+3.500
GL±0.000
+7.200
A-01
+14.400
▸ WORKFLOW · 30/30
↔ Drag · Scroll · Reveal

See
Before
You Build.

A free, no-nonsense workshop on ComfyUI — the open-source AI tool that turns your sketches, CAD files and 3D models into photorealistic renders. Built for designers. Taught by one.

+3.500 · GL±0.000
↓ Scroll
Format
Live · Online
Sessions
6 Workshops
Cost
₹0 — Free
For
Designers & Students
[ Collaboration · Cohort 01 ]
Kala The Art × MODE India
Craft meets narrative. Computation meets atmosphere. Two practices, one workshop.
ComfyUI for Designers · Kala The Art × MODE India · Free Workshop · Nagpur 2025 · Architecture × Narrative · Open Source · ControlNet · Stable Diffusion · SDXL · Flux Models · AI Renders · Free Forever ·
ComfyUI for Designers · Kala The Art × MODE India · Free Workshop · Nagpur 2025 · Architecture × Narrative · Open Source · ControlNet · Stable Diffusion · SDXL · Flux Models · AI Renders · Free Forever ·
[01 / Philosophy]

Knowledge
is
Free.

"Knowledge should be open source —
just like ComfyUI."

Designers already know what they want to build. The problem is showing it. ComfyUI closes that gap — and since it's open-source and free, so is this workshop.

No subscriptions. No paywalls. No gatekeeping. Just the tool, the knowledge, and the community. Every designer deserves access to this.

06

Six Sessions.
Total Control.

01 — 06
01
Inside the Machine
What diffusion models actually do. Why ComfyUI's node graph beats every black-box tool. Full setup — live, together — in under 10 minutes.
BasicsSetup
What you'll learn

An under-the-hood look at how Stable Diffusion actually works — denoising, latent space, samplers — without the jargon. You'll install ComfyUI, load your first workflow, and generate your first image before the session ends.

How diffusion models generate images step by step
ComfyUI interface: nodes, connections, queue
Installing models (SD 1.5, SDXL, Flux)
Your first workflow — text to image in 5 minutes
Cloud setup via Google Colab (no GPU needed)
Duration: 90 min
Session screenshots
Session 1 — ComfyUI fundamentals — example render
Session 1 — ComfyUI fundamentals — example render
Session 1 — ComfyUI fundamentals — example render
Session 1 — ComfyUI fundamentals — example render
02
Prompt Like an Architect
Light, material, atmosphere, scale. Build the vocabulary that makes AI render what you actually mean — not what it guesses.
PromptingSD Models
What you'll learn

Prompting for architecture is a skill. You'll learn the exact vocabulary — material names, lighting terms, camera angles, style references — that reliably produces professional renders instead of generic AI images.

Positive and negative prompt structure
Material keywords: concrete, glass, stone, wood
Lighting: golden hour, overcast, interior ambient
Style references: architectural photography, Dezeen aesthetic
CFG scale and sampler settings that actually matter
Duration: 90 min
Session screenshots
Session 2 — materiality studies — example render
Session 2 — materiality studies — example render
Session 2 — materiality studies — example render
Session 2 — materiality studies — example render
03
ControlNet — Your Sketch, Obeyed
Feed your linework, AutoCAD exports, or Rhino depth maps. Get renders that respect your actual geometry. The session everyone comes for.
ControlNetCAD → AI
What you'll learn

ControlNet is the breakthrough that makes AI renders useful for architects. Instead of a random image, you feed it your actual drawing — and it generates a photoreal render that follows your geometry exactly.

Installing and connecting ControlNet nodes
Lineart mode — from AutoCAD export to render
Depth map mode — from Rhino/SketchUp 3D view
Canny edge detection for section drawings
Combining multiple ControlNet inputs
Duration: 2 hrs
Session screenshots
Session 3 — ControlNet sketch to render — example render
Session 3 — ControlNet sketch to render — example render
Session 3 — ControlNet sketch to render — example render
Session 3 — ControlNet sketch to render — example render
04
Ten Schemes in Ten Minutes
Rapid concept iteration — materials, moods, elevations. One plan becomes ten design directions before the client walks in.
IterationConcepts
What you'll learn

Build a batch generation workflow that takes one floor plan and outputs ten completely different design options in minutes. This is the workflow that changes how you present to clients forever.

Batch nodes — generate multiple outputs from one input
Swapping styles, materials, and seasons automatically
Building a client moodboard with AI in real time
Image-to-image for iterating on existing renders
Seed management for reproducible results
Duration: 90 min
Session screenshots
Session 4 — light and mood — example render
Session 4 — light and mood — example render
Session 4 — light and mood — example render
Session 4 — light and mood — example render
05
Drop It in the Site
Inpainting your building into real photographs. Generating Indian urban contexts. Making proposals look inevitable.
InpaintingContext
What you'll learn

Take a real photograph of your site and composite your building into it — with matching light, shadows, and atmosphere. Also generate contextual backgrounds specific to Indian cities and landscapes.

Inpainting workflow — mask, fill, blend
Matching lighting between render and photograph
Generating Indian street, tropical, and urban contexts
Outpainting — extending a render into a wider scene
People, trees, and entourage with AI
Duration: 90 min
Session screenshots
Session 5 — site context renders — example render
Session 5 — site context renders — example render
Session 5 — site context renders — example render
Session 5 — site context renders — example render
06
From Screen to Portfolio
Upscaling to print resolution. Batch rendering. Integrating with Photoshop and InDesign. A workflow that actually sticks.
OutputWorkflow
What you'll learn

Turn your ComfyUI outputs into print-resolution portfolio pieces. We cover upscaling, colour correction, and exactly how to hand off AI renders to Photoshop and InDesign without losing quality.

4x upscaling with ESRGAN — crisp at A1 print size
Colour grading nodes inside ComfyUI
Exporting to Photoshop for post-production
Batch rendering an entire presentation sheet
Building a reusable workflow template for your studio
Duration: 90 min
Session screenshots
Session 6 — portfolio frames — example render
Session 6 — portfolio frames — example render
Session 6 — portfolio frames — example render
Session 6 — portfolio frames — example render

Who should
attend.

B.Arch and M.Arch students, any year
Architects who present with bad renders
Designers exploring AI + parametric pipelines
Academics preparing to teach emerging tools
Anyone who has felt limited by Midjourney
Tools covered — all free, all open-source
ComfyUICore
Stable Diffusion / SDXL / FluxEngine
ControlNetGeometry
Rhino / SketchUp / AutoCADInput
Google Colab — no GPU neededCloud
[ 02 / Workshop Leads ]
Two Practices.
One Workshop.
→ Most AI-architecture courses teach you how to render. We teach you how to render something worth seeing. Computation meets atmosphere. Craft meets narrative.
Kamal Gyanani
[ Lead 01 · Craft & Computation ]
Kamal Gyanani
Architect · Parametric Designer · Nagpur, India
FOUNDER · KALA THE ART

Practicing architect working across high-rise residential, commercial towers, and parametric design. Uses Grasshopper, Rhino, and ComfyUI on real projects — not just demos.

Every session is built around a real architectural problem. How do you show a client three facade options in 20 minutes? How do you visualise a G+17 tower before the structure is up? That's what you learn here.

"Knowledge should be open source — just like ComfyUI."
Based In
Nagpur, Maharashtra
UDCPR · NBC · IS Codes
Daily Tools
Rhino · Grasshopper
AutoCAD · Revit · ComfyUI
Prabhat Tripathi
[ Lead 02 · Narrative & Concept ]
Prabhat Tripathi
Architect · Designer · Educator · Nagpur, India
FOUNDER · MODE INDIA

A multidisciplinary practice working at the intersection of architecture, spatial storytelling, hospitality design, and experiential environments. Explores architecture not just as built form, but as immersive narrative shaped through atmosphere, materiality, and memory.

15+ years across architecture, visual communication, and creative direction. Professor at IDEAS Nagpur. Master's in Architecture (History & Theory) from University of Westminster, London. Teaches culture, contextual thinking, and experiential spatial systems.

"Spaces are felt before they are understood."
Known For
Heritage Hospitality
Spatial Storytelling · Cinematic Design
Teaches At
IDEAS · Nagpur
Architectural Design Studios
[ 02b / The Pairing ]

Why two
instructors?

Most AI-architecture workshops stop at the tool. You learn the buttons, you make pretty pictures, you go home. This isn't that.

Kamal brings the craft. Six years of practicing architecture in Nagpur, building real high-rises with Rhino, Grasshopper, and ComfyUI. He'll show you how to use these tools the way a working architect actually does — fast, iterative, problem-driven.

Prabhat brings the narrative. Fifteen years of multidisciplinary design, a Master's in History & Theory from Westminster, and an active teaching practice at IDEAS Nagpur. He'll push every student to ask the harder question — not just 'how do I render this?' but 'why is this worth rendering?'

The combination is what makes this workshop different from the YouTube tutorials. You leave knowing the tool and knowing what to do with it.

Teaching
Assistants.

The people who make sure you actually understand it — not just watch it.

Vedika — workshop team
Vedika
Khade
Teaching Assistant
ComfyUI · Workflows
Vansh — workshop team
Vansh
Shelke
Teaching Assistant
ControlNet · Renders
Sarth — workshop team
Sarth
Pawar
Teaching Assistant
Workshop Support
Samadnya — workshop team
Samadnya
Barve
Teaching Assistant
Workshop Support
Kanak — workshop team
Kanak
Kothari
Teaching Assistant
Workshop Support
Anaya — workshop team
Anaya
Deshpande
Teaching Assistant
Workshop Support
Divyanka — workshop team
Divyanka
Latkar
Teaching Assistant
Workshop Support
Ashutosh — workshop team
Ashutosh
Nagpure
Teaching Assistant
Prompting · Models

Common
Questions.

Everything you need before registering.

Do I need a GPU?+
No. We cover cloud alternatives — Google Colab, RunDiffusion — that run on any laptop. Session 1 includes a live setup walkthrough so nobody gets left behind.
Students or working architects?+
Both. Early sessions are built for beginners. Later sessions on workflow integration and client presentation are where practicing architects get the most value.
Do I need coding knowledge?+
Zero. ComfyUI is a visual node graph. If you've used Grasshopper, you already have the instinct. If not, you'll pick it up in the first 30 minutes.
Will sessions be recorded?+
Yes. All registered participants get recordings. In keeping with the open-source philosophy, sessions will eventually be made publicly available.
Why is this completely free?+
The tool is free. The knowledge should be too. Architecture education is expensive enough — this is a contribution back to the community.
[03 / Schedule]

Six
Saturdays.
Mark your calendar.

→ All sessions are Saturday evenings · 8:00 PM IST
→ Live online via Google Meet · Recordings within 24 hrs
→ 6 weeks from 13 June → 18 July 2026
→ Can't attend live? Register anyway — recordings are yours forever.
▸ Session 01
Sat 13 Jun2026 · 8 PM IST
Inside the Machine
▸ Session 02
Sat 20 Jun2026 · 8 PM IST
Prompt Like an Architect
▸ Session 03
Sat 27 Jun2026 · 8 PM IST
ControlNet · Sketch Obeyed
▸ Session 04
Sat 04 Jul2026 · 8 PM IST
Ten Schemes
▸ Session 05
Sat 11 Jul2026 · 8 PM IST
Drop It in the Site
▸ Session 06
Sat 18 Jul2026 · 8 PM IST
Screen to Portfolio
→ Cohort 01 · Inaugural Batch. First-ever run of this workshop, co-hosted by Kala The Art × MODE India. Limited to 50 seats so we can keep it interactive. Once registration closes, the next cohort runs late 2026.
FREE

One Spot.
Zero Cost.
All Access.

Sessions are capped to keep quality high. Your spot is free to claim and free to keep. No payment. No upsell. No conditions.

Knowledge should be open source

No spam · No fees · No conditions · Ever