W Creative / Self-Initiated Product
Year
2026
What I did:
- Product Design
- UX Design
- Workflow Design
- AI-Assisted Development
- Visual Design
- Prompt System Design
- AI-Assisted Development / Vibe Coding
Deliverables
- Functional Web Application
- UX Workflow
- Interface Components
- Structured Prompt System
- Generated Product Scenes
Creating Consistent AI Product Scenes With Less Prompt Work
Creating a good AI product mockup was easy. Creating a repeatable set of accurate, consistent ecommerce images was not.
Existing tools produced attractive results, but product placement, scale, composition, and scene details could drift between generations. Reducing that variation required increasingly detailed prompts, turning what should have been a quick task into a repetitive production process.
At the same time, evolving AI disclosure requirements created another consideration for sellers using generated people or partial human imagery. I wanted a workflow that could reliably produce polished, human-free product imagery while giving users more control over the final scene.
Ways of Working
Tools & Methods
I started by taking inventory of each step in my workflow, identifying where things slowed down or drifted, and asking: How might I make the process faster, more repeatable, and easier for someone else to use?
I evaluated several approaches before building. Manus offered agent-based automation, but unpredictable credit usage and the setup required from each user made it difficult to turn into a consistent product. Canva Code was also considered, but the workflow needed more control and independence than I wanted to place inside a design platform.
I chose Base44 for rapid development and standalone deployment, using ChatGPT to help define requirements, evaluate approaches, structure prompt logic, troubleshoot, test edge cases, and refine the interaction.
I also used lightweight community testing rather than designing only for myself. I shared working tools and workflows with members of the ecommerce creator community, asked them to complete the process independently, and used their questions and friction points to inform subsequent iterations.
My process was iterative: define → build → test with users → refine → repeat.
Tools: ChatGPT · Base44 · Canva · AI image generation
Solution
Building The Generator
I designed and built Object Scene Studio, a web-based product imagery tool that moves the complexity of detailed prompting behind a structured interface.
Users can create styled lifestyle scenes, clean studio images, isolated or floating product presentations, and pattern previews while controlling product details, composition, lighting, environment, and human presence.
Instead of rebuilding a complex prompt for every image, users make their choices once through a guided workflow and the system constructs the generation instructions.
Project Outcome
Streamlined Workflow
Object Scene Studio is now part of my regular ecommerce workflow. In one recent production task, a process that took about 20 minutes through conversational prompting took about 2 minutes using the structured workflow. It has reduced the back-and-forth I spend refining prompts, correcting drift, and recreating product imagery.
Now Available
Object Scene Studio is live and continues to evolve with new features, including studio-style product photography and pattern simulation.
View Object Scene StudioView Object Scene Studio