automated content pipeline: a centralized system for generating product imagery and automating social media distribution. (WIP)

project description

A backend workflow that manages the entire content lifecycle for DIY papercraft sculptures. It extracts styling data from reference photos, renders the product into those environments, and automatically schedules the final images for social media publishing. 

the problem

Creating and publishing daily marketing content is repetitive and time-consuming. Manually generating mockups, managing files, writing captions, and scheduling posts drains hours of creative time. The process needed to be centralized so one database could handle both asset creation and distribution. 

the architecture

The system replaces manual data entry and posting with a unified Airtable database connected via n8n:

• data ingestion: Dropping a reference photo into Google Drive triggers an API to extract layout and lighting details into structured text.

• database assembly: Airtable categorizes this data to build reusable room presets.

• asset generation: The system sends a structured request to ComfyUI, placing the product into the selected environment. Relying on the trained data of the image model so as to get the style and a ControlNet with an image made in Blender so as to guide the composition. The final render saves directly back to the database.

• social distribution: Once an image is marked "Approved," a secondary n8n workflow pulls the file, formats the caption based on database tags, and publishes it to social media on a set schedule.

core takeaway

Treating marketing as a system rather than a daily chore saves hours of manual work. Centralizing generation and distribution into one automated database shifts the focus from repetitive posting to actual design and strategy. 

current status & next steps

• phase 1 complete (data ingestion): The front-end automation is fully operational. Reference images dropped into Google Drive are successfully parsed, structured into JSON, and routed directly into the Airtable database. The base rendering pipeline (ControlNet with a refinement pass) is built and functional.

• infrastructure expansion: The immediate next step is deploying the rendering engine to a cloud server (RunPod or Modal) for headless execution. From there, the n8n workflows will be expanded to trigger the cloud renders and handle the final social media publishing sequence.

• visual pipeline upgrades: To achieve exact 1:1 product accuracy, the rendering workflow will be upgraded by either training custom LoRAs or implementing a localized compositing step (Qwen Edit) to perfectly place the specific physical product into the generated mockups. Finally, an automated color correction pass will be integrated to standardize the final marketing assets.