Home Chef Greatest Hits | Art Direction

Concept 4 - Naan, Cuisine _ Flavor Profile_ Indian.jpg

PROJECT

Using targeted recipe imagery to power a customer survey that helped Home Chef identify and bring back its most-loved meals.

ROLE

Senior Photographer, in-house

COMPANY

Home Chef

STAKEHOLDER(s)

Insights Team,
Menu Strategy Team

SCOPE

Generative AI image creation and art direction using Adobe Firefly
Reference image curation from existing archive
Prompt development and refinement
Quality review and final selection across all concepts

The Problem

The Insights and Menu Strategy teams needed new imagery that could isolate ingredient and meal concept variables visually. The same dish, shot consistently, with only the key ingredient changed between versions. Clean, controlled, and convincing enough to feel real to a survey respondent.

The challenge was producing that volume of controlled imagery quickly and cost-effectively, without running a full photo shoot for every variation.

376021Crispy Fig Chicken - with blue cheese mashed potatoes_ecomm2.jpg

The creative solution was generative AI, specifically Adobe Firefly (with Google Gemini as the model), used as a production tool under deliberate art direction rather than as a shortcut.

Photo Art Direction

Concept 3 - Naan, Protein Swaps_ Sliced Beef.jpg

For each variant, I worked from existing reference imagery in the Home Chef archive, built prompts in Adobe Firefly (Model: Google Gemini) that isolated the specific ingredient change, evaluated outputs against the visual standard of the control image, and made art direction decisions about what was acceptable, what needed refinement, and what needed to be regenerated.

The final selections had to be believable as Home Chef photography before they went anywhere near a customer survey.

CONTROL IMAGE

Concept 1 - Skillets, Protein Swaps_Ground Turkey.jpg

The approach was methodical. Starting from Home Chef's existing image library, I identified reference photos that matched the plating style, lighting, and framing standards of each Greatest Hits meal. Each concept had a control image and multiple variants, giving the Insights team a clean set of comparable visuals for the survey.

PROMPT:

Using image 1 as the reference, replace the meat in image 1 with the sliced chicken from image 2

Reference Image

FINAL IMAGE

Concept 3 - Naan, Protein Swaps_ Chicken Breasts.jpg

CONTROL IMAGE

PROMPT:

Using image 1 as the reference, replace the ingredients in the bowl with the meat/veg mixture from image 2, topping with a drizzle of each sauce from image 3

Reference Image 2

Reference Image 3

FINAL IMAGE

Concept 2 - Skillets, CuisineFlavor Profile_Argentinian.jpg

The goal was visual consistency across control and variant images so that any difference a survey respondent perceived came from the ingredient change, not from inconsistencies in how the food was photographed or styled. That level of control required real art direction judgment, not just prompt generation.

The Results

The project delivered a full set of survey-ready imagery across four meal concepts and multiple protein and cuisine variants, produced without a single additional shoot day. The Insights team had the visual assets they needed to run a controlled customer survey testing how ingredient changes affect craveability and cart behavior, with results that could directly inform future menu decisions.

The project also established a repeatable workflow for using generative AI as a production tool within Home Chef's existing visual standards, with art direction embedded at every step to maintain quality and brand consistency.