Content-Aware Fill predates the 2012 CS6 spotlight it got

The shot

Adobe's own archived product pages date the headline content-aware feature of Photoshop CS6 to 2012, not the introduction of Content-Aware Fill itself. A snapshot of Adobe's Photoshop page from late April 2012, as CS6 messaging replaced the prior version's, leads with 'Content-Aware Patch' and states 'New Content-Aware features take image retouching to a new level of ease and control' -- Patch, not Fill, is the tool Adobe introduces that year. An earlier snapshot from April 2011, a year before CS6, already describes Content-Aware Fill as established: 'Remove any image detail or object and watch as Content-Aware Fill magically fills in the space left behind. This breakthrough technology matches lighting, tone, and noise.'

What the documents show

The two snapshots, both Adobe's own marketing copy, establish that Content-Aware Fill predates the 2012 CS6 cycle often used as shorthand for Adobe's content-aware era; what CS6 added, on Adobe's own account, was Content-Aware Patch, plus retouching and crop improvements described as 'groundbreaking new Content-Aware tools.' A pair of Adobe blog posts documents the tool's next leap: a 2018 post describes a 'sneak peek of Content Aware Fill in After Effects at Adobe MAX 2018,' built under the name 'Project Cloak,' and a 2019 post confirms the shipped result: 'Use Content-Aware Fill to remove unwanted objects from your footage, such as boom mics, logos, and even people,' in After Effects 16.1.

The workflow

The image-based tool samples other regions of the same still frame to fill a selection. Extending that to video required solving a problem a still image lacks: keeping the fill consistent as camera and background move, which is why Adobe ran it as a multi-year research project before shipping a video version years after the image tool was already standard. A production still chooses which tool fits the shot; a locked-off frame differs from a moving handheld one, and that remains an editor's or compositor's call.

What the tool does not change

Content-Aware Fill and its video descendant rearrange pixels already present in the frame; they do not generate new imagery the way a diffusion or transformer-based model does, and Adobe's materials for both versions do not claim otherwise. That is a real difference from the masking and generation tools covered elsewhere on this site, including Rotobrush 3, and treating them as one lineage would overstate what Adobe's documentation describes. This is editorial: a sampling-based fill and a learned generation model solve different problems.

• Does a Content-Aware feature sample existing pixels in the frame, or generate new ones outside the frame's own content?
• Is the shot a locked-off frame or a moving one, and does that change which fill or masking tool is right?
• When a vendor calls a tool a breakthrough, does its documentation name the method, or only the result?

The 2012 CS6 cycle is a real milestone in Adobe's content-aware line, but Adobe's own earlier materials show the fill tool was already in artists' hands before that cycle began, with the video extension arriving as a separately dated research effort years later.