Fully updated August 2026. This post was originally written in 2022 about Instagram AR filters. Meta shut down Spark AR in January 2025, ending branded filter creation on Instagram and Facebook entirely. The eight filter types are still valid — the platforms and tools have just moved on. TikTok (Effect House) and Snapchat (Lens Studio) are now where branded AR lives, and WebAR has become a serious alternative for campaigns that need to reach everyone without an app download.
Custom AR filters used to be Instagram’s thing. Now they’re TikTok and Snapchat’s thing, and they work better there anyway — the audiences are bigger, the sharing loops are faster, and the tools brands have access to are more sophisticated than Spark AR ever was.
If you’re a brand thinking about commissioning a filter, or an agency spec-ing one out, here are the eight types you’ll encounter. Each one works differently, serves a different goal, and costs a different amount to build well.
If you want to understand the cost side, we wrote a full breakdown of AR filter pricing here. And if you’re trying to decide between TikTok effects, Snapchat lenses, and WebAR for your next campaign, this comparison covers the differences in detail.
1. Colour and Tone Filters
The simplest category. Colour filters adjust the camera feed in real time — shifting hues, boosting saturation, applying a tone or mood to everything the camera sees.
They sound basic, but a well-built colour filter tied to a brand’s visual identity is immediately recognisable and shareable. The effect is instant: anyone who turns it on looks like they’re inside your campaign’s colour world.
TikTok’s colour filters are particularly effective because the platform’s video format rewards visual consistency. A brand with a strong colour signature — a specific shade of green, a warm analogue tone, a high-contrast monochrome — can build name recognition just through users sharing footage shot in their filter.
These are among the lowest-cost filters to build, but they reward strong creative direction on the colour itself.
2. Beauty and Makeup Filters
Beauty filters apply virtual makeup and skin effects in real time — smoothing, contouring, adding lip colour, eyeshadow, lashes, or entirely new looks layered over the user’s face.
For cosmetics and skincare brands, these are the obvious choice. A user can try your product’s signature red, your season’s eyeshadow palette, or your foundation range — without touching anything. The try-before-you-buy mechanic works especially well for new product launches.
Beauty filters on Snapchat are technically advanced. Lens Studio supports full facial mesh tracking with precise UV mapping, which means the makeup sits correctly on real skin tones and moves naturally as the user moves. TikTok’s Effect House has comparable tools.
Outside cosmetics, beauty filters have been used effectively by music artists (giving fans a “show look” to share), fashion brands (adding accessories like sunglasses or jewellery), and film releases (character transformations).
3. Random and Personality Filters
These are the viral mechanic filters. The user taps, something cycles on screen, and it lands on a random result: “Which of our products are you?” “What flavour are you?” “Which character are you?”
They work because of the gap between what you hope to get and what you get — and because the result is worth sharing to your followers either way. The format originated with “Which Disney character are you?” filters on Instagram Stories and has never really gone out of use because the basic human interest in personality categorisation doesn’t go away.
For brands, the format maps well to product ranges, character rosters, seasonal campaigns, and team promotions. The key to making them stand out in 2026 is animation quality and creative framing — the base mechanic is familiar to everyone so the execution has to earn the share.
4. Quiz and Interactive Filters
A step up from random filters in complexity. Instead of a single spin-the-wheel outcome, quiz filters present a sequence of choices — typically three to five — and deliver a personalised result based on the user’s answers.
The filter can ask about preferences, reactions to prompts, or emoji choices. The payoff is a result that feels earned rather than arbitrary, which tends to drive higher engagement and more considered sharing.
Quiz filters require more development time because the branching logic needs to be mapped out and each path needs its own visual result. They work best when the outcome options are genuinely distinct and the path to each feels surprising or satisfying.
Good use cases: product recommendations, “which [brand] personality are you” campaigns, advent-style reveals, and audience segmentation campaigns where the filter result directs users to different parts of a campaign.
5. Game Filters
Game filters are interactive experiences — the user doesn’t just look at the filter, they play it. The most common format is catching falling objects (products, brand icons, characters), but good game filters go further: racing mechanics, reaction tests, rhythm games, physics puzzles.
On TikTok particularly, game filters have strong sharing mechanics because the format naturally produces comparison content — “I got 47, beat me.” Leaderboard mechanics and score-sharing are built into the TikTok Effect House toolkit.
The quality bar has risen significantly since the early Instagram days. A basic catching game with no animation polish, no difficulty ramp, and no second mechanic will not get plays in 2026. The filters that gain traction are the ones with a hook inside the first three seconds and a reason to replay.
Game filters are among the more expensive to build correctly, but they are also the type most likely to generate organic reach without paid amplification. For branded AR campaigns where engagement is the primary metric, games consistently outperform passive filter types.
6. World AR Filters
World AR filters use the rear camera and place virtual objects into the user’s real environment — on a surface, in the air, next to a person. The virtual object appears to exist in the real space: it has scale, it casts a shadow (if the filter is well built), and it tracks as the user moves.
The format became famous with the viral PS5 box filter, which let users “receive” a PlayStation 5 in their living room before the console was available. The mechanic — placing a product you don’t own yet into your own space — is uniquely compelling.
Brand applications: product previews, retail scale demonstrations, architectural or event previews, character or mascot appearances in branded environments, and campaign activations that place a brand asset somewhere it wouldn’t normally exist.
World AR filters require solid 3D modelling of the asset being placed. For brands with existing product models (retail, automotive, electronics), the barrier is lower. For brands without 3D assets, the modelling cost is part of the production.
World AR is also the natural bridge to WebAR — the same spatial placement mechanic, but delivered through a browser link rather than a social platform filter. For campaigns where reach beyond TikTok or Snapchat audiences matters, WebAR covers the gap.
7. Target Tracker Filters
Target tracker filters are triggered by a specific image — a product package, a poster, a QR-adjacent graphic, or a print ad. The filter only activates when the camera sees that trigger image, and it overlays animation, information, or an interactive experience on top of it.
The result is packaging or print that comes alive. A coffee cup sleeve triggers a steam animation and a promotional reveal. A poster triggers a character walking out of it. A festival wristband triggers a show-time overlay.
Target trackers require a high-contrast trigger image with distinctive features — flat colour backgrounds and fine text do not track reliably. The filter developer will assess the trigger image before committing to the build.
This type has strong crossover with WebAR use cases. Brands that want the target tracker experience without limiting it to Snapchat or TikTok users typically build it as WebAR — which works via a link or QR scan in any phone browser.
8. Immersive Filters
Immersive filters place the user inside a scene. Rather than adding an element to what the camera sees, the filter replaces the background entirely — the user’s face is composited into a different world.
Strong examples: a brand’s TV ad environment becomes a filter the audience can step into; a sports team’s filter puts fans on the pitch at the stadium; a musician’s filter puts fans inside the music video.
The immersive format works well for launches, cultural moments, and fan engagement because the emotional pull of “being inside” the campaign is stronger than watching it. Users who share immersive filter content are sharing an experience they participated in, not just a product they applied.
Production-wise, immersive filters require well-designed 3D or composited scenes and careful masking of the user’s figure against the new background. The quality of that cutout — especially around hair and edges — determines whether the filter feels premium or cheap.
Which type is right for your campaign?
The right filter type depends on your goal. For brand awareness and reach, game filters, immersive filters, and personality filters consistently perform. For product try-on and consideration, beauty filters and world AR work best. For campaign activations tied to events or physical media, target trackers and world AR are the natural fit. For audience engagement and data, quiz filters and game filters give the most back. For ambient brand presence, colour filters and beauty filters are the low-maintenance choice.
Most campaigns use one type cleanly. Some — particularly larger brand activations — combine a game mechanic with a world AR element, or a personality filter with an immersive reveal.
We build all eight on TikTok and Snapchat, and most of them as WebAR experiences for campaigns that need to reach beyond social platform audiences. If you know which type you want, or if you’re trying to figure out which fits your brief, we’re the people to ask.