Trade show renderings are more painful than you think
Foshan's Taobo Road — the hub of building materials trade show companies. One firm here specializes in建材 booth design, 20 people, serving ceramic, sanitary ware, and stone brands. Peak season hits during the annual Ceramics Expo, Canton Fair, and Shanghai Building Materials Exhibition.
Their pain point: every trade show project requires 8-15 booth renderings. Using 3ds Max modeling + V-Ray rendering, one image takes 3-5 hours, a full set takes 3 days minimum. Client wants changes? Another 3 days. During peak season, designers pulling all-nighters for rendering is routine.
The owner, Mr. Lin, started getting anxious last year — competitors were using AI to produce renderings, at absurd speeds. He feared falling behind. But he'd tried Midjourney directly, and the problems were big: booth dimensions were wrong, product textures were AI-invented, and while the lighting looked nice, it didn't resemble a real building materials showroom.
From 3ds Max to Midjourney + ComfyUI
Lin found a designer who knew ComfyUI and spent two weeks building a workflow. Core idea: AI handles environment and atmosphere, products use real textures, structure is locked with ControlNet.
Step 1: Sketch for direction
The designer hand-draws or builds a rough SketchUp model, screenshots it into Midjourney V8.1, and enters a prompt with booth dimensions, materials, and lighting style. Midjourney outputs 4 direction images in 2 minutes. The client picks one.
This step replaced the old "build 3D white model, then adjust lighting and materials" workflow, jumping straight to "atmospheric direction images." Clients make decisions much faster.
Step 2: ComfyUI refinement
The chosen image goes into ComfyUI, using ControlNet to lock booth structure and perspective, then swaps in real product textures — tile patterns, stone finishes, sanitary hardware. This step is critical: don't let AI invent products, only let it build environments.
Refining one image: 30-40 minutes. A full set of 8 images: delivered within 2 hours.
Lin's words: "Before, producing renderings was like squeezing toothpaste. Every time the client pushed, I'd panic. Now it's an assembly line. Client changes? No sweat."
Three pitfalls hit
Pitfall 1: AI got tile patterns wrong. Midjourney doesn't recognize specific building material brand colors. The rendered tile patterns didn't match actual products. Client saw it immediately: "That's not our tile." Solution: switch to ComfyUI's image-to-image mode, using real product photos as input. AI handles only lighting and environment. Problem solved.
Pitfall 2: Client suspected cutting corners. Delivering too fast bred distrust. One client questioned: "3 days to 2 hours — are you using templates to fob me off?" Lin's response: show clients the ComfyUI workflow node graph, proving each image is custom-generated, not templated. He also attached design notes explaining the lighting logic and material choices for each image.
Pitfall 3: Designer resistance. Two senior designers felt AI was stealing their jobs and refused to learn ComfyUI. Lin didn't force it. He let the younger designers who learned first produce more and earn more commission. Once the old hands saw the efficiency gap, they learned voluntarily. Now all 6 designers use ComfyUI, and the senior designers have become the main force in reviewing AI output quality — their experience spots details AI didn't handle right.
Numbers
- Rendering speed: 3 days/set → 2 hours/set (10x)
- Per-image cost: 800 RMB (including labor) → 60 RMB (Midjourney + electricity)
- New clients in 3 months: +12 (previously 5-6 per quarter)
- Designer overtime: ↓70%
Tool costs: Midjourney Pro $30/mo, ComfyUI free (runs on local RTX 4060). 6 designers, monthly cost about 1,300 RMB.
Why more new clients? Because with fast rendering, Lin dares to take rush jobs. Before, when a client said "need images next week," he'd decline — 3 days rendering + 3 days revising = can't make it. Now with 2-hour rendering, he shows first drafts the same day. Competitiveness jumps immediately.
Another unexpected effect: designers used to spend so much time on rendering that creativity got squeezed. Now rendering goes to AI, and designers have more time to study booth circulation, lighting strategy, and client brand identity. Lin says the team's design quality actually improved — "They used to be draftsmen. Now they're designers."
Lin said something real: "AI isn't here to replace designers. It's here to replace designers' overtime hours." Now his team spends the saved time running client meetings, planning proposals, and researching new materials and processes. That's the real competitive edge for a trade show company.