A Seasonal Banner-Set Workflow

A seasonal campaign that launches across six channels needs banners in at least a dozen distinct dimensions: a 1920x600 website hero, a 1440x400 secondary web banner, a 600x200 email header, a 1200x628 Facebook Open Graph image, a 1080x1080 Instagram feed square, a 1080x1350 Instagram portrait, a 1080x1920 Story frame, a 1500x500 Twitter/X header, a 1128x191 LinkedIn cover, and various Google Display Network sizes (300x250, 728x90, 160x600). Producing all of these individually from scratch would require each to be designed separately, which means each is an opportunity for the seasonal theme to drift — a slightly different shade of the "winter" palette here, a slightly different typographic weight there.

The traditional solution is to design one master banner and then scale and crop derivatives from it. But a single master rarely contains the right compositional elements at every aspect ratio. A landscape master with a cozy fireplace scene on the left and product imagery on the right will not crop well to a square format without losing either the fireplace atmosphere or the product. A portrait master built for Stories will not crop to a panoramic website hero without the scene feeling claustrophobic and cropped.

The Floniks seasonal banner-set workflow takes a different approach: it generates a style seed that establishes the seasonal visual identity without being locked to any specific composition, then generates each banner format independently while being constrained by the shared style seed. The result is banners that are visually unified at the theme level but compositionally optimized for each format.

The style seed is a single image generated at a square 1:1 format that captures the essence of the seasonal theme without committing to any specific compositional layout. Think of it as a mood board compressed into a single reference image. For a winter holiday campaign, the seed prompt might be: "warm holiday atmosphere, deep evergreen and burgundy color palette, soft golden candlelight and fairy light bokeh, snow-dusted pine branches in the foreground, warm interior ambient light in the background, luxury editorial aesthetic, no products, no text, no human subjects." For a summer campaign: "bright coastal energy, aquamarine and coral color palette, natural sunlight with soft shadows, beach grass and warm sand texture in foreground, golden hour light, clean and airy editorial aesthetic."

The style seed generation node runs first, before any other banner generation begins. Its output is the visual reference that constrains all downstream banner generation. Connect the style seed output to a Style Extractor node that analyzes the image and produces a structured style descriptor: dominant colors (as hex codes), lighting direction, color temperature, atmospheric mood, and a compressed style embedding that can be used as image conditioning for downstream generation nodes. This structured descriptor travels through the workflow alongside the style seed image itself.

Review the style seed before running the downstream banner generation. The seed should feel unmistakably seasonal without being literal or cliche. A winter holiday seed should feel warm and celebratory without resorting to clip-art snowflakes or generic tinsel imagery. If the seed output does not capture the intended seasonal feel, adjust the prompt and regenerate before triggering the full banner set run. Spending an extra few minutes on the seed saves significant rework downstream because every banner in the set is conditioned on it.

With the style seed reviewed and approved, the workflow branches into parallel generation paths — one per banner format. Each path uses a Banner Generation node configured for the target dimensions and aspect ratio. The style seed image is connected as a conditioning input to each Banner Generation node with an image strength of 0.3 to 0.4 — sufficient to enforce the color palette, lighting direction, and atmospheric mood while allowing the composition to be optimized for each specific aspect ratio.

Each Banner Generation node receives a format-specific prompt that specifies the compositional needs of that format. The wide landscape web hero (1920x600) prompt emphasizes horizontal composition: "wide cinematic scene, seasonal atmospheric elements distributed across horizontal canvas, focal product area in right third, open space for headline text overlay in left third, same color palette and lighting as style reference." The square Instagram format prompt emphasizes centered balance: "square composition, seasonal atmosphere, centered focal arrangement, equal visual weight left and right, same color palette and lighting." The portrait Story format prompt emphasizes vertical energy: "vertical composition, seasonal environment flowing from top to bottom, product or focal area in center vertical third, visual momentum flowing downward, same color palette and lighting."

For banners that need headline text and call-to-action overlays, add a Text Overlay node downstream of each Banner Generation node. The text configuration is stored in a shared Typography Config node that feeds every Text Overlay node identically — same font, same weight, same color, same hierarchy rules. Only the text content (headline copy and CTA text) varies per banner. The text content is drawn from a Campaign Copy node that stores all approved headline and CTA variants for the seasonal campaign.

Open /editor and create a new workflow. Add a Campaign Config node as the root. Enter the seasonal theme description, campaign name, target channels, and the approved headline and CTA text variants. Connect the Campaign Config to a Style Seed Generation node. Enter the style seed prompt for the seasonal theme and configure the output at 1080x1080 pixels. Add a Style Extractor node downstream of the seed generation. Configure it to output dominant hex colors, lighting direction, color temperature, and a style embedding.

Add one Banner Generation node per required format. For each node, set the target dimensions, write the format-specific compositional prompt, and connect the Style Seed output as a conditioning image at strength 0.35. Connect the Style Extractor color output to a Color Palette Constraint on each generation node to enforce the extracted palette. Add a Text Overlay node downstream of each Banner Generation node, connected to the Campaign Config for copy and to a shared Typography Config for formatting.

Connect all Text Overlay outputs to a Format Export node configured with the correct encoding for each channel: JPEG quality 92 for web and social, PNG for ad networks that require lossless, and WebP for modern web deployments. Add a Naming Convention node that generates filenames in the format "[campaign-name]-[format-label]-[dimensions].[ext]." Run the complete workflow and review the outputs across all formats side by side. If any format deviates significantly from the seasonal theme, adjust the format-specific prompt on that branch and re-run that branch only. Save the workflow as a template named "SeasonalBanners-[Season]-[Year]-v1" for use in all future seasonal campaigns of the same type.