Self-paced Intermediate Creative Cloud 2026

Photoshop AI and Generative Fill

A working day on Generative Fill and Firefly features for people who ship real artwork

$99.00

per person, including materials

Start the moment you buy it, work at your own pace, and keep access for twelve months.

How you learn

About this course

Generative Fill demonstrates beautifully and then disappoints on the job. The first result looks convincing on screen, and at print size the texture is soft, the perspective is slightly wrong and the client asks why the shadow falls the other way. This course is about the difference between a generated result and a delivered one.

You start with what the tools actually do. Generative Fill, Generative Expand, Generate Background, Generate Similar and the Remove tool each behave differently, and each fails differently. You learn to recognise the failure modes early, so you stop regenerating and start correcting. From there the day is mostly prompting practice: describing subject, lighting, lens and material rather than typing a noun, using the shape of your selection as part of the instruction, working through variations, and conditioning a generation on a reference image so the result matches the rest of a campaign.

The afternoon is production work. You extend a photograph to a new aspect ratio so one shoot covers a print advert and a vertical social post, remove cables, reflections and bystanders from a location shot, build a background for a product photographed on a messy desk, and then repair what the model got wrong by hand: matching grain, rebuilding texture and cleaning up edges with the ordinary Photoshop skills that still decide whether the file passes.

The last module covers the part clients ask about. You inspect the Content Credentials Adobe attaches to generative output, see what provenance information travels with a file, and work through a simple internal position on disclosure, client sign-off and what to check in your own licence before generative work goes into paid advertising. Adobe states that its Firefly models are trained on Adobe Stock and openly licensed or public domain content, and offers indemnification on some business plans; the course shows you where to verify that for your own subscription rather than taking it on trust.

The class runs live online or in the classroom in Colombo, with pricing in LKR for Sri Lanka and USD internationally. You keep the exercise files, the prompt library built during the day and the class recording for twelve months.

What you'll be able to do

Choose correctly between Generative Fill, Generative Expand, Generate Background and the Remove tool for a given job.
Write prompts that specify subject, lighting, lens, material and framing instead of a single noun.
Shape a selection deliberately so the marquee itself guides what the model generates.
Generate, compare and manage variations, keeping the generative layer stack tidy and editable.
Condition a generation on a reference image so the result matches an existing campaign style.
Extend a photograph to a new aspect ratio for both a print advert and a vertical social post.
Remove cables, reflections and unwanted people from a location shot without leaving a smeared patch.
Build a clean product background and correct the generated result for grain, texture and edge quality.
Inspect the Content Credentials attached to a generated file and explain what provenance data it carries.
Draft a short internal position on disclosure, client sign-off and licence checks for AI-assisted images.

What we cover

The on-demand library is being recorded. Leave your email on any course and we will tell you when it lands.

Module 1 — What the generative tools actually do

The feature set in the current release, the model behind it, and how each tool fails.

Generative Fill, Generative Expand, Generate Background and the Remove tool 25 min
How Adobe describes Firefly training data, and what that means for commercial work 25 min
Recognising a bad generation early: perspective, texture, hands, type and repetition 25 min

Module 2 — Prompting inside Photoshop

Getting a usable result in two or three attempts rather than twenty.

Prompt structure: subject, lighting, lens, material and framing 35 min
Selection shape as instruction: how the marquee changes the result 25 min
Variations, ratings and keeping the generative layer stack editable 25 min
Reference images and matching an existing campaign style 20 min

Module 3 — Production work with generative tools

Four jobs that come up weekly, taken from prompt to a file that would pass review.

Extending a photograph to print and vertical social aspect ratios 30 min
Removing cables, reflections and bystanders from a location shot 25 min
Backgrounds and set extensions for product photography 30 min
Repairing artefacts by hand: grain, texture, edges and colour match 20 min

Module 4 — Rights, disclosure and handover

The questions a client or a legal team will ask, and how to answer them with the file in front of you.

Content Credentials: inspecting what provenance data travels with the file 25 min
Disclosure, client sign-off and checking your own licence terms 25 min
File naming, versioning and handing generative work to a print or web team 25 min

What's included

Class recording, available for 12 months
Exercise images, including the awkward files used for artefact repair
Prompt library built during the day, structured by job type
Checklist for reviewing generative work before it goes to a client
MyLearnPlus certificate of completion
Instructor email support after the class for questions on the exercises

Need this for a team?

We run this course privately, on your dates, at your office or online, tailored to the work your team actually does. Tell us what you need and we will send a quote.

Get a quote