Automating a Product-Catalog Workflow for Hundreds of SKUs

E-commerce brands with large catalogs face an acute production bottleneck: every SKU needs at least one high-quality product image, often multiple images showing different angles, environments, and lifestyle contexts. Traditional photo shoots require physical product samples, studio time, photography equipment, and post-production editing. For a catalog of 200 SKUs, this means either a massive upfront investment or a perpetual backlog of unimaged products.

AI product photography workflows change the economics fundamentally. Instead of shooting every product individually, you define the visual style once — background color, lighting setup, composition rules, brand aesthetic — and then apply that defined style to every product image automatically. The workflow is the shoot. Once the pipeline is built and validated on a handful of products, scaling to 200 or 2,000 SKUs requires only supplying new input images and triggering a batch run. The creative and technical decisions are made once and reused everywhere.

A production-ready product-catalog workflow in the Floniks /editor typically has four layers. The input layer accepts the raw product image for each SKU — either a simple photo against a plain background or a raw shot straight from a mobile device. This is the only node that changes per SKU.

The background-replacement layer removes the original background and composites the product onto the branded background defined in your style configuration — a seamless white, a textured surface, a lifestyle environment, or a color gradient matching your brand identity. The lighting-enhancement layer applies consistent studio lighting to the composited image: fill light, highlight, and shadow that make the product look professionally lit regardless of the original shot’s lighting conditions. The quality-finalization layer runs upscaling and sharpening to bring every output to a consistent resolution and crispness, ensuring uniform quality across the entire catalog. Each layer is a node; each node’s output feeds the next.

Start in the Floniks /editor by adding a batch input node configured to accept product images. This node is the entry point for each SKU. Connect it to a background-removal node, which isolates the product from its original background and outputs a masked product image with a transparent background.

Wire the background-removal node’s output to a background-composition node. Configure this node with your brand’s background — upload a background image or specify a color and texture. The composition node places the isolated product onto the new background with appropriate scale and positioning. Connect the composition node to a lighting-enhancement node configured with your brand’s lighting style. Finally, wire the lighting node’s output to an upscaling node that brings every image to your target catalog resolution. Connect the upscaling node’s output to a batch output collection node. Test the complete pipeline on three to five representative SKUs before scaling to the full catalog.

The visual consistency of your catalog depends entirely on how precisely you define the style configuration in each workflow node. The background-composition node should specify: background color or image, product placement (centered, lower-third, corner), product scale as a percentage of frame width, and shadow type (drop shadow, reflection, none). Vague background configuration produces inconsistent product scale and placement that is immediately obvious when catalog images are viewed together.

The lighting-enhancement node configuration should specify: main light direction (front, 45-degree, side), fill-light intensity, highlight strength, and shadow depth. For most e-commerce catalogs, front-weighted three-point lighting with soft shadows is the standard that reads most clearly at small catalog thumbnail sizes. Specify your lighting configuration in the node settings and lock it — do not vary it between SKUs unless you have a deliberate visual strategy for doing so.

Once your workflow is configured and validated on a test set, trigger a batch run by uploading all SKU images to the batch input node at once. The workflow engine fans out the pipeline across all inputs, running multiple SKUs in parallel batches. Monitor the run in the /editor dashboard — the execution log shows which SKUs are processing, completed, or encountered errors.

After the batch completes, review outputs using a systematic quality-check protocol. For every 50 SKUs, manually inspect 5–10 outputs for: (1) background-removal accuracy at product edges and transparent materials, (2) consistent product scale across different product sizes, (3) lighting uniformity across different product colors and surfaces, and (4) sharpness and resolution at the target export size. If systematic issues appear — for example, a category of products with curved edges consistently shows background-removal artifacts — adjust the relevant node configuration and re-run only that batch segment.

Once your baseline product-catalog workflow is validated, save it as a named template in /editor. This template becomes a reusable asset: any team member can instantiate it, supply new product images, and run a consistent pipeline without understanding the node topology. Templates also make it straightforward to create variant workflows — for example, a "lifestyle context" variant that places products in real-world environments rather than on studio backgrounds, or a "seasonal" variant that swaps backgrounds for holiday themes.

For catalog refresh cycles (new season, new products, updated brand guidelines), update the relevant nodes in the template once and all future runs inherit the update. This is the compounding advantage of the workflow approach: as your brand’s visual standards evolve, you update one workflow and every future production run reflects the change. The alternative — re-briefing a photographer or photo editor on updated brand guidelines for every new shoot — is slower, more expensive, and less consistent.