13:15
Photoshop Generative Fill
The video, day 1 of a free PHLEARN AI series, demonstrates that Photoshop's Generative Fill — powered by Adobe's Firefly Image 3 and partner models like Google's Gemini 2.5 Nano Banana — can remove, add, and replace objects in a photo through a rough selection plus a one-line text prompt, replacing manual retouching with single-click AI generation.
Object removal: paint a selection with the selection brush tool, open Generative Fill, leave the prompt blank, choose Firefly Image 3, and click Generate to erase the selected object.
Object addition: paint a selection over empty space and type a short prompt (e.g. 'garden gnomes') into Generative Fill to insert new content with Firefly Image 3.
Object replacement: select the entire image (Select > All) and use Gemini 2.5 Nano Banana with a natural-language instruction (e.g. 'change her existing boots to sunflower boots') rather than a tight local selection.
Firefly Image 3 always returns three variations per generation; partner models like Gemini return only one variation at a time, though you can click Generate repeatedly.
Generative Fill creates separate, independently toggleable layers, so before/after can be compared by clicking the layer eye icons.
Unwanted side effects of a fill (e.g. it altering the subject's sunglasses) can be fixed by clicking the fill layer's mask, pressing B, and painting black to restore the original pixels there.
The AI reads context from the surrounding image (recognizing ocean and sand) to stylistically blend new content, as shown when the prompt 'tropical' produces beach-appropriate trees.
The separate Remove tool is offered as a free-standing alternative to Generative Fill for pure object removal, with a dedicated video promised later in the series.
All sample images, PSDs, and a follow-along calendar are distributed free via a link in the description.
The Generative Fill entry point lives in the contextual taskbar, which must be enabled via Window > Contextual Taskbar if not already visible.
Selection Brush Tool — A Photoshop tool that lets you paint a rough circle or freeform area over a region to create a selection, automatically filling in the center. Apply: Grab the selection brush tool from the toolbar and paint imprecisely over anything you want to remove, add to, or change before invoking Generative Fill.
Generative Fill — Object Removal (blank prompt) — Using Generative Fill with no text typed into the prompt field, which signals the AI to remove the selected object rather than add anything. Apply: Select the unwanted object with the selection brush, open Generative Fill, leave the prompt blank, choose Firefly Image 3, and click Generate.
Generative Fill — Object Addition (prompted) — Using Generative Fill with a short text prompt describing a new object to generate inside the selected area. Apply: Paint a selection over the empty area, type a prompt like 'garden gnomes' into Generative Fill, select Firefly Image 3, and click Generate to get three variations.
Generative Fill — Object Replacement (whole-image Nano Banana workflow) — Using Google's Gemini 2.5 Nano Banana partner model with a full-image selection and a natural-language instruction to change one specific existing element. Apply: Select the entire image via Select > All, open Generative Fill, pick Gemini 2.5 Nano Banana, and describe the desired change in words (e.g. 'change her existing boots to sunflower boots') instead of painting a tight local selection.
Firefly Image 3 — Adobe's own AI image-generation model built into Photoshop's Generative Fill, which always returns three variations per generation. Apply: Select it from the partner-model dropdown in Generative Fill whenever you want multiple simultaneous variations to browse, as used in the removal, addition, and beach-cleanup examples.
Gemini 2.5 Nano Banana — A Google partner model available inside Photoshop's Generative Fill that works from a whole-image selection plus a descriptive text instruction, returning one variation per generation. Apply: Use it for targeted, describable edits (like swapping boots) rather than precise local painting, and click Generate repeatedly to get additional single variations.
Remove Tool — An alternative Photoshop tool to Generative Fill for deleting selected objects, mentioned as the subject of a separate upcoming video in the series. Apply: Click Remove instead of Generative Fill on an active selection when you just want the object gone and don't need multiple generated variations.
Layer Mask Correction — A technique for fixing unwanted side effects of a generative fill (such as it altering part of the subject, like sunglasses) by editing the fill layer's mask. Apply: Click the layer mask on the generative fill layer, press B for the brush tool, and paint black over the affected area to restore the original image there.
Contextual Taskbar Setup — The floating Photoshop toolbar that surfaces Generative Fill and other AI actions once a selection is active. Apply: If it isn't visible, go to Window and enable 'Contextual Taskbar,' then optionally drag it to a preferred screen position and pin that position via the menu item.
Model choice dictates selection strategy, not just output quality: Firefly Image 3 needs a tight local selection around the object, while Gemini 2.5 Nano Banana is described as working better from a whole-image selection paired with a verbal description of the change — the video treats this as a simple rule of thumb rather than flagging it as a limitation of either model.
An empty prompt field functions as an implicit 'remove' instruction in Firefly-based Generative Fill — the tool infers removal intent purely from the absence of text, not from an explicit command.
The one-variation-vs-three-variation split between partner models and Adobe's own Firefly is presented as a minor heads-up, but it structurally changes how much post-generation curation a user gets per click.
The replacement example shows the AI handling partial occlusion correctly: Nace points out that even the boots visible under the subject's lace dress hem were updated to match, evidencing more than a flat object swap.
«Hello everyone, my name is Aaron Nace of Flearn and in this series we're diving into the deep world of artificial intelligence for photo editing in both Photoshop and Lightroom.»
— 00:00
«Now, the selection brush tool, really, all you have to do is simply create a circle.»
— 01:29
«If you want to just simply remove the object from your image, you don't type anything in at all. Just leave this completely blank.»
— 02:31
«So, if you're using Adobe's AI, it's always going to give you three variations. The partner models are only going to do one at a time.»
— 06:29
«So, we've seen that generative fill is really good at removing objects from your photos. It's great at adding objects. and it's also good at replacing them as well.»
— 09:44
«Rather than just saying sunflower boots, I'm going to say change her existing boots to sunflower boots.»
— 10:20
«The subject looks the exact same. The rest of the image is maintained.»
— 12:29
«We'll see you tomorrow where we show you how to use generative tip expand to literally change the boundaries of your photos.»
— 13:06
Reception
Viewers appreciate the tutorial quality but express significant frustration with Adobe's unclear credit pricing and broader concerns about AI's role in photography.
As a day-1 tutorial the video is clear and concretely demonstrated across three worked examples with visible before/afters, but it functions as promotional walkthrough content for Adobe's AI stack rather than an evaluation of it.
13:15