Shokunin Studio: AI Cloned Photoshop Brushes Perform Far Worse Than the Original
After months of tracking, Shokunin Studio points out that there are currently about a dozen AI-assisted attempts to clone Photoshop and other Adobe applications, along with numerous forked versions, with goals extending to Microsoft, Avid, Autodesk, and game development tools.
These projects have invested thousands of dollars in AI tokens (API fees); the largest attempt has over 70 developers (increased to 165 and still rising) using Claude to submit hundreds of pull requests daily, with token expenditures estimated to exceed $50,000 in the past two weeks.
Despite the investment, no project has achieved a brush engine that rivals Photoshop, with performance collapsing when the canvas exceeds 8K×8K pixels; Photoshop can maintain smooth brushes on 100-megapixel images from Hasselblad or Phase One, while copy-pasting and undo stacks on large canvases in cloned versions are extremely slow and can even freeze the application temporarily.
These projects also struggle to achieve 100% file format compatibility and lossless editing, with filter previews often downsampled to low resolutions to maintain speed, unstable color profile rendering and soft proofing, limited support for a wide range of colors and print profiles, and some platforms unable to use common fonts due to licensing issues, reverting overall to a pre-1996 lack of standardization.
It takes about $2,000 to generate a Photoshop competitor, while Affinity Photo is free and Photoshop's annual fee is about $200; participants criticize the subscription model while subscribing to more expensive AI services, resulting in low application usability and the need for ongoing high-cost maintenance.
Historically, open-source software has rarely posed a real challenge to commercial software; Krita has existed for years, but Clip Studio Paint subscriptions have brought millions in annual revenue to Celsys; a 2026 survey of 150 CG production companies in Japan shows that among the main 3D animation suites, Maya holds 58%, 3ds Max 17%, Blender 12%, and Cinema 4D 7%, with commercial applications leading innovation while open-source often follows.
Market Mechanism: AI companies earn continuous revenue from cloning projects through token fees, while cloners attempt to divert Adobe and other companies' users and revenue by replicating interfaces and functions at low costs; professional users still prefer commercial software due to performance, compatibility, and support issues, while cloning projects struggle to form effective alternatives due to quality gaps, with Adobe facing public pressure but still maintaining core engine barriers.
Source: Public Information
ABAB AI Insight
Shokunin Studio has been tracking the professional creative software ecosystem for a long time, focusing this time on the phenomenon of AI-assisted cloning; historically, similar behaviors include the early 2000s Adsense spam sites that copied pages in bulk to steal traffic, as well as long-standing software piracy and interface imitation, with Adobe applications being a major target for piracy since the 1990s.
In terms of capital pathways, participants have invested thousands to tens of thousands of dollars into AI services like Claude to generate code and pull requests, with AI companies receiving direct token revenue; cloning projects attempt to replicate mature products at extremely low marginal costs, diverting paying users, while original manufacturers maintain continuous revenue through cloud services and subscriptions to cover high complexity engine and support costs.
Similar cases include open-source image tools like GIMP and Krita that have existed for years but have not replaced Photoshop's commercial status, as well as Blender following rather than leading commercial suites like Maya; current AI cloning is in a stage of rapid generation but lacks core performance and compatibility, with interfaces easily replicated but engine-level optimization still relying on long-term engineering accumulation.
This essentially tests the boundaries of technological substitution. AI significantly compresses the time for interface and function realization, reducing cloning costs from team annual investments to individual thousands of dollars, but brush engines, large canvas performance, complete file format compatibility, and color management still require deep systematic optimization, making it difficult for generative methods to fully replace them in the short term; the result is accelerated superficial copying, while core barriers and commercial support models continue to persist.
ABAB News · Cognitive Laws
Interfaces can be quickly copied, but engine performance determines long-term barriers.
Generation speed does not equal engineering depth; equivalent human time may still fail.
Low-cost imitation diverts traffic, while high complexity support maintains paid users.