End to End AI Lifestyle Workflow. Metrovac

As an Art Director, I’m responsible for creating guidelines and processes that improve how the team works and collaborates. A key part of my role is automating time-consuming tasks and workflows, allowing us to focus more energy on projects that truly demand deeper creative and strategic thinking. 

I’m constantly driven to refine and improve processes, not only to keep learning, but also to share that knowledge with others the methods, insights, and experiences that have helped me consistently deliver optimal results.

That’s why I’m sharing my workflow. I hope you find it useful and inspiring. 

Credits: Brand: Metrovac

Product: IT Vacuums and Blowers

Project: Realistic Lifestyle Product Images 

Media Channel: Amazon Listing 

Art Director: Mark Zuniga 

AI Prompt Designer: Mark Zuniga Generative 

AI Tool: Freepik 

Project Completion Date: 202

END TO END AI LIFESTYLE IMAGE. PRODUCTION WORKFLOW

Documentation of the full process: folder structure, reference selection, prompt building, generation, editing/retouching, and final client delivery.

Standardized Project Folder Structure

Standardized organization for storing briefs, working PSDs, resized/output files, and final deliverables making it easy to access each stage of the process.

Final Deliverables (Exported Files)

Export-ready results for delivery: final JPG/PNG/TIFF as requested, correct resolution, final naming convention, and approved versions only. No working files just finished outputs.

Project Intake & Requirements

Store everything received at the start: client brief, specs, required dimensions (e.g., 4000×2093), style references, deliverables list, naming/SKUs, notes, deadlines, and feedback history.

AI Generation & Retouch (Working PSDs)

Your “work-in-progress” PSDs where you fix what the AI didn’t get right: edge cleanup, shadows, deformations, weird textures, artifact removal, color correction, compositing, and any manual refinements. Fully editable files live here.

Format Adaptation (Resized/Output PSDs)

PSDs that are already cleaned up but adjusted to the requested format: reframing, background extension, layout alignment for 16:9 or custom ratios, safe margins, and consistency across the full set. This is the final PSD stage before export.

Reference Image Selection (Starting Point)

Examples of the images we chose to guide the AI from the start and keep product, background, pose, and style consistent.

Product reference preparation

To give the AI a strong product reference, we removed unnecessary elements from the original image. This helps the model focus on the essential product features and work more intelligently instead of copying the photo exactly. In some cases, a direct copy wouldn’t work because certain angles or elements may not fit the final composition.

Background guidance

To support the AI when generating backgrounds, we sometimes provided empty backgrounds with no people. This allowed the AI to build the environment more creatively, while we controlled the context and level of detail through the prompt.

Pose, models, and animals

For posture, human models, and dogs, we also supplied reference images so the AI had a closer visual target. This included specifying the exact dog breed we wanted to appear in the final result.

Prompt: Reference Mapping (Image → Role)

Each reference has a defined role and is cited in the prompt using @image # to control composition, camera angle, lighting, and realistic integration.

FREEPIK generation (Model + Output Format)

Screenshot of the Freepik Image Generator platform showing generated variations, the prompt, the selected model, and the output format/resolution settings for export.

Available Formats by AI Model

Comparison chart showing the supported formats and resolutions for each model to help identify the closest match to the client’s requirements. The selected format is highlighted in green. When using the Nano Banana or Seedream model, it provides the closest output size to the *client-requested delivery dimensions.

Final Results: Selection, Retouch, and Delivery

Shows the selected image (closest to the expected result), the marked areas requiring retouching in an editing program, and the final approved version delivered to the client.

Additional AI-Generated Variations (Lifestyle + Product in the Foreground)

Samples of additional AI-generated images for this project, keeping a lifestyle look with the product in the foreground, in use, and clearly visible.