clean plates, object removal, plate extension.
+ buffer sessions
Dedicated time for errors and troubleshooting. Learning to read console logs and debug a broken pipeline when a shot is due in an hour is half the job.
End of Term 1, examination when the cohort is ready.
Term 2
The Controls
Learn to condition and constrain generation
ControlNet
Depth, canny, normal. Constrain generation to match a plate.
LoRAs + Upscaling
LoraLoader, stacking, strength, upscale models, tiled output for VFX resolution.
Digital Matte Painting
Outpainting + depth ControlNet + tiled upscale from a real plate.
Capstone: Single Frame
Full pipeline: plate, generated BG, Nuke/Resolve handoff.
+ buffer sessions
Dedicated time for errors and troubleshooting. Learning to read console logs and debug a broken pipeline when a shot is due in an hour is half the job.
End of Term 2, examination when the cohort is ready.
Term 3
The Timeline
Move from frames to shots
I2V Foundations
How I2V differs from I2I, motion prior, prompt structure, key parameters.
First & Last Frame + Style Transfer
Anchor frames → controlled motion, applying visual style to a plate.
Animating Stills + AI Paintout
Matte painting → animated plate, mask propagation across frames.
Capstone: Full Shot
Real plate, AI video pipeline, DaVinci/Nuke final handoff.
+ buffer sessions
Dedicated time for errors and troubleshooting. Learning to read console logs and debug a broken pipeline when a shot is due in an hour is half the job.
End of Term 3, examination and demo reel when the cohort is ready.
Demo reel
60–90 sectxt2img, img2img, inpainting/outpainting, basic DMP, one AI video clip.
Exit criteria
You can pick up any new model release, identify its components, wire a basic pipeline, and apply it to a VFX task independently.
What's not covered
Rough comp cleanup, relighting, look development, video-to-video restyling, FX generation, face replacement, advanced prompting, LoRA training, fine-tuning, LLM integration, Python API, and custom nodes are covered in Comfy II & III.
Compositor
Course goal
Build compositor-grade image and video pipelines. You leave with production-ready workflows deployable on real shots.
Term 1
AI Image Pipelines for Production
Single-frame compositor-grade workflows
Workflow Architecture
Modular design, subgraphs, reusable components, project hygiene.
Advanced Prompting Techniques
Prompt weighting, attention control, negative prompt strategy, cinematic language in prompts, cross-frame prompt consistency.
Rough Comp Cleanup
img2img on a Nuke/Resolve output. Edge fringing, spill, inconsistency cleaned up with denoise discipline.
Relighting
Lighting ControlNet, normal pass input. Match FG illumination to a generated or replaced BG.
Look Development
Translate a script or brief into a frame-accurate visual narrative. Style reference, img2img, sky replacement as sub-task.
Face Enhancement & Replacement
Segmentation mask + inpainting + quality pass via Impact Pack.
Capstone: Full Single Frame Comp
Plate → BG → relight → look dev → face cleanup → Nuke/Resolve handoff.
+ buffer sessions
Dedicated time for errors and troubleshooting. Learning to read console logs and debug a broken pipeline when a shot is due in an hour is half the job.
End of Term 1, examination when the cohort is ready.
Term 2
AI Video Pipelines for Production
Shot-level temporal workflows
I2V Production Workflow
Temporal consistency, prompt discipline, parameter control at shot level.
Video-to-Video Restyling
Keyframe and reference-based control for temporal coherence across extended sequences.
BG Replacement (Video)
Depth ControlNet → I2V, maintaining frame consistency across a shot.
FX Generation
Proxy FX → img2img → video. Fire, smoke, water from a hand-animated proxy.
Multi-Pass Shot Assembly
Combine AI-generated elements (BG, FX, face) into a coherent shot.
Capstone: Full VFX Shot
Real plate → full AI-assisted pipeline → DaVinci/Nuke delivery.
+ buffer sessions
Dedicated time for errors and troubleshooting. Learning to read console logs and debug a broken pipeline when a shot is due in an hour is half the job.
End of Term 2, examination and demo reel when the cohort is ready.
Demo reel
90–120 secFull single-frame comp, video-to-video restyle, full AI-assisted VFX shot.
Exit criteria
You can build and execute a full AI-assisted VFX pipeline on a real shot, from plate to final handoff.
What's not covered
Python API, custom nodes, LoRA training, model fine-tuning, LLM integration, creature motion, and studio-scale pipeline design are covered in Comfy III.
Supervisor
Course goal
Pipeline thinking at studio scale. You leave able to design, automate, supervise, and evaluate AI-assisted VFX pipelines across a team.
Term 1
Pipeline Architecture & Advanced Technique
Design systems, train models, integrate AI at depth
ComfyUI API Fundamentals
Submitting via Python, polling results, batch processing.
Folder-Watcher & Batch Automation
Automated ingestion → processing → output pipelines.
Custom Node Development
Building nodes for studio-specific tasks.
LoRA Training
Dataset curation, captioning, trigger words, training parameters, integration into pipeline.
Model Fine-Tuning
DreamBooth, full fine-tune, evaluating output quality, when to fine-tune vs LoRA.
LLM Integration
Using LLMs for prompt generation, shot sequencing, and creative ideation inside a ComfyUI pipeline.
Capstone: Automated Pipeline
End-to-end automated pipeline: folder in → processed shot out.
+ buffer sessions
Dedicated time for errors and troubleshooting. Learning to read console logs and debug a broken pipeline when a shot is due in an hour is half the job.
End of Term 1, examination when the cohort is ready.
Term 2
Studio-Scale Supervision
Lead AI-assisted VFX at team and studio level
Model Evaluation & Selection
Assess a new model for a specific shot. SDXL vs Flux, LTX vs Wan, and others.
Round-Trip Integration
ComfyUI ↔ Nuke ↔ Blender ↔ DaVinci. EXR pipelines, metadata, versioning.
Creature & Character Motion Studies
Generative video for motion exploration on creature and character work.
Pipeline Design for Teams
Modular architecture others can operate, version control, documentation, QC checkpoints.
Integrating AI into a Traditional VFX Pipeline
Where AI fits, where it doesn't. Managing client, director, and team expectations.
Cost & Resource Management
GPU budgeting, cloud vs local decisions, when to scale.
Capstone: Supervise a Full Shot
Design, delegate, and QC a full AI-assisted VFX shot as a supervisor.
+ buffer sessions
Dedicated time for errors and troubleshooting. Learning to read console logs and debug a broken pipeline when a shot is due in an hour is half the job.
End of Term 2, examination and demo reel when the cohort is ready.
Demo reel
90–120 secAutomated pipeline output, video restyle at sequence scale, creature/character motion study, multi-pass shot with supervisory documentation.
Exit criteria
You can design, automate, and supervise an AI-assisted VFX pipeline at studio scale.
How it works
Built for working artists.
To make sure you actually master the material and ship your shots, each cohort stays strictly small. No giant, anonymous webinars.
Small, elite cohorts
Small cohorts mean you get direct feedback on your work, not a slot in a 200-person Zoom call.
Ultimate flexibility
Every session is recorded in HD. Miss a class? Catch up at your own pace.
Direct 1-on-1 sessions
Get time one-on-one. Debug your custom nodes, fix errors, or break down a real shot you're working on.
At a glance
The full ladder.
Comfy I
Foundation
3 terms
12 sessions
Comfy II
Compositor
2 terms
13 sessions
Comfy III
Supervisor
2 terms
14 sessions
Total
Full programme
7 terms
39+ sessions
What you leave with
Production-grade deliverables.
Production-ready modular workflows
Custom ComfyUI pipelines that work on real shots, not toy demos.
A studio-grade VFX demo reel
Before/after pipelines that show you can do the work under production pressure.
Official Comfy Compositing Certification
An industry-recognized credential that matches what studios actually hire for.
An elite alumni network
A closed network of technical, node-based compositors working at top studios around the world.