This workshop is designed for participants who are new to ComfyUI but have a basic understanding of Rhino3D. We will begin by exploring different generative AI models to identify which ones best suit our workflow. Newer models often produce higher-quality results but take longer to generate, while older models are faster but may compromise on quality.
Participants will then create short animation sequences in Grasshopper and convert them into architectural spaces using AI models such as Flux, SDXL, and Gemini 2.5 inside ComfyUI. Throughout the process, we will use concepts like upscaling, ControlNet, IPAdapter, and image refinement. Finally, selected frames will be morphed onto real sites chosen by the participants.
By the end of the workshop, participants will have developed an iterative workflow that connects parametric logic, generative AI interpretation, and contextual design visualization.
The workshop aims to move beyond standalone generative AI tools and integrate them directly into design processes such as parametric modeling. Participants will use Grasshopper3D to create parametric animations, ComfyUI to convert those animations into architectural spaces placed on real sites, and Premiere Pro to combine them into cohesive video sequences.
This workshop aims to move beyond standalone generative AI tools and integrate them directly into design processes such as parametric modeling. Participants will use Grasshopper3D to create parametric animations, ComfyUI to convert those animations into architectural spaces placed on real sites, and Premiere Pro to combine them into cohesive video sequences.
Stage 1 | Parametric Animations
Participants will learn key Grasshopper3D concepts like data trees, color coding, and wire connections, and explore how to animate frames using sliders and different plugins. In this stage, we will be working with abstract patterns, but participants can choose their own designs as well. The focus is on building an understanding of animation within the Grasshopper interface.
Stage 2 | Animation to ComfyUI Render
This stage introduces core AI concepts such as checkpoint models, VAEs, and prompts before translating Grasshopper frames into architectural renders using ComfyUI. Participants will use ControlNet models (Depth and Sketch) and IPAdapter to guide and style their outputs.
Stage 3 | Render to Site
Participants will select preferred designs and integrate them onto real sites using upscaling with SDXL and FLUX, followed by compositing through Gemini 2.5 Flash Preview. Image editing will refine lighting, weather, and time of day for contextual realism.
Stage 4 | Combining in Adobe Premiere Pro
Participants will bring their sequences together in Premiere Pro, learning basic editing, keyframing, adding text, and making video adjustments to create smooth animations.
Program:
Day 1 – Parametric Animation to Rough Architectural Space
Tools: Rhino3D, Grasshopper3D (Plugin – Heteroptera), ComfyUI (SDXL Turbo, SDXL Depth + Sketch ControlNet, IPAdapter), Adobe Premiere Pro
Generating Parametric Animation from Grasshopper
-
Introduction to data trees, color coding, sliders, and wire connections
-
Adjusting camera location for better output
-
Extracting animation frames
Animated Frames to Architectural Space
Combining Frames into Video in Adobe Premiere Pro
- Introduction to the Premiere Pro interface, importing stills, and learning key frames
- Adjusting motion properties to get a smooth animation
- Assignment 1 + Q&A
- Outcome: Final video output
Day 2 – Site Implementation of the Design
Tools: ComfyUI (Flux.1 Dev, Gemini 2.5 Flash Image (Nano Banana), Flux Kontext Max), Adobe Premiere Pro, Kling or WAN 2.2
Upscaling One Frame from Day 1
Merging the Upscaled Design with the Site
Image Editing
Combining Everything
- Video generation with Kling or WAN 2.2
- Combining in Adobe Premiere Pro
- Assignment 2 + Q&A
- Outcome: Final video and 3D model output
