Seasonal Bouquets & Gifts 2026
Owen Group used AI image generation to produce photorealistic bouquet imagery for P&O Cruises and Cunard, working from florist stem lists alone: no photoshoots, no studio time, no compromise on quality.
Carnival UK needed seasonal bouquet and gift package imagery for their P&O Cruises and Cunard onboard retail channels. Three campaigns: Valentine's Day, Mother's Day, and Festive. Each bouquet had a precise specification from their professional florist, listing exact flower varieties, stem counts, foliage types, and packaging.
The challenge: produce imagery that matched these specifications closely enough to satisfy a professional florist's review, delivered in both P&O Cruises (blue background) and Cunard (white background) brand variants, plus gift package compositions. All without a single photoshoot.
Generate photorealistic flower arrangements from text-only stem lists, accurate enough to pass professional florist review across three seasonal campaigns.
P&O Cruises and Cunard guests browsing onboard gift and bouquet catalogues. Images needed to align with existing asset bank photography standards.
Deliver production-ready imagery across two brand identities and three seasonal themes, matching the quality and accuracy of traditional studio photography.
Owen translated detailed florist stem lists into structured AI prompts, specifying every variety, stem count, colour, foliage type, and packaging detail. The AI generated initial compositions, which were then reviewed by the client's professional florist for botanical accuracy. Feedback was precise: specific stem counts adjusted, flower varieties corrected by proper horticultural name, individual blooms repositioned. This iterative loop between AI generation and expert florist review produced final images indistinguishable from studio photography.
Where the AI occasionally resisted adding or removing a specific stem, traditional retouching filled the gap: a practical, hybrid approach that delivered nine final hero images across both brands without booking a single studio session.
The creative and production agents behind photorealistic AI imagery.
Creative Ignition. Takes validated territories and generates multiple creative routes before briefing design.
Translated raw florist stem lists into structured AI image prompts, capturing botanical accuracy, arrangement style, and brand-specific presentation.
Each bouquet specification ran to a dozen or more lines: exact flower varieties (Explorer Ecuadorian Roses, Zantedeschia Pink, White Asiatic Lilies), precise stem counts, supporting foliage, and packaging details. Spark converted these into layered prompts that controlled composition, colour balance, lighting, and that crucial "florist's hand" quality the client specified: natural flow, not rigid arrangements.
The prompts were iteratively refined based on the florist's feedback. When she noted that Calla Lilies should be referred to as Zantedeschia, or that the Peach Avalanche Roses needed six to seven visible rather than four, those corrections fed directly back into the prompt structure for the next generation pass.
Copy at Scale + Asset Reskinning. Same story, new wrapper, without briefing design every time.
Took a single validated bouquet composition and produced brand-variant outputs for both P&O Cruises and Cunard, plus gift package versions, without starting from scratch each time.
Once a bouquet composition was approved by the florist, Chameleon handled the brand adaptation. The same arrangement appeared on P&O's signature blue background and Cunard's clean white, with lighting and shadow adjustments to suit each treatment. CopyMate ensured all accompanying product descriptions matched the visual output for catalogue consistency.
The gift package versions layered in prosecco bottles, chocolate boxes, and branded packaging elements around the core bouquet: a composition task that would traditionally require a separate styled photoshoot for each combination.
QA: Production. Regulatory and brand compliance checking across every asset.
Cross-referenced every generated image against the original stem list, checking flower variety accuracy, stem counts, and botanical correctness before client review.
Before any image reached the client's florist, Sentinel ran a compliance pass against the original specification. Stem counts were verified. Flower varieties were checked against proper horticultural names (not the common names used in the initial brief). Background colour consistency was validated across brand variants.
This pre-flight check caught discrepancies early, reducing the number of florist review rounds and accelerating the approval cycle.
The images below show the client's existing professionally photographed assets alongside the AI-generated equivalents Owen produced from stem lists alone. The AI imagery was reviewed and refined by the client's own professional florist, who provided stem-level feedback on botanical accuracy, flower variety identification, and arrangement composition.
Professionally photographed in studio. This is the benchmark our AI imagery was measured against.
AI-generated from a florist's stem list. No photoshoot. Reviewed and approved by the client's professional florist.
Existing asset bank photography. The standard our AI output needed to match for onboard retail consistency.
AI-generated Mother's Day bouquet on P&O blue. Zantedeschia Pink, Peach Avalanche Roses, White Hydrangea: every stem specified, every bloom placed.
Each campaign was delivered in three variants: P&O Cruises (blue background), Cunard (white background), and a gift package composition featuring prosecco and chocolates. Blue backgrounds serve P&O Cruises. White backgrounds serve Cunard. All generated from the same underlying stem list.
12 Explorer Ecuadorian Roses, Cinerea Eucalyptus, Pale Pink Genista, Mini Cymbidium Orchids, Calla Lily Red ×5, dark red and black packaging.
Zantedeschia Pink ×3, Lily Hybrid Pink ×3, Peach Avalanche Roses ×9, Pink Picotee Lisianthus ×5, White Hydrangea ×5, Ranunculus White ×5, kraft paper and satin ribbon.
Red Amaryllis, Red Roses ×5, White Lily ×3, Hypericum White ×3, Carnations Green ×3, Ilex Berries, Pussy Willow, Pine Cones, Cinnamon Sticks, Nobilis Pine, and Eucalyptus Gold.
The real test: a professional florist reviewed every image and gave stem-level feedback on botanical accuracy. The corrections were specific: exact stem counts, proper horticultural names, individual bloom placement. That is the standard AI imagery needs to meet, and this project met it.
From stem list to production-ready imagery, in a fraction of the time and cost of traditional photoshoots.