The best virtual try-on API depends on who should wear the product in the result: a catalogue model or the shopper. Photoroom’s Virtual Try-On API puts a garment on a person you supply, such as a shopper, for $0.10 an image on its $100 a month Plus plan. For on-model catalogue photos made from a flat-lay garment photo, Photta’s virtual try-on API returns a 2K or 4K photo of the garment on an AI model for about $0.21 to $0.32 an image on its monthly plans, and it has separate endpoints for jewelry and eyewear. If the person in the picture must be the shopper, pick an API that accepts any person photo, such as FitRoom, Perfect Corp or Google Vertex AI. If cost per call matters most, Magic Hour, SnapEdit and the try-on models hosted on fal.ai list the lowest prices, and Amazon Nova Canvas suits teams already on AWS. Every fact here was read on the vendors’ own pages on 15 November 2026, Photta’s prices on 18 November 2026.
This list was made from each vendor’s own pages. Photta is a sister product of pSeoMatic (same founder). Photoroom is listed first and Photta second; the order of the other eight is not a ranking.
Photoroom vs Photta vs Google Vertex AI vs fal.ai for clothing virtual try-on by API
| Tool | Best for | Takes and returns | Covers | Starts from | Free start |
|---|---|---|---|---|---|
| Photoroom | Try-on plus background editing in one API | Garment photo plus person photo; returns an image | Clothing | $100 a month; $0.10 an image | Sandbox with 1,000 images |
| Photta | On-model catalogue photos | Garment photo plus a model ID; returns a 2K or 4K photo | Clothing, jewelry, eyewear | $19 a month; about $0.21 to $0.32 an image | 20 credits, watermarked |
| FitRoom | Low-cost clothing try-on on any model photo | Clothing photo plus model photo; returns an image up to 2048 px | Clothing | $18 a month; about $0.09 an image | Free key, limited credits |
| Perfect Corp | Clothing plus accessories from one vendor | User photo plus product image; returns an image URL | Clothing, shoes, bags, jewelry, watches | $24 a month; about $0.10 an image | 40 trial units |
| Magic Hour | Lowest listed cost per call | Person photo plus garment photo; returns an image | Clothing | $19 a month; about $0.03 to $0.06 an image | 400 sign-up credits |
| LightX | Fast responses in a shopper flow | Person photo plus outfit photo; returns a JPEG | Clothing | ₹579 a month; about ₹11.58 an image | 25 credits |
| SnapEdit | Choosing speed or quality per call | Model image plus cloth image; returns an image | Clothing | $10 a month; from about $0.05 an image | Free credits, amount not stated |
| fal.ai | Several try-on models behind one key | Person image plus garment images, or live video | Clothing, accessories on one model | $0.04 an image | Not stated |
| Google Vertex AI | Teams on Google Cloud | Person image plus clothing image; returns 1 to 4 PNGs | Clothing | $0.06 an image | Not stated |
| Amazon Nova Canvas | Teams on AWS who want mask control | Source, reference and mask images; returns up to 5 images | Clothing, footwear, accessories | $0.04 an image | Not stated |
Two kinds of virtual try-on API: a library model or any person’s photo
A virtual try-on API takes a picture of a product and a picture of a body and returns that body wearing the product. The first thing to understand is where the body comes from.
The first kind dresses a model. You send a garment photo and the ID of a model from the vendor’s library, and you get a finished catalogue photo. Photta’s API works this way. For backgrounds and packshots, see the list of AI product photography APIs.
The second kind takes any person photo. You send the shopper’s own picture and the product image in the same call, so it can sit behind a “try it on” button in a shop or an app. FitRoom, Perfect Corp, Magic Hour, LightX, SnapEdit, Photoroom, Google and Amazon all work from a person photo you supply.
Both kinds share one limit: none pastes your product pixels onto the body. The model redraws the garment from the photo, so fine print, logos, stitching and exact colour can drift. That is true of Photta as well.
The 10 best virtual try-on APIs for developers
1. Photoroom: a virtual try-on API that also edits the photo
Photoroom’s Virtual Try-On API places a garment on a person you supply: your own model, a brand ambassador or a shopper. Optional fields set a scene and a pose, and the same key reaches its background removal and resize endpoints.
- Takes and returns: garment photo plus person photo (PNG, JPEG, WEBP or AVIF, up to 30 MB and 5,000 px); returns one image per call, up to 60 images a minute by default.
- Price: included in the Plus API plan at $100 a month, which Photoroom states as $0.10 an image. It is not in the Basic plan.
- Free start: a sandbox mode with 1,000 free images; whether the sandbox covers try-on is not stated.
- Not for: small tests on a budget, since try-on needs the $100 plan. Its try-on page describes clothing only.
2. Photta: a virtual try-on API for on-model catalogue photos
Photta’s virtual try-on API for apparel takes a flat-lay garment photo and a mannequin ID and returns an on-model photo at 2K or 4K. What goes in: a front image (URL or base64, JPEG, PNG or WebP, up to 10 MB), an optional back image that requests a back shot, and a product type of top, bottom, top and bottom, or one-piece. What you choose: the model (listed with a resource endpoint, uploaded as your own photo, or generated as a new AI model), a plain backdrop colour by name or hex value, ten aspect ratios, body size from XS to XXL, facial expression, optional footwear, and a short free-text instruction. What comes out: a job ID at once, then an image URL and a thumbnail when the job completes. Photta documents 1.5 to 4 minutes per generation, polling every 3 seconds, 600 requests a minute per account, and refunds for failed jobs. Separate endpoints cover jewelry (necklace, ring, bracelet, earrings) and eyewear at the same price. The garment is redrawn from your photo, so compare each result with the real product.
- Takes and returns: garment photo plus a mannequin ID; returns one 2K or 4K on-model photo, asynchronously.
- Price: the API is included on every plan and uses the same credits. Plans are Starter $19 or $29 a month, Seller $49 or $89, Brand $119, $219 or $369; a 2K try-on works out at about $0.21 to $0.32, and 4K is on Seller and Brand. Without a plan, one-time packs of 50 credits for $12 or 100 for $20 do not expire.
- Free start: 20 credits on sign-up with no card, enough for four try-ons; photos made before paying carry a watermark.
- Not for: a live try-on button, since a job takes minutes and the API sends no webhooks (Photta sells a separate storefront widget, Photta Business, for shoppers). Its pages say poses and photo backdrops were retired from apparel try-on in August 2026.
3. FitRoom: a virtual try-on API priced at one credit a call
FitRoom’s API takes a clothing image (flat lay, hanger, ghost mannequin or product photo) and a model image and returns the model wearing the outfit. Its page lists upper body, lower body, full body and a combined top and bottom mode.
- Takes and returns: clothing photo plus model photo; returns an image up to 2048 px, “in seconds” by the vendor’s wording.
- Price: from $18 a month for 200 credits, one credit per try-on, so about $0.09 an image. Non-expiring pay-as-you-go credits are offered; their price was not readable on the page.
- Free start: a free API key with a limited number of credits; the amount is not stated.
- Not for: accessories. Its pages describe clothing only.
4. Perfect Corp: a virtual try-on API family for clothing and accessories
Perfect Corp’s YouCam API offers an AI clothes try-on endpoint and lists further try-on APIs for shoes, bags, scarves, hats, rings, bracelets, earrings, necklaces and watches. The clothes endpoint accepts full-body, half-body and selfie-style photos up to 10 MB.
- Takes and returns: user photo plus a garment image (product-only or worn by a model); returns a try-on image URL and a result ID.
- Price: billed in units, two units per clothes try-on. The vendor lists $24 a month for 500 units (about $0.10 an image) or $27.50 for 500 pay-as-you-go units (about $0.11).
- Free start: 40 trial units, enough for 20 clothes try-ons.
- Not for: teams that need a published output resolution or unit costs for the accessory endpoints; the pages read do not state them.
5. Magic Hour: a virtual try-on API with the lowest listed cost per call
Magic Hour’s AI Clothes Changer endpoint takes a person image and a garment image, with a garment type of upper body, lower body or entire outfit. It ships SDKs for Python, Node.js, Go and Rust.
- Takes and returns: person photo plus garment photo; returns one image per call.
- Price: 25 credits per clothes change. The Creator plan is $19 a month, or $144 a year for 144,000 credits, which is about $0.03 an image; a one-time pack of 4,000 credits costs $10, about $0.06 an image.
- Free start: 400 sign-up credits with no card; the page says some endpoints and resolutions need a paid plan.
- Not for: accessories. Its pages do not state an output resolution for this endpoint.
6. LightX: a virtual try-on API with a documented 15-second response
LightX’s virtual outfit try-on API takes a person photo and an outfit reference, which can be a flat lay, a mannequin shot or a photo of someone wearing the outfit. A parameter sets upper body, lower body or full body. Images must first go through LightX’s upload endpoint (JPEG or PNG, at most 5 MB).
- Takes and returns: person photo plus outfit photo; returns a JPEG, with a documented average response time of 15 seconds and status polling every 3 seconds.
- Price: two credits per generation; failed orders are not charged. Prices are listed in Indian rupees: ₹579 a month for 100 credits (about ₹11.58 an image) or ₹864 for a one-time 100 credits.
- Free start: 25 credits on API sign-up, no card.
- Not for: external image URLs, which its documentation says return an error. Output resolution is not stated.
7. SnapEdit: a virtual try-on API with three quality modes
SnapEdit’s virtual try-on model takes a model image, a cloth image, a cloth type (upper, lower or full) and a mode.
- Takes and returns: model image plus cloth image; returns one image in Normal, HD or Ultra mode.
- Price: 23, 45 or 114 credits per image by mode. Credits cost $0.002 each on a $10 monthly subscription, so about $0.05, $0.09 or $0.23 an image; a one-time 5,000 credits costs $15.
- Free start: free credits with API key sign-up, no card; the amount is not stated.
- Not for: accessories. The page does not state output resolution, file limits or rate limits.
8. fal.ai: several virtual try-on API models behind one key
fal.ai is a model hosting platform, and its try-on collection lists several models from different makers. Its own image try-on takes a person image URL and a clothing image URL. Bria’s try-on model accepts one to three garment or accessory images. Decart’s Lucy 2.1 VTON model transforms a live webcam feed from a reference garment image.
- Takes and returns: person image plus one to three garment images, returning one image; or a live video stream on the realtime model.
- Price: $0.04 an image for fal’s own try-on and for Bria’s; $0.02 per second of streaming for the realtime model. Pay per use, no plan.
- Free start: not stated on the pages read.
- Not for: teams that want one vendor answerable for results. Each model has its own maker, and the model pages read do not state licence terms.
9. Google Vertex AI: a virtual try-on API for teams on Google Cloud
Google’s Vertex AI has a model named virtual-try-on-001. You send a person image and a clothing product image and choose how many results to generate.
- Takes and returns: one person image plus one clothing image; returns one to four images as base64 PNG.
- Price: $0.06 an image on the Vertex AI price list.
- Free start: not stated on the pages read.
- Not for: accessories or a specified output size. Its documentation shows clothing only and does not state output resolution or input limits.
10. Amazon Nova Canvas: a virtual try-on API with mask control on AWS
Amazon Nova Canvas, reached through Amazon Bedrock, treats try-on as guided inpainting. You send a source image, a reference image of the product and a mask, which the model can compute from a garment class (upper body, lower body, footwear or full body) or from a text description. Options keep or regenerate the person’s pose, hands and face.
- Takes and returns: source image, reference image and mask; returns one image by default and up to five.
- Price: Nova Canvas is listed at $0.04 to $0.08 an image depending on resolution and quality; a separate try-on rate is not listed.
- Free start: not stated on the pages read.
- Not for: prompt-driven edits, since the try-on task accepts no text prompt.
How to check virtual try-on API output before you publish
Run the same few products through each API you shortlist, then check every result against the real item.
| Check | What to look for |
|---|---|
| Print, logo and text | Letters still readable and in the right place |
| Colour | Matches the real garment next to a neutral background |
| Construction | Collar, cuffs, buttons, pockets and hem length unchanged |
| Body and face | Hands, neck and hair intact; the person still looks like the input photo |
| Accessories | Stones, engraving and metal colour match the real piece |
| Personal photos | Where shopper photos are stored and for how long |
Which virtual try-on API fits your shop
- You need catalogue photos of garments, jewelry or eyewear on a model: use Photta.
- You need shoes, bags or watches on a person: use Perfect Corp.
- You need a shopper-photo try-on at the lowest listed price: use Magic Hour or SnapEdit.
- You need the quickest documented response for a shopper flow: use LightX or FitRoom.
- You need try-on and background editing from one key: use Photoroom.
- You already run on a cloud platform: use Google Vertex AI on Google Cloud or Amazon Nova Canvas on AWS.
- You want to compare several models behind one key: use fal.ai.
- You run Shopify and would rather install than build: see the virtual try-on apps for Shopify.
FAQ: virtual try-on APIs
What are the best virtual try-on APIs for clothing for developers?
Photoroom’s API takes any person photo and also edits the image, Photta is the pick for on-model catalogue photos from a flat-lay image, and FitRoom, Perfect Corp and Google Vertex AI take any person photo too and suit a shopper-facing feature. Magic Hour and SnapEdit list the lowest prices per call.
How much does a virtual try-on API cost per image?
Between about $0.03 and $0.47 on the prices listed on 15 November 2026. Magic Hour works out at about $0.03 on its yearly plan, fal.ai lists $0.04, Google Vertex AI $0.06, FitRoom about $0.09, Photoroom $0.10, and Photta about $0.21 to $0.32 for a 2K photo.
Can a virtual try-on API put clothes on a shopper’s own photo?
Yes, if the API accepts a person image with each call. FitRoom, Perfect Corp, Magic Hour, LightX, SnapEdit, Photoroom, Google Vertex AI and Amazon Nova Canvas all take a person or source photo. Photta’s API dresses a model chosen by ID, so it fits catalogue production better than a live shopper feature.
Which virtual try-on API works for jewelry, shoes or glasses?
Photta has jewelry and eyewear endpoints alongside apparel. Perfect Corp lists try-on APIs for shoes, bags, rings, bracelets, earrings, necklaces and watches, and Amazon Nova Canvas has a footwear garment class. Most of the others describe clothing only on their pages.
Will a virtual try-on API change how my product looks?
It can. Every API in this list generates the garment again from your photo instead of pasting the original pixels, so prints, logos and colour may shift. Check each result against the real product, Photta included.
Sources for this comparison of virtual try-on APIs
- Photta: the virtual try-on API, jewelry try-on API, API pricing and API rate limit pages, and llms.txt, read on 15 November 2026. Photta’s pricing page and llms.txt were read again on 18 November 2026 for its current prices.
- FitRoom, Perfect Corp, Magic Hour, LightX, SnapEdit and Photoroom: each vendor’s API product page, pricing page and, where published, API documentation, read on the same date.
- fal.ai: its try-on collection, three model pages and pricing page. Google: the Vertex AI try-on documentation and price list. Amazon: the Nova user guide and the Bedrock price list. All read on 15 November 2026.
- Cost per image is our own arithmetic from listed plan prices: plan price divided by included credits or units, times the credits or units one try-on takes.
- Nothing was tested hands-on.