🔢Stalecollected in 2h

Photoshop's Remove Tool Now Supports On-Device AI Models

PostLinkedIn
🔢Read original on 少数派

💡Photoshop moves to on-device AI, proving that high-quality generative editing can now run locally without cloud latency.

⚡ 30-Second TL;DR

What Changed

Photoshop Remove tool now utilizes on-device AI processing.

Why It Matters

This move signals a broader industry trend toward edge AI, reducing latency and infrastructure costs for image editing tasks. Practitioners should evaluate local model deployment for privacy-sensitive workflows.

What To Do Next

Check the latest Photoshop update and test the Remove tool on high-resolution assets to benchmark local inference speed against cloud-based alternatives.

Who should care:Creators & Designers

Key Points

  • Photoshop Remove tool now utilizes on-device AI processing.
  • Improved performance for object removal tasks.
  • Enhanced user privacy by reducing cloud data dependency.

🧠 Deep Insight

Web-grounded analysis with 35 cited sources.

🔑 Enhanced Key Takeaways

  • The on-device AI model for the Remove tool was officially released with Photoshop version 27.7 on May 21, 2026.
  • Utilizing the on-device model requires substantial hardware, including an Apple Silicon M1 Max or higher with 24 GB RAM and macOS Tahoe (macOS 26) or later on Mac, or an NVIDIA GeForce RTX 30-series or higher with at least 14 GB VRAM on Windows, and an approximate 5 GB model download.
  • Users have the flexibility to switch between on-device and cloud processing for the Remove tool, with the 'Device' option automatically becoming available only on compatible systems.
  • This update also features enhanced integration between Photoshop and Firefly Boards, enabling direct document sharing and surfacing Firefly Boards as a destination within the Variations panel for moodboarding and iteration.
  • The Remove tool now includes a 'Find Distractions' feature, offering automated detection and removal of general distractions, people, or wires and cables.
📊 Competitor Analysis▸ Show
Product/FeatureAI Object RemovalGenerative Fill/ExpandPlatformPricing Model
Adobe PhotoshopYes (on-device/cloud)YesDesktop, Web, MobileSubscription (Creative Cloud)
Adobe LightroomContent-Aware Removal, AI MaskingNo (less deep AI than Photoshop)Desktop, Web, MobileSubscription (Creative Cloud)
Luminar NeoYes (GenErase, Object Removal AI)Yes (Sky Replacement, Face AI)Desktop (macOS, Windows)One-time purchase or Subscription
CanvaYes (Magic Eraser)Yes (Magic Edit, Magic Media)Web, MobileFree tier, Paid subscription
PhotoroomYes (Magic Retouch, object removal)No (focused on product visuals)Web, MobileFree tier, Pro subscription
Cleanup.picturesYes (browser-based)NoWebFree tier
Samsung Object EraserYes (on-device)Yes (Generative Edit)Mobile (Samsung Galaxy S24+)Included with device
Fotor AI EraserYesYes (AI Image Generator)OnlineFree tier, Paid subscription
Pixelcut.aiYes (Magic Eraser)NoWebFree tier, Paid subscription for advanced
Claid.aiYes (Brush tool, AI Edit)Yes (AI Photoshoot, AI upscaler)WebPaid subscription (ecommerce-focused)

🛠️ Technical Deep Dive

  • The on-device model for the Remove tool is a generative AI model.
  • It is a GPU-intensive feature, requiring GPU acceleration and specific 'Remove Tool: Faster' processing flags to be enabled in Photoshop settings.
  • The model download size is approximately 5 GB.
  • Minimum hardware requirements are stringent:
    • Mac: Apple Silicon M1 Max or higher, 24 GB RAM minimum, and macOS Tahoe (macOS 26) or later.
    • Windows: NVIDIA GeForce RTX 30-series or higher with at least 14 GB VRAM, or a discrete GPU with 12 GB VRAM (Intel systems with integrated GPU only require 32 GB system RAM).
  • Adobe's broader AI capabilities, including those powering the Remove tool, are built upon Adobe Sensei and Firefly technologies, which leverage machine learning and deep neural networks.
  • On-device AI processing enhances user privacy by ensuring that sensitive image data remains on the local device and is not transmitted to cloud servers for processing.

🔮 Future ImplicationsAI analysis grounded in cited sources

Adobe will likely expand on-device AI capabilities to more Photoshop features and other Creative Cloud applications.
The successful implementation of on-device AI for the Remove tool, coupled with Adobe's existing Sensei and Firefly platforms, sets a precedent for local processing of other computationally intensive AI features, especially given the privacy and performance benefits.
The high hardware requirements for on-device AI will drive demand for more powerful consumer and professional workstations.
The need for M1 Max/RTX 30-series GPUs and significant RAM for the current on-device model indicates that users seeking the best performance and privacy will need to invest in high-end hardware, influencing future PC and Mac upgrade cycles.
On-device AI will become a key differentiator for professional creative software, emphasizing privacy and offline workflow.
As AI editing tools become more ubiquitous, the ability to process sensitive creative work locally, without constant internet dependency or cloud data transfer, will be a significant advantage for professionals and enterprises concerned with data security and workflow reliability.

Timeline

2010
Photoshop introduces Content-Aware Fill, an early AI-powered feature for object removal.
2016
Adobe Sensei, Adobe's AI and machine learning platform, is launched, integrating AI across its software suite.
2020
Photoshop launches its first AI-powered 'Neural Filters.'
2023-05
Adobe Photoshop introduces AI-based generative capabilities with Generative Fill, powered by Firefly AI, in beta.
2024-10
Photoshop's Remove Tool is updated with generative AI capabilities powered by Adobe Firefly, including 'Find Distractions' for people, wires, and cables.
2026-05
Adobe releases Photoshop version 27.7, introducing an on-device generative AI model option for the Remove tool.
📰

Weekly AI Recap

Read this week's curated digest of top AI events →

👉Related Updates

AI-curated news aggregator. All content rights belong to original publishers.
Original source: 少数派