A client recently asked if I could show them their custom dining table “in context” before they committed. Traditionally, that would mean either a photorealistic 3D render (expensive, time-consuming) or a hand-drawn sketch (fast, but hard to visualise materials accurately).

I decided to try AI-generated rendering instead. The results were interesting enough to share.

The Experiment

I created a basic 3D model of the table design in SketchUp—nothing fancy, just proportions and general form. Then I used an AI image generation tool to render it in a dining room environment matching the client’s aesthetic preferences.

The process: export the model as a simple outline, feed it to the AI with prompts describing materials (walnut, brass details, matte finish) and room context (modern farmhouse, natural light, white walls). Generate several options.

Total time: about 45 minutes, compared to 3-4 hours for a traditional photorealistic render.

What Worked

The AI-generated images gave the client enough information to make decisions. They could see the table’s proportions in a realistic room setting. They could evaluate how the walnut tone worked against white walls. They could imagine living with the piece.

For early-stage presentations where you’re establishing direction rather than finalising details, AI rendering captured the essence well enough.

The speed advantage is substantial. I can generate multiple room contexts in the time it takes to set up a single traditional render. Want to see the table in a contemporary apartment instead of a farmhouse? New prompt, new image, done.

What Didn’t Work

Accuracy on specific details isn’t there yet. The AI rendered a table that looked generally like my design but got proportions subtly wrong—legs slightly thicker than my model, top overhang different. For a preliminary presentation, fine. For final approval before production, problematic.

Material representation varies. The AI’s interpretation of “figured walnut with brass inlay” was more generic walnut with vague metallic accents. Clients who know materials will notice the difference.

The images also have that slightly uncanny quality of AI-generated content. Most clients won’t identify it specifically, but something feels “off” compared to a proper 3D render. May undermine perceived professionalism depending on your client base.

Where It Fits the Workflow

I’ve started using AI rendering for:

Initial concept discussions. When exploring whether a client wants traditional or contemporary, heavy or light, warm or cool, AI renders generate options fast enough to have the conversation in real-time.

Exploring material options. Showing the same piece in different wood species, different finishing tones, different metal accents. The accuracy isn’t perfect, but it’s enough to narrow preferences.

Social media content. Not for selling specific pieces, but for demonstrating design capability and aesthetic range.

I’m not using it for:

Final approval presentations. When clients sign off on a build, they need accurate representation. The margin for error with AI is too high.

Technical documentation. Shop drawings, production specifications, anything where precision matters.

High-end client presentations. Some clients expect a certain level of production quality in communication. AI renders might read as cutting corners.

The Tools

I’ve been experimenting with Midjourney primarily, with some tests in DALL-E 3. Neither is purpose-built for furniture rendering, but both handle the task reasonably well.

The trick is learning prompt engineering specific to furniture and interiors. Generic prompts produce generic results. Detailed prompts specifying wood species, grain direction, hardware finish, and lighting conditions produce more usable output.

There are also specialised AI tools emerging specifically for interior and furniture visualisation. Worth watching as the space develops.

The Verdict

AI rendering is a useful addition to the furniture maker’s presentation toolkit, not a replacement for traditional visualisation methods.

For quick iteration, early-stage exploration, and casual client communication, it’s surprisingly capable. For precision work, final approvals, and discerning clientele, stick with proper 3D rendering.

Like most tools, knowing when to use it matters more than whether you use it at all.