Capstone: Automating a Real Batch-Retouching Workflow

Raster Editing Fundamentals' own capstone worked one photo, end to end, by hand — and named this course's own capstone directly as where that exact discipline scales up to a real production batch, automated via scripting. This capstone does exactly that: build one proven retouching recipe on a single reference photo, then automate it across an entire batch of product photos, using every technique from Chapters 2 through 9.

The Scenario

A client has delivered 60 raw product photos, each shot against a cluttered studio background, needing: the product masked out cleanly (including some fine, hair-like fringe on a fabric sample), composited onto the client's own consistent branded background, surface texture retouched to remove shipping-handling marks without losing real material texture, a subtle branded filter look applied consistently, and every file watermarked and exported — with all 60 files needing to go out the same day.

Step 1 — Building the Mask on a Reference Photo (Chapter 2)

Working on one representative photo first, build a precise channel-based mask around the product, using whichever color channel shows the strongest contrast against the studio background, refined enough to hold up against the fabric sample's own fine fringe detail.

Step 2 — Compositing Onto the Branded Background (Chapter 3)

Composite the masked product onto the client's branded background, matching color and lighting direction between the two, and feathering the mask's edge slightly so the fringe detail preserved in Step 1 reads as naturally photographed rather than pasted on.

Step 3 — Non-Destructive Retouching and the Brand's Filter Look (Chapters 4 & 8)

Convert the composited product layer to a Smart Object before applying the brand's signature filter stack — a specific, previously agreed sharpen-then-subtle-vignette combination — so the whole stack stays reopenable and adjustable per photo if the client requests a tweak later, rather than being baked in destructively.

Step 4 — Frequency-Separation Cleanup (Chapters 5 & 9)

Remove the shipping-handling marks from the product's surface using frequency separation, correcting tone on the Low Frequency layer without touching the real material texture on the High Frequency layer. In GIMP, this step uses the Wavelet Decompose script installed back in Chapter 9 to build the frequency layers, rather than assembling them by hand each time.

Step 5 — Recording the Recipe (Chapter 6)

With the full sequence proven on this one reference photo, record it as a Photoshop Action, or write it as a GIMP script calling the same sequence of PDB operations — turning a manually-proven workflow into something repeatable without re-performing every step by hand on each of the 60 files.

Step 6 — Running the Batch (Chapter 7)

Test the recorded Action or script against five representative files first, per Chapter 7's own tip — confirming the mask, composite, filter stack, and frequency-separation cleanup all hold up correctly before committing to the full run. Then run the Batch command (or the equivalent headless GIMP command-line invocation) across the remaining files.

Step 7 — Catching the Files That Didn't Match Assumptions (Chapters 6 & 7)

Two of the 60 files were shot at a different angle than the rest, and the recorded recipe's masking step doesn't hold up cleanly on them — exactly the silent-failure risk both Chapter 6 and Chapter 7 warned about. Because the batch run was logged, these two files are caught and flagged rather than shipped incorrectly, and are corrected individually by hand before final delivery.

Step 8 — Final Export

Export all 60 finished files — the corrected two included — to the client's required format and naming convention, with the fully-layered working files for each preserved separately in case of a future revision request.

Hands-On Exercises