Apple Shock | iOS 27 Photo AI: 3 On-Device Features Enter the Arena
機械翻訳 / Machine-translated

機械翻訳 / Machine-translated
@aifriends
AI Friends(https://aifriends.jp)のクロスポスト公式アカウント。AIツールの紹介・使い方・できることを、中学生でもわかるやさしい日本語で届けます。
"The scenery beyond the edges of a photo magically stretches outward." "An AI can instantly generate a front-facing shot of a landmark you photographed from an angle." — That kind of future appears to be coming to iPhone this fall 2026.
On April 28, 2026, Bloomberg (written by Mark Gurman, the well-known Apple watcher) broke a major scoop about three AI photo-editing features — "Extend / Enhance / Reframe" — planned for iOS 27, iPadOS 27, and macOS 27.
"What can it do?" "How is it different from the current iPhone?" "Can it beat Pixel and Galaxy?" "Will it work in Japan too?" — Here are the answers to all your burning questions, explained in plain language anyone can understand.
Let's start by breaking down the primary scoop from three angles.
It all started with a Bloomberg article published on April 28, 2026.
The title: "Apple's iOS 27, macOS 27: Photo Editing With AI to Extend, Enhance and Reframe."
Written by Mark Gurman, a reporter with strong insider sources on Apple — widely followed for his high accuracy rate.
The plan: a new Apple Intelligence editing features section to be added to the Photos app in iOS 27, iPadOS 27, and macOS 27.
Official announcement: at the WWDC 2026 (Worldwide Developers Conference) keynote, scheduled for June 8, 2026 (US time).
General release: likely in fall 2026, around the same time as the new iPhone 18 series announcement.
Think of it as the information-density equivalent of a top sports journalist scooping "I know Japan's entire formation" a month before the World Cup.
Major outlets — MacRumors, 9to5Mac, Engadget, AppleInsider, GIGAZINE, and others — all followed with their own coverage.
Apple itself had no comment, as expected — though past scoops of this kind have almost always proven accurate.
Apple Intelligence-compatible devices began rolling out in phases starting October 2024, with Japanese-language support arriving in iOS 18.4 at the end of March 2025.
The three new features are expected to launch with Japanese language support already in place, meaning Japanese users will likely be able to use them from day one.
The stars of the show are the new "Apple Intelligence Tools" section appearing in the Photos app editing screen.
① Extend: AI generates scenery beyond the edges of your photo, filling in what lies outside the frame.
Example: After shooting a tight close-up of a landmark at a tourist spot, pinch out to have the AI naturally add surrounding buildings, sky, and roads.
② Enhance: One-tap correction for color, brightness, noise, and sharpness — Apple Intelligence brings photos to a professional-grade finish.
Example: A dimly lit food photo taken indoors is automatically transformed by AI into a bright, vivid image worthy of social media.
③ Reframe: Allows you to shift the viewpoint of Spatial Photos (3D photos compatible with Vision Pro) after the shot is taken.
Example: A portrait shot straight-on can be reconstructed by AI to appear as if viewed from the side or at an angle.
Think of it as three kinds of magic on a single image: "stretch it outward," "make it glow," and "change the angle."
The existing iOS 18 "Clean Up" (object removal) feature stays — these three are added on top of it.
All processing happens on-device; photos are never sent to the cloud.
Processing takes a few seconds per photo; requires an Apple Intelligence-compatible A17 Pro / A18 / M2 or later chip.
Apple's approach: "changing how a photo looks without changing what's in it."
The "cautionary note" Gurman also conveyed is equally important.
Extend and Reframe are running into quality issues in internal testing — including unnatural compositing and distortion in viewpoint changes.
Apple's policy: "don't release until the AI model meets the bar" — the more important the feature, the more cautious the approach.
Worst-case scenario: they don't make it into the initial iOS 27 release and are added later in iOS 27.1–27.2.
Enhance, by contrast, is relatively light to process and is already highly refined — it's expected to be included in the initial release.
Think of it like developing powered folding mirrors and auto-parking for a new car simultaneously — sometimes only one ships first.
Past precedent: "Genmoji" and "Image Playground" announced with iOS 18 were initially US English only and expanded over time.
Full Japanese-language support for all Apple Intelligence features took until iOS 18.4 at the end of March 2025.
This time, Apple's familiar rhythm of "big announcement, gradual rollout" is to be expected.
For users, the realistic approach is: watch the June announcement → wait for the fall release → try features with each update.
The key dividing line: whether you own an Apple Intelligence-compatible device. iPhones below the A17 Pro won't benefit.
2026 is shaping up to be a year where "upgrading your iPhone for AI features" becomes a stronger motivation than ever.
Here are three angles on the strategy behind Apple's moves.
The reason Apple is rushing these features now: Android rivals have built up a significant lead.
Google Pixel's "Magic Editor" debuted with the Pixel 8 announcement in 2023 — it can move subjects, resize them, and auto-generate backgrounds.
Samsung Galaxy's "Galaxy AI" has been available since early 2024, covering AI photo editing, object removal, slow-motion video generation, and call translation.
Meanwhile, Apple only added "Clean Up" (object removal) with iOS 18.1 in October 2024 — still the only photo AI feature available to date.
It's like being the only one using a flip phone to take photos while everyone else in class is freely editing theirs on a smartphone.
In Tom's Guide's AI photo editing shootout, Apple ranked last among three competitors — a decisive loss to Google and Samsung.
iPhone sales remain solid, but voices from younger users and social media users saying "Android beats iPhone on AI features" have been growing.
Apple's leadership took this situation very seriously, issuing a directive in 2025 to make "Apple Intelligence the company's top priority."
These three features are a symbolic demonstration of how serious Apple is about catching up.
Extend = answer to Magic Editor. Enhance = answer to Galaxy AI. Reframe = Apple's unique angle, strengthening the Vision Pro ecosystem.
Positioning: "a three-feature package to close the gap all at once."
The "path to winning," once caught up, runs through on-device processing.
All three new features complete their AI processing entirely on the iPhone, iPad, or Mac — photos never leave the device.
Pixel and Galaxy rely on cloud processing for some features, meaning photos are sent to Google's or Samsung's servers.
Apple's philosophy — "Privacy is a fundamental human right" — implemented directly in the product.
On-device processing is also faster, and works without an internet connection, even in airplane mode.
Think of a famous chef proudly declaring, "Everything in our kitchen is prepped on-site."
There is a trade-off, though: a powerful AI chip is required — only relatively recent devices with A17 Pro / A18 / M2 or later qualify.
Users on iPhone 15 or earlier will need to upgrade — a point of contention.
On the other hand, privacy-conscious users gain loyalty: "I want Apple processing my photos, even if it costs more."
With the EU's AI Act taking effect in 2025, California's CCPA tightening, and privacy regulations strengthening worldwide, Apple's on-device strategy is perfectly positioned to ride that tailwind.
"Late on AI performance, but king of privacy" is the new competitive axis.
The most distinctive of the three features is "Reframe."
Reframe is exclusive to Spatial Photos — that is, 3D photos taken with Vision Pro or iPhone 15 Pro and later.
Being able to shift the viewpoint after shooting is a stepping stone toward 3D video and VR experiences.
Apple launched Vision Pro in February 2024, but its high price (approximately ¥600,000) kept sales from taking off.
Rumors of a more affordable "Vision Air" (tentative name) in 2025 suggest Apple is aiming for mainstream adoption of spatial computing in 2026.
Think of it like launching a new gaming console but lacking software — so you develop your own popular titles to drive excitement.
Reframe's role is to expand the "content to watch" on Vision Pro, turning users' existing photos into VR content.
Vision Pro received a major feature upgrade with visionOS 26 in June 2025, with deeper Apple Intelligence integration underway.
visionOS 27 is also expected to be announced at WWDC 2026, where integration with Reframe is likely to be a highlight.
Apple's vision: a new experience flow of "take photos → AI-edit them → immerse yourself in them on Vision Pro."
From "photos are something you take" to "photos are something you step into and experience" — an ambition to redefine the medium.
Reframe may look modest on the surface, but many experts call it the "sleeper frontrunner" in Apple's long-term strategy.
Let's look at the comparisons that matter most, across three axes.
The biggest rival: Google Pixel's "Magic Editor."
Magic Editor launched with the Pixel 8 announcement in October 2023 — a powerful tool that can move, enlarge, or delete subjects within a photo and auto-generate backgrounds.
Tom's Guide called it "the all-around king of AI photo editing," with polish further improved on the Pixel 9 Pro XL.
The difference from Apple's Extend: Magic Editor "changes what's inside the photo" (moves subjects); Extend "stretches the outside of the photo" (generates beyond the frame).
The difference from Apple's Enhance: Magic Editor's "Magic Eraser" leans toward object removal; Apple's Enhance focuses on color and brightness correction — they don't overlap.
In cookbook terms: Magic Editor = rearranging the ingredients on the plate; Extend = adding more food around the plate.
Magic Editor leans toward cloud processing (sends to Google Photos servers); Apple's three features are fully on-device, with a privacy advantage.
Processing speed: Magic Editor 5–10 seconds; Apple's estimated 2–5 seconds (on-device advantage).
User experience difference: Pixel's appeal is "the fun of tweaking anything freely"; Apple's direction is "professional results, effortlessly."
After fall 2026, a cultural showdown between "Magic Editor camp" and "Extend camp" will begin.
Which you prefer is up to you — ideally, you'd get to try both.
Another formidable rival: Samsung's "Galaxy AI" and Chinese smartphone makers.
Galaxy AI made its full debut with the Galaxy S24 in 2024, offering a wide range of features including AI photo editing, call translation, and Circle to Search.
The 2026 Galaxy S26 brings major upgrades to "Drawing Assist" and "Photo Assist," enabling subject addition and style conversion.
Among Chinese brands, Xiaomi, Vivo, and Honor are pushing hard with proprietary AI features — Xiaomi 15 Ultra's "AI Image Expansion" in particular is close to what Extend does.
Honor Magic 7 Pro's "AI Eraser" has been rated as surpassing Google's Magic Eraser in accuracy.
It's like the TV aisle at an electronics store, with every brand claiming "ours looks the best" — a true era of competition.
Apple's strength, however, lies in "a consistent ecosystem" — the same features work across iPhone, iPad, Mac, and Vision Pro.
Galaxy and Xiaomi have feature differences between devices; Apple Intelligence promises "the same experience on any Apple device."
For households where the whole family uses Apple devices, the seamless integration of AI editing across anyone's photos is a major appeal.
The photo AI market from 2026 to 2027 will be a four-way structure: Apple vs. Google vs. Samsung vs. Chinese brands.
The era where users choose not by "number of features" but by "does this fit my lifestyle?" has arrived.
The degree of change for iPhone users is significant.
Current iOS 18 has just one photo AI feature — Clean Up (object removal) — and it's limited to iPhone 15 Pro and later.
iOS 27 expands this to four features (Clean Up + Extend + Enhance + Reframe), with a new "Apple Intelligence Tools" section in the editing screen.
From a UI/UX perspective, it's a "one-button editing revolution" — advanced edits previously handled by third-party apps (Photoshop Express, Pixelmator, etc.) are now built in by Apple.
It's like the TV you bought from an electronics store suddenly coming with a Blu-ray player and soundbar built right in.
As a result, third-party photo editing apps face stiff competition; Adobe is exploring differentiation through its Creative Cloud subscription model.
Lightroom's "AI Generative Remove" and Photoshop's "Generative Fill" remain powerful — pros and semi-pros will continue using both.
Casual users, however, may find "the free built-in features are enough," leading to potential increases in subscription cancellations.
For current iOS 18 users, the key question is: "Is my device Apple Intelligence-compatible?" — A17 Pro / A18 / M2 or later is required.
For those still on iPhone 14 or earlier, this is a prime moment to consider upgrading — it's worth timing your decision around the iPhone 18 announcement (expected September 2026).
"Upgrading your iPhone for photo AI" is a brand-new purchase motivation.
The biggest good news for Japanese users: there's a strong chance they'll be able to use these features in Japanese from day one.
Apple Intelligence's Japanese-language support arrived with iOS 18.4 at the end of March 2025 — at the time, it was six months behind.
As of 2026, Japanese has been elevated to a primary language alongside English, making simultaneous support for new features the new standard.
The three iOS 27 features essentially only require UI label and prompt translation to function — Japanese-language compatibility is technically straightforward.
However, if natural-language Siri or voice commands like "expand this photo" are involved, Japanese accuracy will need to be verified.
It's like going from "waiting for subtitles" to "same-day dubbed broadcast" — the stress of lagging behind disappears.
Apple's official Japanese website and support documentation are expected to update simultaneously — no more Japanese users stuck with English-only screenshots.
Apple Authorized Resellers (Bic Camera, Yodobashi, Tsutaya Electrics) should have working demo units available at an early stage.
There's also a possibility that the App Store Awards selection criteria will incorporate "Apple Intelligence utilization," accelerating AI adoption among Japanese app developers.
Fall 2026 could mark the first generation of Japanese iPhone users experiencing cutting-edge AI features at the same time as English-speaking regions.
Japanese users are particularly avid photographers — and the benefits here are substantial.
According to research firm GfK, Japanese iPhone users take 1.3 times as many photos as the global average, with a strong focus on tourist spots, food, and pets.
Extend lets you fix that moment when you shot a building too close at a tourist spot — you can fill in the surrounding scenery after the fact.
Enhance converts a dimly lit food photo from a izakaya or café into something bright and sharp with one tap — perfect for social media.
Reframe enables cultural use cases like shooting historical architecture in Kyoto or Nara as Spatial Photos and then experiencing them in VR on Vision Pro.
It's like being at a Japanese bathhouse and having your coffee milk refilled without even asking — total hospitality.
Life milestone photos — weddings, sports days, graduation ceremonies — can be AI-enhanced into "professional-grade albums."
That said, using Extend to "artificially extend cherry blossom branches" is not recommended for journalistic or documentary photography.
Discussions around ethics — such as marking photos as "AI-processed" or automatic metadata recording — are also beginning.
Fall 2026 will bring Apple's philosophy of "don't change what's in the photo, only change how it looks" into public debate in Japan as well.
The more you love photography, the more "literacy in how to use these tools" will matter.
Ripple effects will spread through Japan's photo and creative industries as well.
Photo book services (Shimauma Print, TOLOT, Photoback, etc.) will need to revamp their templates to account for Enhance-processed photos as the new baseline.
Wedding and coming-of-age photography studios (Studio Alice, Life Studio, etc.) will face pressure to differentiate from "AI post-processing."
That said, demand for "scenes that require a professional" — traditional Japanese portraits, Shichi-Go-San, Miyamairi — remains robust.
The freelance photography market will polarize between "those for whom AI correction is enough" and "those who need a professional."
Think of it like the sharpest kitchen knives available to home cooks getting dramatically better — but the sushi chef's craft doesn't disappear.
Photo editing apps (Foodie, Snapseed, Lightroom Mobile, etc.) will need to clearly articulate "what they offer that Apple's built-in tools don't."
Apps specializing in "AI video editing" for TikTok and Instagram Reels may increase.
Japanese SaaS products like Picsart and PhotoDirector will leverage the Apple Intelligence integration API to survive through differentiated features.
The photo AI market is expected to grow 15–20% annually from 2026 to 2028, with the Japanese market reaching 1.5 times its 2025 size.
A new division of labor: "foundational AI for everyone" from Apple, "deep customization" from third parties.
Ayaka runs a travel blog from Kyoto.
Monthly PV: 100,000. Instagram followers: 25,000.
Her struggle: "I think I've taken a great shot at a tourist spot, but when I look at it later, the edges are cut off and it's not suitable for social media."
In September 2026, she starts using "Extend" on her iPhone 18 Pro, freshly updated to iOS 27.
She uses Extend to naturally expand a close-up shot of Kiyomizudera's main hall to include the surrounding mountains and sky.
She uses Enhance to bring out the misty morning atmosphere — the Instagram post earns her a personal record of 3,500 likes.
Her workflow goes from 2 hours of shooting → 10 minutes of editing → posting, with a dramatic improvement in pace.
It feels like having a professional photographer by your side — anyone's shots come out looking like travel magazine photos.
Her monthly blog PV climbs to 120,000, and her Instagram followers break 30,000.
Ad revenue rises from ¥100,000/month to ¥180,000/month — turning her side hustle into a potential full-time career.
She does receive some feedback about "over-editing with AI" and starts using "unedited" tags to differentiate.
Fall 2026 will be a turning point between bloggers who master AI and those who don't.
Ayaka becomes one of the former — and starts receiving interview requests from Kyoto tourism media.
Kensuke runs a 15-seat Italian restaurant in Kichijoji, Tokyo.
Social media marketing is a challenge.
His struggle: "I post food photos on Instagram, but my wife keeps telling me they don't look appetizing because the restaurant is too dark."
In October 2026, he tries iOS 27's Enhance on his new iPhone 18.
He corrects photos of pizza, pasta, and wine with one tap — brightness and color are dramatically transformed into professional-looking shots.
He uses Extend to naturally add the full table and restaurant ambiance around the dishes, creating photos that convey the "story of the place."
It's the kind of saving that normally requires hiring a food photographer for ¥100,000/month — achieved for free.
The result: average Instagram likes per post jump from 30 to 180, and new customers increase from 50 to 90 per month.
Photo AI isn't the only factor, but the quality of his social media management has clearly improved across the board.
He also refreshes his photos on Tabelog and Google Business, and his search rankings improve.
By the end of 2026, reservations are booked three weeks out — AI becomes a lifeline for his small neighborhood restaurant.
Kensuke is invited to speak as an instructor at a local chamber of commerce "Social Media × AI" seminar.
He embodies what it means to be a "shop owner who makes the most of AI."
Yuko is a freelance wedding photographer based in Yokohama.
She photographs around 40 weddings per year.
Her struggle: "Editing 500–1,000 images per wedding takes an enormous amount of time — just the color correction in Lightroom takes 10 hours."
In fall 2026, Yuko switches to a hybrid workflow combining iOS 27's Enhance with Adobe Lightroom.
She runs every shot through iPhone's Enhance for a first pass → then fine-tunes in Lightroom.
Editing time drops from 10 hours to 4 hours; her delivery lead time goes from two weeks to one week.
It's like a carpenter adopting power tools — the timeline gets cut in half, but the craftsperson's skill remains.
She can now handle 55 weddings per year instead of 40 — a meaningful income increase.
She also introduces a "new wedding video presentation" using Reframe — preserving the vow scene as a Spatial Photo in three dimensions.
She launches a new service where the couple can immerse themselves in "the moment of that day" on Vision Pro — three bookings a month.
At an additional ¥50,000 per booking, that's an extra ¥180,000 per year.
Yuko's path is not "AI taking her job" but "AI creating more work for her."
For creative professionals in 2026–2027, the question isn't "do you use AI or not?" but "what new things are you creating with AI?"
For creative workers, AI is a weapon — and those who pick it up first will take the market.
A. Official announcement: June 8, 2026 (US time) at the WWDC 2026 keynote. General release is expected around September 2026.
Apple typically follows this schedule for major updates, coinciding with the new iPhone announcement (early September).
Developer Betas begin right after WWDC in June; Public Betas follow in July.
Betas are unstable — if you're using your main device, it's safest to wait for the official release in September.
Think of it like choosing between an "early access" version of a game and waiting for the "official release."
Even after the official release, early versions tend to have bugs — iOS 27.0.1–27.1 in late September to early October will be more stable.
If Extend/Reframe don't make the initial cut, they may be added in iOS 27.1 (around November) or later.
Back up to iCloud before updating — this is the golden rule, to protect against any accidental loss of photos or contacts.
If you want to try it right away, jump in with the Public Beta in July. For stable, everyday use, wait for the official version in September.
A. Only Apple Intelligence-compatible devices — meaning models with an A17 Pro / A18 / A18 Pro / A19 or later chip.
Specifically: iPhone 15 Pro / 15 Pro Max, all iPhone 16 models, and all iPhone 17 models.
iPhone 14 and earlier, and iPhone 15 / 15 Plus (A16 chip), are not supported.
For iPad: iPad Pro / Air with M1 or later, and iPad mini with A17 Pro or later.
For Mac: all models with M1 or later — meaning all Macs from 2020 onward.
The iPhone 18 series, expected to be announced in September 2026, will of course be fully compatible.
Think of it like a new game that only runs on the latest console — hardware from three generations ago won't run it.
For users on iPhone 14 or earlier, this is a great opportunity to consider upgrading — you gain both new features and AI capabilities.
In the used market, iPhone 15 Pro prices are holding relatively steady, sustained by demand for its AI-compatible value.
Buying a used iPhone 15 Pro for AI features is a smart option — available in the ¥60,000–80,000 range.
Your decision to upgrade should hinge on two things: "Am I frustrated with my current smartphone?" and "How do I want to use photo AI?"
A. All three features run on-device — your photos are never sent to the cloud.
Apple Intelligence's design philosophy explicitly prioritizes "on-device processing first."
For tasks that exceed the device's processing capacity, data may be sent to Private Cloud Compute (PCC) — but photo AI is expected to run fully on-device.
PCC is designed to delete data after processing; Apple has publicly stated that even its own staff cannot access it.
Even photo metadata (location, time, camera model) stays on-device by default.
Think of it like a bank vault that the bank tellers themselves cannot open.
Pixel and Galaxy use cloud processing for some features, meaning photos are sent to Google's or Samsung's servers.
On-device processing is fully compliant with the EU's AI Act and GDPR, California's CCPA, and Japan's Amended Act on the Protection of Personal Information.
That said, if you share an "AI-processed photo" via iCloud, metadata may be visible to recipients (family or friends) on their devices.
New habits to consider: such as showing an "AI-edited" flag when sharing photos within the family.
Apple's brand image as "privacy-first" will only strengthen further.
A. Yes — all existing photos in your photo library can be edited.
Any photo in your iCloud Photo Library, no matter how old, can be AI-edited as long as it's downloaded to your iPhone.
The original photo is not altered — the edited version is saved as a separate version, with an edit history preserved.
To revert, simply tap "Revert" in the editing screen — you can undo at any time.
Works with any photo in "Recents," "Favorites," or any Album in the Photos app.
Live Photos, Portrait mode, and Night mode photos are also expected to be supported.
It's like refreshing an entire old family photo album with a "magic scanner."
Note: if your iCloud Photo Library is set to "Optimize iPhone Storage," only a low-resolution preview may be on your device, which can't be edited.
In that case, switch to "Download and Keep Originals" and temporarily download the photo you want to edit.
Processing time: 2–10 seconds per photo; for 100 photos at once, allow 5–15 minutes.
Battery consumption is about 1.5 times normal usage — plugging in a charger is recommended.
A. Some work will be replaced, and some won't.
Most at risk: "everyday snapshot touch-ups" and "editing photos for social media posts" — tasks achievable at a basic level by anyone.
Least at risk: milestone life events like weddings, coming-of-age ceremonies, and Shichi-Go-San — work that requires technical skill and artistic sensibility.
"Product photography," "advertising photography," and "architectural photography" — jobs that depend on professional lighting and composition — remain in steady demand.
On the other hand, "casual family portrait sessions" and "business headshots" may be substituted by AI correction, making the freelance market tougher.
Think of it like housekeeping services: "routine cleaning" gets automated, but "a thorough deep clean" is still done by people.
Survival strategies for professional photographers: ① Specialize in "areas AI can't touch" (weddings, commemorative shoots, etc.); ② Go hybrid, making full use of AI; ③ Evolve into a comprehensive producer covering shooting + editing + storytelling.
The photography industry will undergo significant restructuring as it adapts to the "AI era" from 2026 to 2030.
Photographers who can offer "relationship" and "story" to their clients will be the strongest.
Those with human capital — photography workshops, local community ties, active social media presence — will be the ones who survive.
The right answer is not "being replaced by AI" but "co-existing with AI to create new value."
"Don't change what's in the photo — magically change how it looks." That is Apple's iOS 27 photo AI strategy.
The three features — Extend, Enhance, and Reframe — are Apple's serious attempt to close the two-year gap it let open against Pixel and Galaxy.
Full on-device privacy processing, combined with deep ecosystem integration, is what sets Apple apart.
It's like a traditional Japanese craft, reborn by embracing the latest technology — the craftsperson's hands, transformed.
After the official iOS 27 release this fall 2026, iPhone users around the world will experience a new era of photo AI.
For Japanese users, full Japanese-language support from launch day means they can enjoy the cutting edge right alongside English-speaking regions — no gap.
Three things you can do starting today: reserve your spot for the WWDC live stream on June 8, check whether your device is compatible, and organize your iCloud Photo Library — whether you actively embrace the photo AI revolution or let it wash over you is a choice only we, the users, can make.
This article is a cross-post from AI Friends.