2D Image Generator
Product Design
COMPANY
Replica Studios
PRODUCT
2D Image Generator (Game Studio)
ROLE
Lead Product Designer ✦ Product Owner
YEAR
2024 – 2025
Creating high-quality 2D game assets remains one of the biggest time and resource bottlenecks for small studios and indie developers. Existing solutions often fall short: they are either too rigid, too technical, or too slow to scale across diverse art styles and game genres.
Industry data shows that game developers leveraging generative AI for 2D asset creation can reduce their time to market by up to 30%, while cutting concept art production time by 40-60%*. This efficiency gain translates directly into faster prototyping, reduced costs, and greater creative freedom, which are critical to small and medium-sized studios competing in fast-moving markets.
We set out to build a tool that would change that. An intelligent, fast, and visually expressive generation system that empowers developers to create production-ready 2D assets using natural language.
2D Image Generator, part of our Game Studio suite (alongside SFX and Localisation), was our answer. A prompt-driven generation tool that made concepting and producing game visuals as easy as describing them.
User viewing generations flow and interactions
Users Assset list showing 2D Image generations
MY ROLE
As Lead Product Designer, I was responsible for both the product strategy and delivery of the MVP, from early research and competitor analysis through to interface design, prototyping, and technical build.
In this role, I led the end-to-end product design and delivery of the MVP, including:
Conducting product and UX research with Game Studio customers to understand pain points and workflows
Designing the interface, interactions, and creative flow using insights from user interviews, market gaps, and competitive analysis
Vibe-coding the MVP leveraging v0 along with OpenAI and fal.ai APIs, rapidly iterating and testing hypotheses
Delivering a fully functional MVP that was handed off to engineering for integration with authentication, credit tracking, and backend infrastructure
DESIGN APPROACH
The 2D Image Generator was designed to be accessible yet powerful, with a focus on creative control, minimal friction, and game-production practicality.
✦ Prompt-First Generation, Enhanced Automatically ✦
Users begin by describing what they need, for example, "battle worn robot mech that is blue and gold" Behind the scenes, our system enriches that prompt with AI to increase clarity and output fidelity. This allowed users of all experience levels to get high-quality results without perfect prompt engineering.
✦ Type-Aware Generation ✦
Users can generate from five core asset types, each with structured subtypes:
Characters (hero, enemy, NPC, robot, creature)
Items (weapons, props, tools, wearables, etc.)
Backgrounds (interiors, landscapes, fantasy, abstract, etc.)
Textures (tileable patterns, terrain, fabric)
Each category also supported automatic type detection based on the input prompt, streamlining workflows for users who wanted to move quickly.
✦ Style, Theme and Palette Control ✦
To support creative consistency across asset libraries, users could fine-tune:
Art Styles (Pixel, Cartoon, Realistic, or Auto)
Themes (Fantasy, Sci-Fi, Steampunk, Casual Fun, Cyberpunk, and more)
Color Palettes (Vibrant, Dark, Pastel, Mystic, Retro)
This made it easy to match existing game art directions or explore entirely new ones.
✦ Asset Management and Export ✦
The generator included a full asset browser with:
Filtering by type, style, theme, and palette
Multi-select and bulk actions such as delete, download, and regenerate
Download single assets or batch export as ZIP
These features turned a generative playground into a serious production tool, ready for integration into any custom pipeline.
Asset view interactions
User generated asset
Empty state with quick start generation prompts
Generation prompt options and settings
OUTCOME + REFLECTION
The MVP shipped rapidly and validated the product vision on multiple fronts, delivering measurable value and shaping our longer-term strategy:
Rapid development and launch: Designed, coded, and deployed a fully functional MVP in just one week leveraging v0, OpenAI, and fal.ai APIs, demonstrating the power of tight integration between AI services and agile design processes.
Strong early validation: Initial user testing across internal teams and select external partners confirmed robust demand, with users praising the generation quality, ease of use, and creative flexibility. This feedback validated the core UX flow and AI output fidelity.
Informed backend integration and scalability: MVP insights directly influenced plans to integrate the generator with Game Studio’s credit system, user management, and backend infrastructure, ensuring a seamless experience for a broader user base and enabling sustainable scaling.
Strategic proof of concept: Beyond immediate deliverables, the project provided compelling evidence for the viability and value of AI-assisted asset pipelines, influencing our broader AI product roadmap and accelerating investment in generative tooling across multiple domains within the studio.
Catalyst for innovation culture: The success of the MVP fostered enthusiasm and confidence in leveraging AI creatively within the product team, inspiring cross-functional collaboration and a more experimental mindset around future game development tools.
Expected real-world application: We anticipate that both internal creative teams and external developers will adopt the tool for rapid prototyping, visual pitch decks, and early-stage production workflows, underscoring its potential to move beyond experimentation into practical, everyday use.
2D Image Generator was one of the most hands-on, full-stack design experiences of my career. I was not just shaping the UX but prototyping, coding, and shipping the actual product.
This allowed me to test ideas in real time, balance creativity with constraints, and align the design direction with both business goals and technical realities. It also ensured we were building something real developers could use immediately, not just a concept on paper.
Ultimately, this product expanded the creative toolkit of our users, bringing speed, flexibility, and AI augmentation to one of the hardest parts of game development.
* Game asset creation with generative AI can reduce time to market by up to 30%, and cut concept art production time by 40 to 60 percent. Source: Grid Dynamics, Game Asset Creation With Generative AI →