Wearlook – Virtual Try-On Fitting Room for WooCommerce
| 开发者 | gopalbogati |
|---|---|
| 更新时间 | 2026年11月2日 13:11 |
| PHP版本: | 7.4 及以上 |
| WordPress版本: | 7.1 |
| 版权: | GPLv2 or later |
| 版权网址: | 版权信息 |
详情介绍:
- A visitor clicks Try It On and uploads a photo of themselves. The photo never leaves their browser.
- If the product has more than one photo, they choose which image to try on — a different angle, or a different colour.
- The garment is fitted onto their photo automatically — Wearlook measures the visitor's shoulders and scales the garment to them, and if the product photo is shot on a model it cuts the garment off the model first. They can still drag, resize and rotate to perfect the fit, press Auto-fit to go back to the measured placement, then download the look.
- (Optional premium) Businesses can connect a Wearlook Cloud account to add an AI render button that produces a photorealistic image of the garment on the visitor's body.
- Any WordPress site — add
[wearlook]to any page or post with a featured image, or point it at any media-library image:[wearlook image="123"] - WooCommerce (optional) — the button appears automatically on every product page with a featured image. Per-product disable checkbox included.
- Blogs, portfolios, lookbooks, boutiques, designers — anywhere clothes are shown.
- Products with several photos show a thumbnail strip at the top of the try-on window.
- It offers the featured image, the full product gallery, and — on variable products — the variation images, so a shopper can try the same garment in another colourway.
- Switching image keeps their uploaded photo and the fit they have already set, so comparing two colours on themselves takes a single tap.
- Products with a single image are unchanged: no picker, no extra clicks.
- Choose Google Gemini, FASHN.ai or fal.ai and paste your own API key — renders are billed to you by that provider, at their prices (both are paid services; expect a few cents per image).
- fal.ai lets you pick between several try-on models, so you can trade cost against quality.
- Or point the plugin at a Wearlook Cloud server.
- Leave the engine on None and the plugin stays completely free, with no external calls at all.
- Free (everyone): unlimited in-browser fitting room, automatic garment placement and background removal, download-your-look. Zero configuration, zero data collection.
- Premium (businesses): photorealistic AI renders via Wearlook Cloud credits, with a consent checkbox and per-visitor rate limiting to protect your balance.
安装:
- Upload the plugin ZIP via Plugins → Add New → Upload, and activate it.
- That's it — on WooCommerce product pages the button appears automatically. On any other page, add the
[wearlook]shortcode (uses the featured image) or[wearlook image="123"]. - (Optional premium) Go to Settings → Virtual Try-On and enter your Wearlook Cloud API URL and key to enable photorealistic AI renders.
屏幕截图:
![Works on any WordPress site with the [wearlook] shortcode — no WooCommerce needed.](https://cdn.jsdelivr.net/wp/wearlook/assets/screenshot-3.png)
常见问题:
Can I show the button only on clothing?
Yes. Under Settings, Virtual Try-On, set "Show the button on" to "Only products recognised as clothing" for a hands-off setup, or pick the exact categories. Choosing a category includes everything beneath it. If a garment with an unusual name is missed, set its garment type under Product data, Advanced and the button appears.
How does Wearlook know a product is trousers and not a shirt?
It reads the product name, then its categories, for garment words such as jeans, shorts, skirt, dress or jacket, and uses the last one it finds: a "jean jacket" is a jacket, "polo shorts" are shorts. Sleeve lengths are ignored, so "short sleeve tee" stays a top. Each product's edit screen shows what was detected, and you can override it. Developers can add words in other languages with the wearlook_garment_keywords filter.
Does it need WooCommerce?
No. It works on any WordPress site via the [wearlook] shortcode. If WooCommerce is active, product pages get the button automatically.
Is it really free?
Yes — the fitting room (automatic garment placement, drag/resize/rotate, download) is free and unlimited with no account. Only the optional photorealistic AI render uses a paid Wearlook Cloud subscription.
Do visitor photos leave their device?
Not in the free fitting room — everything runs in the visitor's browser. A photo is only transmitted (over HTTPS, after explicit consent) when the visitor requests a premium AI render, and it is processed in memory, never stored.
How does the automatic fitting work?
Wearlook reads the visitor's photo, separates them from the background and measures their shoulders, then scales the garment to that measurement and hangs it from their shoulder line — so the same garment comes out larger on a close-up selfie and smaller on a full-length photo. The visitor can still fine-tune with drag and the size/rotate sliders. All of it runs in their browser; nothing is uploaded. If the photo is unusual — a busy background, or the head cropped out of frame — Wearlook says so to itself and falls back to a sensible fixed placement rather than guessing. The sliders are always there, and the Auto-fit button re-runs the automatic placement at any time, so no amount of dragging is a dead end. A new photo is always fitted automatically. Switching to another product image keeps the size and rotation already dialled in, so comparing two colourways on yourself stays one tap — press Auto-fit if you would rather start that image fresh.
How much does the on-device AI add to page load?
Nothing on an ordinary product page. The pose model is only fetched the first time a shopper actually opens the fitting room and prepares a photo. That first use pulls about 17 MB: roughly 11 MB of MediaPipe runtime and 5.5 MB for the Pose Landmarker Lite model. The browser then caches both, so it is a one-off per shopper, and the files come from your own site rather than from Google. Measuring a photo takes about half a second the first time and under a tenth of a second after that. Shoppers on a slow connection can still place the garment by hand while it loads.
My product photos are shot on a model, not laid flat. Does that work?
Yes. When Wearlook finds a person in the product photo it cuts the garment out of it — the model's head, arms and trousers are removed and only the garment is placed on the visitor. Flat-lay photos are used as they are, with the studio background knocked out.
Can I use any image as the garment?
Yes — [wearlook image="URL-or-attachment-ID"] works with any image in your media library. Pass several, separated by commas, to let the visitor choose: [wearlook image="12,13,14"].
Can the visitor choose which product image to try on?
Yes. When a product has more than one image, a thumbnail strip appears at the top of the try-on window. It offers the featured image, the product gallery, and the variation images of variable products — so a visitor can try the same garment in another colour or from another angle. Switching images keeps their uploaded photo and current fit, and the choice is validated server-side against that product's own images.
Can I use my own AI API key?
Yes. Under Settings → Virtual Try-On pick Google Gemini or FASHN.ai as the AI engine and paste your own key. Renders are billed to you directly by that provider, and the plugin adds no fee. Your key is stored in this site's database.
Which AI engine should I choose?
fal.ai is the most flexible: one key gives you a choice of purpose-built try-on models, so you can pick a cheaper or a higher-quality one from the settings dropdown. Google Gemini is usually the cheapest per image and the key takes a minute to create in Google AI Studio. FASHN.ai is purpose-built for garment try-on. Both are paid services. AI image generation is not included in Google's free API tier, so your Google account needs billing enabled; FASHN.ai is likewise a paid API. Expect roughly a few cents per render. Check each provider's current pricing before enabling renders on a busy store, and use the "AI renders per visitor per hour" limit to cap what visitors can spend on your behalf. If you would rather not pay anything, leave the engine on None — the fitting room stays free and unlimited.
Can I disable it for some WooCommerce products?
Yes — every product has a "Disable virtual try-on" checkbox under Product data → Advanced.
How do I stop one visitor using all my AI credits?
Premium renders have a per-visitor hourly rate limit, configurable in settings (default: 10 per hour).
更新日志:
- Wearlook now knows what kind of garment each product is. Jeans, shorts, skirts and leggings are fitted on the hips; tops, jackets and knitwear from the shoulders; dresses and jumpsuits from the shoulders down. It reads the product name and categories, so there is nothing to set up.
- Override any single product under Product data, Advanced, where Wearlook also shows what it detected.
- New "Show the button on" setting: all products, only products recognised as clothing, or only chosen categories. Stores that also sell homewares, accessories or gift cards no longer get a Try It On button on a frying pan.
- On-model product photos of trousers and skirts are now cut out at the hips, so the shopper gets the jeans and not the shirt above them.
- Better cut-outs from on-model photos in general: pale denim, grey and white garments separate cleanly from studio walls and floors, and skin is left out of yellow and tan garments.
- Hidden products are refused by the AI render and MCP abilities too, not just hidden from the page.
- AI renders and MCP try-ons now send each product's own garment type to the AI provider.
- The [wearlook] shortcode accepts type="bottoms" (or tops, one-pieces).
- Declares WooCommerce 11.2 and Cart and Checkout blocks compatibility.
- AI assistants can now use Wearlook over MCP. Three abilities are registered with the WordPress Abilities API on WordPress 6.9 and later: check status, list a product's try-on images, and render a try-on.
- Works with the official WordPress MCP Adapter. Nothing is exposed until you install an MCP server plugin yourself.
- Try-on over MCP requires an administrator capability, because renders are billed to your own AI provider account.
- No change on WordPress 6.8 and earlier — the abilities are simply not registered.
- Bundled on-device shoulder/hip landmark detection for automatic alignment, including busy backgrounds. No external model service or API key is needed for the free preview.
- Clear manual-adjustment guidance when a single person's landmarks cannot be confidently detected.
- Fixed stale image loads overwriting the selected colour or reappearing after closing the fitting room.
- Reset photo consent between sessions, preserve drag adjustments across image changes, and release downloaded image URLs.
- Align tilted garments around the collar and use measured hip positions for the bottoms setting.
- Automatic fitting now measures the visitor. Wearlook separates them from the background, finds their shoulders and scales the garment to that width instead of always using a fixed fraction of the photo — so a close-up selfie and a full-length photo both come out right.
- Product photos shot on a model are handled properly. The garment is cut out of the model photo — head, arms and trousers removed — instead of the whole model being pasted onto the visitor.
- Studio padding around a product photo is trimmed, so the garment is sized by its own edges rather than by the whitespace around it.
- New Auto-fit button. Drag the garment or move the sliders as much as you like, then press Auto-fit to snap straight back to the measured placement.
- Drag, size and rotate are unchanged and still available on every fit.
- When a photo cannot be read confidently — busy background, head out of frame — the previous fixed placement is used rather than a bad guess.
- Plugin title now describes what it does, so shops can find it by searching. Same plugin, same settings, nothing to change.
- Plugin author details updated. No functional changes.
- Bring your own AI key: choose Google Gemini, FASHN.ai or fal.ai as the render engine and pay that provider directly — no Wearlook account needed.
- fal.ai gives a choice of try-on models (FASHN v1.6, Kling Kolors v1.5) so you can trade cost against quality.
- New "AI engine" selector; only the key field for the chosen engine is shown.
- Clear provider error messages are passed through (for example an invalid API key now says so).
- Existing sites using a Wearlook Cloud URL and key keep working exactly as before.
- Visitors can now choose which product image to try on, instead of always getting the first one. Requested by a store owner selling garments photographed from several angles and in several colours.
- The picker offers the featured image, the product gallery, and — on variable products — the variation images, so colourways can be tried on directly.
- Switching image keeps the uploaded photo and the current size/rotation, so comparing colours is one click.
- The shortcode accepts a list:
[wearlook image="12,13,14"]. - The chosen image is validated server-side against that product's own images.
- Camera photos, including browser-decodable HEIC images, are resized and converted to JPEG before AI upload.
- White garment details are preserved more reliably during browser background removal.
- Failed AI-provider requests no longer consume a visitor's hourly allowance.
- Works on every WordPress site — WooCommerce is now optional.
- New
[wearlook]shortcode: featured image, attachment ID, or image URL. - Free tier needs zero configuration; premium AI section only appears when a cloud account is connected.
- Settings moved to Settings → Virtual Try-On.
- Multiple try-on buttons per page supported.
- First release: fitting room with automatic garment placement, optional AI renders, consent checkbox, per-visitor rate limiting, per-product disable.