Meta Spark AR Development: Platform Discontinued, and What Replaced It
Meta Spark AR is discontinued. Meta shut the platform down on 14 January 2025, and named no replacement. Meta Spark Studio, Meta Spark Hub and Meta Spark Player stopped working on that date. Every AR effect built by a brand or a community creator was removed from Facebook, Instagram and Messenger, and only Meta's own first-party effects remain. As of August 2026, Instagram and Facebook still accept no third-party AR effects at all.
This page used to sell Spark AR development. It no longer can, and we would rather say so than quote for a platform that has been closed for nineteen months. What follows is where this work actually goes now, what to do if you still hold Spark assets or a live 8th Wall campaign, and how to buy AR in a category where two of the largest delivery platforms died inside fourteen months of each other.
What happened to Meta Spark AR
- Closed on 14 January 2025 — Meta ended the platform, describing it as a decision to prioritise the products it believed would best serve future consumer and business needs.
- Tools withdrawn — Meta Spark Studio (authoring), Meta Spark Hub (publishing and analytics) and Meta Spark Player all stopped functioning. Existing projects cannot be opened, edited or republished.
- Effects removed — third-party AR effects disappeared from Facebook, Instagram and Messenger. Meta-owned effects were retained.
- No successor named — Meta pointed third-party AR creators at nothing, and no Meta replacement has appeared since.
Meta's AR investment moved to hardware rather than to a new effects platform. Its smart glasses line with EssilorLuxottica keeps expanding, but that is a device business. It is not a surface brands can publish effects to.
Where brand social AR went instead
Social AR consolidated onto two platforms, both free to build on.
- Snap Lens Studio — the largest remaining social AR surface, and Snap's authoring tool is free to download and use. Its GenAI Suite generates ML models and 2D and 3D assets inside the authoring tool.
- TikTok Effect House — free to build and publish, with paid placement on the trending tab available separately through Branded Effects. Reach depends on the algorithm rather than on media spend alone, which makes forecasting harder and upside larger.
- WebAR you host yourself — no install, entered from a link or QR code, and the only one of the three you actually own. This is where anything that must outlive a campaign belongs.
On budget: creative production for a custom lens is quoted separately, with paid distribution budgeted on top of it. Spark's closure did not make social AR cheaper. It made the market narrower and the platform risk more concentrated.
WebAR after the 8th Wall shutdown
The second platform death matters as much as the first. The hosted 8th Wall platform was retired on 28 February 2026, and existing published experiences continue to run only until 28 February 2027, after which they go offline permanently. The tooling moved to 8thwall.org, where Image Targets, Face Effects, Sky Effects, the utilities and the example projects are fully open source under the MIT License. The XR Engine itself is not: it is distributed as a binary, free to use including commercially under a limited use license, and World Effects and Absolute Scale are available only through that binary. You can keep shipping, but the tracking core stays closed source.
If you have live 8th Wall work, that is a hard deadline and migration needs to be budgeted now rather than in late 2026. We audit what the experience actually depends on, because image-target and face-effect builds usually port cleanly while world tracking that leans on the closed engine needs a different plan.
The open web stack that replaces it is in better shape than it was. WebGPU now ships in Chrome, Edge and Safari 26 and later, but it has not reached Baseline — web platform status still records it as limited availability, so a WebGL2 fallback is mandatory rather than optional. WebXR is a W3C Candidate Recommendation. One caveat kills naive pitches: Safari does not implement the WebXR Device API on iOS, so there is no WebXR AR on iPhone or iPad. Camera-based WebAR with our own tracking still runs on iOS. WebXR AR does not.
Face tracking, consent and the law in August 2026
Face effects are the most legally exposed part of this work, and the rules moved twice in the last eighteen months.
- Illinois BIPA — written notice and a written release are required before collecting face geometry. A prevailing party may recover liquidated damages of $1,000 or actual damages, whichever is greater, for a negligent violation, and $5,000 or actual damages, whichever is greater, for an intentional or reckless one, plus reasonable attorneys' fees and costs and other relief including an injunction (740 ILCS 14/20).
- EU AI Act Article 50, live since 2 August 2026 — you must disclose that a person is interacting with an AI system, mark synthetic image, audio, video and text in machine-readable form, and notify anyone exposed to emotion recognition or biometric categorisation. Providers that placed generative AI systems on the market before 2 August 2026 get a four-month grace period, to 2 December 2026, to adapt their marking.
- EU AI Act Article 5, in force since 2 February 2025 — inferring emotions from biometric data in workplaces and education institutions is prohibited, as is biometric categorisation that infers race, political opinion, trade union membership, religious or philosophical belief, sex life or sexual orientation.
- The high-risk deadline moved, and many teams budgeted for the old one — Regulation (EU) 2026/1744, the Digital Omnibus on AI, entered into force on 27 July 2026 and pushed Annex III standalone high-risk obligations from 2 August 2026 to 2 December 2027, and Annex I embedded systems to 2 August 2028. Transparency duties were not deferred with them.
In delivery this means consent captured before the camera starts, landmark processing kept on device wherever the effect allows it, a written retention and deletion policy, and a jurisdiction matrix in the statement of work rather than in a legal annex nobody reads. MediaPipe Face Landmarker is our usual on-device baseline, returning 478 3D landmarks, blendshape coefficients and transformation matrices for rendering, and we tell clients plainly that Google still labels it a Solutions Preview early release rather than a stability guarantee.
When social AR is the wrong choice
Social AR is rented distribution. The effect lives on someone else's platform, under their policy, for as long as they choose to run it, and Spark proved the notice period can be short. That is a reasonable trade for a six-week campaign and a poor one for anything else.
- Do not put durable tools on a social platform — a product configurator, a try-on that supports sales all year, a training aid or an in-store fixture should be a web build you host and can move.
- Do not build a native headset app for consumer reach — the installed bases for Apple Vision Pro and Samsung's Galaxy XR remain small next to phones, and next to smart glasses. Headsets are a development and enterprise target, not a reach channel.
- Do not use AR when a video answers the question — if the experience does not need the user's own space, face or body to work, AR adds friction and cost without adding meaning.
- Do not accept proprietary asset containers — keep captures and models in glTF, PLY and 3D Tiles so the next shutdown does not strand your library. Gaussian splatting is now the better format for photoreal real-time visuals, and Khronos has published KHR_gaussian_splatting for glTF 2.0, but that extension is still at Release Candidate status rather than ratified. We commit to a migration path rather than claiming a ratified standard, and we keep photogrammetry where the deliverable needs metric, survey-grade geometry.
How we run an AR engagement now
Mixcore Studio has been building software for over eight years, with a team of around twenty and more than 320 projects delivered. On AR work the sequence matters more than the tooling.
- Form factor before budget — we argue the case for social AR, hosted WebAR, mobile ARCore or ARKit, or nothing at all, before anyone approves a number.
- A compatibility matrix up front — which devices and browsers get the full experience, which get the WebGL2 path, and what the degraded and no-camera behaviour is. We write it before build, not after the first bug report.
- Consent and compliance in the build — notice, consent capture, retention limits and jurisdiction handling designed into the flow rather than bolted on at legal review.
- An exit plan in the proposal — where the source, assets and analytics live, what portable formats they are in, and what happens to the experience if the host platform closes. After Spark and 8th Wall, we consider a proposal without this section incomplete.
Our expertise
- Spark AR and 8th Wall migration
- Snap Lens Studio development
- TikTok Effect House builds
- WebAR on WebGPU and WebXR
- Face tracking and computer vision
- Biometric consent and compliance
Frequently asked questions
Is Meta Spark AR still available in 2026?
No. Meta shut the Spark platform down on 14 January 2025. Meta Spark Studio, Meta Spark Hub and Meta Spark Player no longer function, and every third-party AR effect was removed from Facebook, Instagram and Messenger on that date. Only AR effects owned by Meta remain available. Any agency offering Spark AR development in 2026 is quoting for a platform that no longer exists.
Can brands publish AR effects on Instagram or Facebook now?
Not as third parties. Meta retained its own first-party effects but did not reopen the platform to brands or creators, and there is still no submission route as of August 2026. If a campaign requires reach on Meta surfaces, that has to come from conventional media formats rather than from an AR effect.
What replaced Meta Spark AR?
Meta named no successor. In practice the work split three ways. Snap Lens Studio and TikTok Effect House absorbed most social AR, both free to build on, and self-hosted WebAR took the experiences that need to outlive a campaign. Which one fits depends on the audience and on whether you need to own the experience or only rent distribution for a few weeks.
We built our WebAR campaign on 8th Wall. What happens to it?
It has a hard end date. The hosted 8th Wall platform was retired on 28 February 2026, and existing published experiences go offline permanently on 28 February 2027. Image Targets, Face Effects, Sky Effects and the utilities were open-sourced under the MIT License at 8thwall.org, but the XR Engine was not — it ships as a free binary under a limited use license, and World Effects and Absolute Scale are available only through it, so self-hosting does not hand you the tracking core as source. Image-target and face-effect experiences usually migrate cleanly; world tracking needs a different approach and should be scoped now.
Do face-filter AR effects need biometric consent?
In several jurisdictions, yes. Illinois BIPA requires written notice and written consent before collecting face geometry, with statutory damages of $1,000 per negligent and $5,000 per intentional or reckless violation. In the EU, Article 50 of the AI Act has required notice to people exposed to emotion recognition or biometric categorisation since 2 August 2026, and Article 5 has prohibited emotion inference in workplaces and education outright since 2 February 2025. We build consent capture and a retention policy into the experience rather than treating it as a legal annex.
Contacts
We are always happy to talk with you.
Feel free to contact us in any suitable way
Request a quote
Let's discuss your project!
Please, provide us with a brief description of what you
already have and what you are going to achieve.
Mail us contact@brainiacminds.com