BOBrian Olsen
AI-assisted product image production

Commercial product visuals built from imperfect real-world source photos.

I combine hands-on photography, Photoshop retouching, AI-assisted image generation, and human-led quality control to turn casual product photos into polished ecommerce, campaign, and presentation-ready assets.

View case studies See workflow
Product truth firstShape, materials, label structure, brand identity, and detail fidelity stay central.
AI heavy liftingBackground cleanup, reconstruction, lighting exploration, and first-pass commercial polish.
Photoshop finishingMasking, compositing, cleanup, small-detail correction, and production-ready refinement.
Creative-director QAReview for realism, packaging truth, typography risk, and brand-safe execution.

Selected work

Each example shows the production problem, the transformation, and the QA logic behind the final image.

CPG packaging repair

AI-assisted reconstruction CPG / food packaging Ecommerce hero

A torn, cluttered iPhone photo of a candy package was transformed into a clean front-facing product image. The key production challenge was not merely removing damage — it was restoring correct front-package logic and removing exposed back-label content that did not belong on the front design.

Original damaged Twizzlers package photo on a cluttered tabletop.
Sourcetorn wrapper / cluttered phone photo
Intermediate AI repair of Twizzlers package with incorrect nutrition facts still visible on the front.
QA catchpolished but packaging logic was wrong
Final repaired Twizzlers package on a clean ecommerce background.
Finalcorrected ecommerce product image
Production lesson: AI can create a polished image that is still commercially inaccurate. Human QA caught the exposed nutrition/back-label information and redirected the edit toward a truthful intact front package.

Objective

Repair a damaged branded package photo and create a clean ecommerce-ready hero image.

Source problem

Torn front wrapper, fingers in frame, distorted packaging shape, exposed back-label content, cluttered tabletop, and uneven phone lighting.

Workflow

AI-assisted wrapper reconstruction, background cleanup, damage removal, lighting correction, and iterative QA prompts.

Fidelity / QA

Preserved logo structure, red glossy wrapper, strawberries, candy window, white crimped ends, wrinkles, and package proportions. Small text would require manual production review.

Premium wearable spec hero

Spec concept Reflective product Reference-based QA

A casual iPhone photo of a smart ring was transformed into a premium wearable-tech hero image. A separate closeup reference was used to preserve the real block-style inner sensor, polished inner band, front notch, and product geometry.

Original black smart ring photo on a brass surface with cluttered background.
Sourcecluttered phone photo
Closeup of smart ring interior showing block-style sensor and inner markings.
Detail referencesensor / interior fidelity
Final premium smart ring spec product hero image.
Finalpremium wearable spec hero
Disclosure: Spec concept only. Not affiliated with Oura. Brand-inspired elements are shown only to demonstrate product-fidelity workflow, art direction, and premium wearable-tech presentation.

Objective

Create a premium commercial hero image from a poor iPhone product photo while preserving product-specific details.

Source problem

Mixed lighting, reflective brass surface, background clutter, visible wear, noisy phone detail, and no campaign composition.

Workflow

AI-assisted isolation, seamless studio background, lighting refinement, material correction, sensor-detail QA, and composition refinement.

Fidelity / QA

Preserved matte black exterior, polished inner band, front notch, block-style inner sensor, inner markings, oval perspective, and grounding shadow.

Real commercial food photography

Hands-on photography Manual masking Pre-AI production

Before adopting AI-assisted image workflows, I photographed core Mister Softee SoCal menu/product assets on location using a portable white-box setup, then manually masked and prepared the images for menu and display use.

Mister Softee SoCal menu display screenshot showing product/menu imagery.
Published contextmenu/product imagery usage reference
Production note: Original RAW files are available. The final menu board layout/prints were not designed by me; my role was photographing and preparing core product imagery.

Objective

Create clean product/menu images for real business-facing use.

Source problem

Images were captured in a non-studio environment, requiring practical lighting control and clean product isolation.

Workflow

Portable white-box capture, RAW workflow, manual masking, background cleanup, and Photoshop retouching.

Relevance

Demonstrates hands-on photography, masking, retouching, and commercial asset preparation before AI entered the workflow.

Fast ecommerce cleanup tests

Product isolation Background cleanup Lighting correction

Simple product-photo transformations used to test speed, cleanup, grounding shadows, and ecommerce presentation from ordinary iPhone source photos.

Original coffee cup iPhone photo in a cluttered room environment.
Sourcecoffee cup / cluttered room
Clean studio-style coffee cup product image.
Finalstudio-style product image
Original weekly pill organizer on wrinkled fabric.
Sourcepill organizer / wrinkled fabric
Clean ecommerce-style weekly pill organizer image.
Finalclean ecommerce hero

Objective

Convert casual iPhone shots into clean product-forward ecommerce images.

Source problem

Distracting backgrounds, mixed lighting, phone perspective, clutter, and non-commercial composition.

Workflow

AI-assisted background replacement, lighting cleanup, perspective refinement, product isolation, and shadow correction.

Fidelity / QA

Checked shape, material behavior, transparency, label/detail risk, grounding shadows, and whether the final still looked photographed.

Workflow

The goal is not a magic prompt. The goal is a repeatable production process with human oversight at every point where AI can drift from product truth.

1

Product truth analysis

Identify what must be preserved: geometry, material, label layout, logo placement, text areas, packaging structure, and detail references.

2

AI production pass

Use AI for heavy reconstruction: background cleanup, lighting exploration, damaged surface repair, ecommerce setup, and first-pass polish.

3

Creative QA

Review for hallucinated text, warped logos, false packaging logic, impossible reflections, floating shadows, and over-smoothed material behavior.

4

Manual finishing

Use Photoshop for final detail work: masking, clone/repair, type-area cleanup, perspective fixes, label accuracy, color correction, and delivery prep.

Relevant background

I have hands-on Photoshop retouching experience from commercial field-photo correction, including compositing missing signage, duplicating product facings, repairing display photos, and making images meet business requirements.

I now use AI-assisted tools to accelerate the heavy-lift stages of product image production while relying on Photoshop skill and human QA for the details AI cannot be trusted with.

Production strengths

  • Product photography and practical lighting in imperfect environments
  • Manual masking, background cleanup, and Photoshop compositing
  • AI-assisted image generation, product repair, and background reconstruction
  • CPG package fidelity, label-risk awareness, and brand-safe review
  • Reflective, glossy, matte, transparent, and flexible packaging material QA
General disclosure: Some examples are speculative proof-of-concept exercises using personally captured source photos. Brand names/logos are shown only to demonstrate product-fidelity workflows and are not presented as official brand work unless explicitly noted. Final client delivery would include manual review for label accuracy, trademark compliance, small text fidelity, and production readiness.

Contact

Brian Olsen
Oxnard, California
Product photography, retouching, and AI-assisted commercial image production

brian@olsenautomation.com
(626) 524-8156
LinkedIn Profile