Fenestra logo
Image Generators · Video Generators

Fenestra

Fenestra is browser-based AI rendering software for architects and interior designers. It turns hand sketches, CAD drawings, photos and 3D models into photorealistic renders and cinematic animations in seconds, with no plugin, render engine or dedicated GPU.

Active GDPR compliant Free plan Freemium No public API 18+ Verified by Guidaio
Overview

What is Fenestra?

Fenestra is AI rendering software aimed squarely at architects and interior designers. Published by Designtree AI LLC, a company registered in Sharjah Media City in the United Arab Emirates and founded in 2024, it gathers a full visualisation workflow into a single browser-based studio: no plugin to add to your CAD software, no render engine to configure and no dedicated GPU, so the same work runs from a laptop.

The studio is organised around six tools. Image generates renders from a prompt, a sketch or a reference image. Edit handles conversational retouching in plain language, with rectangle, brush and editable polygon masks, and imports 3D models in GLB, FBX or 3DM up to 150 MB to render straight from the geometry with full camera control. Realtime renders live as you draw. Enhance upscales and sharpens output to 8K. Video animates a still into a cinematic clip using archviz camera presets such as pan, dolly and orbit. A dedicated 3D workflow completes the set.

What distinguishes the product is aggregation combined with control. Nineteen AI models are curated and exposed behind one subscription, among them Nano Banana 2, GPT Image 2, Krea 2, Seedance 2.0, LTX 2.3 and several Flux variants, so users do not juggle separate platforms. The vendor's argument is that generic AI output is not enough for practice work: custom style libraries, reference images and prompts are meant to keep renders recognisably yours, and rendering from an uploaded model or drawing follows the geometry rather than reinventing it.

Accepted inputs are wide: JPG, PNG and WEBP images, PDF drawings, and native Rhino .3dm files alongside GLB and FBX exports from Revit and SketchUp. Usage runs on credits, roughly one per image, two per enhancement and ten per video, with the cost of a generation shown before it runs.

The company claims more than 20,000 architects and designers, 200,000 generations and 11,000 projects, and publishes named testimonials from practices in Australia, Germany and Spain. Twenty-three solution pages cover landscape design, timber construction, heritage, education, real estate and sustainability. The changelog and blog were both current at the time of review.

What it does

  • Turn a hand or digital sketch into a photorealistic render in under 30 seconds
  • Render a Revit, Rhino or SketchUp model directly from its geometry
  • Render 2D CAD plans, sections and elevations imported as a PDF or an image
  • Edit an image in plain language: add people, remove furniture, change a facade
  • Stage an empty interior with furniture, optionally matching real reference pieces
  • Upscale and sharpen an existing rendering up to 8K
  • Animate a still render into a cinematic flythrough clip
Audience

When to use Fenestra / When not to

A quick filter to help you decide if Fenestra is the right fit.

When to use Fenestra

  • Architects who need a client-ready visual from a concept sketch or a Revit model before any 3D scene is finished
  • Interior designers testing materials, lighting and furniture arrangements, or staging empty rooms
  • Real estate and marketing teams cleaning up listing photography and turning a single shot into a short video
  • Landscape architects and urban designers illustrating a scheme on the site as it actually stands
  • Architecture students and solo designers building portfolio visuals without access to a render farm

When not to use Fenestra

  • Teams needing physically accurate lighting or daylight analysis, since output is generative rather than simulated
  • Developers looking to automate rendering, as Fenestra publishes no API and no documented endpoints
  • Anyone wanting a mobile or desktop application, the product being web-only
  • Organisations with strict procurement requirements, given the absence of a DPA, of SOC 2 or ISO 27001 certification and of an EU Article 27 representative
  • Non-English-speaking users who need a localised interface, no other language being offered
Get started

How to use Fenestra

A typical end-to-end flow, from setup to results.

  1. Create an account from the login page, by email or through a Facebook or X account
  2. Start with the five rendering credits granted to every new account, or pick a paid plan
  3. Choose a tool from the top navigation bar: Image, Edit, Realtime, Enhance or Video
  4. Upload your starting point: a sketch, a photo, a PDF drawing, or a 3D model in GLB, FBX or 3DM
  5. For a 2D drawing, select the page and crop to the plan or elevation you want to render
  6. Describe the scene in plain language in the prompt bar, and add style references if you have them
  7. Check the credit cost shown before generating, then run the generation
  8. Refine locally in Edit with a rectangle, brush or polygon mask, leaving the rest of the image untouched
  9. Use the camera widget to orbit, raise, lower or zoom for an additional viewpoint from the same image
  10. Enhance to 8K for delivery, or send the still to Video to produce an animated clip
Quick read

Pros & Cons

Pros

  • Nineteen curated AI models behind a single subscription, with no platform juggling
  • Very wide range of inputs: sketches, 2D CAD, PDF, photographs and native Rhino files
  • Nothing to install and no dedicated GPU required, so it runs on an ordinary laptop
  • Renders announced in under 30 seconds where a traditional engine takes hours
  • Credit cost displayed before each generation, which makes spending predictable
  • Commercial licence included from the entry-level paid plan upward
  • Work is preserved after cancellation, and the product is visibly maintained with a public changelog

Cons

  • No API whatsoever, which rules out automation or pipeline integration
  • Web only: no mobile and no desktop application
  • No DPA, no SOC 2 or ISO 27001 certification, and no EU Article 27 representative despite a non-EU publisher
  • The site is hosted in the United States with no data residency option
  • All purchases are non-refundable, and the licence taken over uploaded content is unusually broad
  • Terms and conditions read as an unfinished template, with a blank free trial duration and an incomplete postal address
  • Pricing is presented inconsistently across the site, and several advertised features are still marked Coming Soon
Pricing

Pricing & Plans

A permanent free plan is available: every new account receives five rendering credits, and an account that cancels a subscription reverts to that free plan with its projects and generations preserved. The lowest paid entry point is the Individual plan at USD 10 per month, which includes 200 credits, or USD 96 per year when billed annually, a saving of 20%. All paid plans carry the same 20% annual discount, and payment is taken in US dollars by card or PayPal through Lemon Squeezy. Purchases are non-refundable.

Plan 1
  • Free — five rendering credits on sign-up
  • the plan an account reverts to after cancellation
Plan 3
  • Pro — USD 35 per month or USD 336 per year
  • 1
  • 000 credits
  • marked as the most popular tier
Plan 4
  • Studio — from USD 60 per month or USD 576 per year
  • with 2
  • 000 to 8
  • 000 credit tiers
Plan 5
  • Team — from USD 250 per month or USD 2
  • 500 per year
  • 5 to 25 seats on a shared credit pool
Plan 6
  • Enterprise — custom pricing
  • including training a model on the practice's own style
Special offers — 20% discount on every paid plan when billed annually · An additional 20% student discount on any plan, granted on request by email · Five rendering credits granted free on account creation · Decreasing credit unit cost: Pro is stated to be 30% cheaper per credit than Individual, and Studio 40% cheaper
Prices and plans listed above may evolve. Always check the official pricing page before subscribing.
Trust & Privacy

Data, GDPR & hosting

A consolidated view of how Fenestra handles your data.

GDPR overview

The privacy notice, dated 1 January 2025, engages with the GDPR explicitly. It sets out the legal bases relied on under the GDPR and UK GDPR (consent, performance of a contract, legitimate interests, legal obligations, vital interests) and lists the rights available in the EEA, the UK, Switzerland and Canada: access, rectification, erasure, restriction, portability, objection and non-subjection to automated decision-making. It names the right to lodge a complaint with a Member State authority or the UK regulator, and rights are exercised by email. Two gaps stand out for a controller established in the United Arab Emirates: no EU representative is designated under Article 27, and no data protection officer is named. No Data Processing Agreement is published or offered, and no security certification is claimed.

Who owns the data?

The terms state that Fenestra asserts no ownership over user contributions: you keep full ownership of the files you upload and of the associated intellectual property rights. That ownership sits alongside a very broad licence granted to the vendor over the same content, described as unrestricted, irrevocable, perpetual, worldwide, transferable and sub-licensable, including for commercial and advertising purposes, with a waiver of moral rights. Feedback and suggestions are treated differently and become the vendor's exclusive property. Commercial use of generated images is included from the Individual plan upward and on every Team seat, and the vendor states that custom-trained models belong to the customer and are not shared.

Reuse rights

Customers may reuse what they generate without asking permission: a commercial licence is bundled with the Individual, Pro and Studio plans for a single user, and with every seat on a Team plan, which the vendor presents as suitable for practices, visualisation studios and real estate teams producing client work. Uploaded material and generated output remain the customer's, and projects, generations and assets are preserved even after a subscription is cancelled and the account reverts to the free plan. On the vendor's side, inputs and outputs are shared with named AI service providers (Hugging Face, Stability.ai, Black Forest Labs and OpenAI) in order to run the models, while payment data is handled entirely by Lemon Squeezy. The vendor states that it does not train on customer data, and that it has neither sold nor shared personal information in the previous twelve months.

Data retention & training

Retention summary
The privacy notice keeps retention tied to the life of the account: personal information is kept only as long as needed for the stated purposes, and no purpose is said to require keeping it beyond the period during which the user has an account. Identifiers, customer records, geolocation and any image or audio data follow that same rule. When you ask for your account to be closed, the data is deactivated or deleted from the active databases, though some may be retained to prevent fraud, troubleshoot, support investigations or meet legal obligations. Anything left in backup archives is isolated from further processing until deletion becomes possible. Projects, generations and uploaded assets survive a cancellation. No retention period is published as a figure.
Trains on customer data
No
Subprocessors disclosed
Yes
GDPR contact

Hosting summary

The terms state plainly that the site is hosted in the United States, and that accessing it from another region amounts to transferring your data there for processing. DNS resolution is consistent with this: the domain points to an anycast node on Amazon infrastructure located in the United States. No data residency option is offered, and no EU or other regional hosting is proposed. Beyond the vendor's own infrastructure, inputs and outputs are processed by named third-party AI providers — Hugging Face, Stability.ai, Black Forest Labs and OpenAI — and payment data is held by Lemon Squeezy, but the vendor does not say where any of those providers host the data they receive. There is no trust or security page, no published hosting certification and no Data Processing Agreement. For a customer subject to European data protection requirements, this is the weakest documented area of the service.

Hosting countries
🇺🇸 United States
Watch-outs

Things to keep in mind

Risks and trade-offs to weigh before adopting Fenestra.

  • The licence granted over uploaded content is perpetual, irrevocable, sub-licensable and exploitable commercially, even though you keep ownership
  • Feedback and suggestions you send become the vendor's exclusive property under the terms
  • The claim that data is not shared with third parties sits awkwardly beside the documented transfer of inputs and outputs to four AI providers
  • Generative output is not a physical simulation: renders can look convincing while quietly misrepresenting light, materials or proportion in a client presentation
  • Leaning on instant AI renders early can erode the drawing and modelling judgement that lets a designer spot when an image is wrong
  • All purchases are non-refundable, and pricing shown on the site is inconsistent enough to warrant checking the amount before subscribing
  • Governing law and jurisdiction are those of the United Arab Emirates, with arbitration seated in Abu Dhabi, which makes any dispute costly to pursue from elsewhere
Setup

Setup & Integrations

Technical difficulty

Very low. There is nothing to install: no CAD plugin, no render engine to configure and no dedicated GPU. You sign up by email or through a social account, receive five credits and can generate immediately. The only real technical step is exporting your model in a supported format — FBX or GLB from Revit and SketchUp, native .3dm from Rhino — within the 150 MB limit. No code and no integration work is involved, and the documentation offers a quick-start path measured in minutes.

Deployment

Web app

Integrations

Revit Rhino SketchUp AutoCAD
Company

Behind Fenestra

Company name
Designtree AI LLC
Founded
04/01/2024
Country of origin
🇦🇪 United Arab Emirates
Headquarters
Sharjah Media City, Sharjah, United Arab Emirates
UBO
Shaun McCallum
UBO country
INFORMATION_NOT_FOUND
Domain registrar country
🇺🇸 United States
Legal contact
Support contact

Social

Official links

Resources

All the official URLs gathered for verification and reference.

Compare

Alternatives

Tools that compete with or complement Fenestra.

L LumionE EnscapeD D5 Render
FAQ

Frequently asked questions

What does Fenestra actually do?
It turns sketches, CAD drawings, photographs and 3D models into photorealistic architectural renders and animated clips, inside a single browser-based studio aimed at architects and interior designers.
Do I need to install a plugin or a render engine?
No. Fenestra runs entirely in the browser. There is no plugin for your CAD software, no render engine to configure and no dedicated GPU, so it works from an ordinary laptop.
Which files can I upload?
Images as JPG, PNG or WEBP, PDF drawings in the Edit tool, and 3D models as GLB, FBX or 3DM up to 150 MB. Rhino files import natively, while Revit and SketchUp scenes are exported as FBX or GLB.
How much does it cost?
A free plan gives five rendering credits on sign-up. Paid plans start at USD 10 per month for Individual with 200 credits, then USD 35 for Pro, from USD 60 for Studio and from USD 250 for Team. Annual billing saves 20%.
How do the credits work?
Each generation draws on your monthly allowance: roughly one credit per image, two per enhancement and ten per video, with higher-fidelity models costing more. The exact cost appears before you generate, and credits refill at the start of each billing cycle.
Can I use the images for client work?
Yes. A commercial licence is included from the Individual plan upward for a single user, and on every seat of a Team plan, which the vendor presents as suitable for practices, visualisation studios and real estate teams.
Is my data used to train the models?
The vendor states that it does not train on customer data and that stored data is not shared with third parties. Note that inputs and outputs are nevertheless passed to named AI providers — Hugging Face, Stability.ai, Black Forest Labs and OpenAI — in order to run the models.
Is there an API or a mobile app?
Neither. Fenestra publishes no API documentation, and the product is a web application only, with no iOS or Android release.
Conclusion

Should you pick Fenestra?

Fenestra is a genuinely specialised tool rather than a general image generator dressed up for architecture. The six studio tools, the nineteen curated models and the breadth of accepted inputs — sketches, PDF drawings, photographs and native Rhino geometry — add up to a workflow that covers concept through to client presentation, and the credit economics are competitive: around 200 renders a month for USD 10. The absence of any installation is a real advantage for practices whose machines are not built for rendering.

The reservations are mostly institutional rather than technical. This is a young, small publisher: a single personal email address serves as the legal, GDPR and support contact, front-line support runs through a chat widget, and the terms and conditions are visibly an unfinished template, with a blank free-trial duration and a postal address containing a gap. There is no DPA, no security certification and no EU Article 27 representative, which will matter to any organisation with a formal procurement process. Pricing is presented inconsistently, the plan cards showing a struck-through figure that matches nothing stated elsewhere on the site. The licence taken over uploaded content is also unusually broad, even though ownership itself stays with the customer.

Set against that, the product is unmistakably alive, with a changelog and a blog both current at review, and named testimonials from practices in three countries. For concept work, iteration and client presentation, Fenestra is convincing; where physical accuracy or contractual guarantees are required, it belongs beside a conventional render engine rather than in place of one.