ArchRender turns elevation drawings, sketches and imported 3D models into photorealistic architectural images in the browser. Season, time-of-day, camera and mood-board controls replace traditional scene setup, and everything stays organised inside projects.
Used this tool? Rate it
Used this tool? Rate it
ArchRender is a browser-based rendering platform that turns elevation drawings, sketches and imported 3D models into photorealistic architectural images without any scene-setup or lighting work. It expects the drawings and models an architectural workflow already produces, and returns presentation-ready visuals of the same building.
The distinction that matters most on first contact is between the two working modes. Render Mode builds a new image from a drawing, sketch or model; Edit Mode changes an image that already exists instead of regenerating it from scratch. Starting from a fresh elevation puts you in Render Mode with every parameter open. Coming back to a render that is almost right — wrong window frames, wrong paving, wrong sky — puts you in Edit Mode, where the rest of the image survives the change.
Iteration is therefore cheap in effort, because you are not rebuilding a scene to test a material. The tool is also opinionated about scope: the vendor positions the product as a complement to precision rendering tools rather than a replacement for them. Construction documents and measured accuracy are not what this is for; early and middle-stage visual decisions are.
The site reports more than ten thousand architects and designers, and the named testimonials come from residential design studios and home builders rather than large commercial practices.
Traditional rendering asks you to build a scene: place lights, set materials, position a camera, wait. ArchRender replaces that with three parameters. Season shows the same design in spring, summer, autumn or winter conditions so a client can read it across a year. Time of day controls how light, shadow and atmosphere fall across the facade. Environment places the building in surroundings ranging from natural landscape to more developed settings. These do the work of lighting rigs and environment maps at a fraction of the setup cost, with correspondingly less control over specifics.
For imported models you set the camera angle, the field of view and the export aspect ratio before anything is generated. Framing decisions made afterwards are expensive — a rendered image cannot be recomposed without generating it again.
The home page names OBJ, FBX and GPL as the accepted model formats, while the official blog states OBJ, FBX and GLB. The two official statements do not agree on the third format, so anyone with a model to import should confirm the accepted extensions against the application rather than either page.
Materials, style, mood and lighting can be described in natural language. The alternative matters more for design work. A single inspiration board is enough to carry colours, finishes and style into the final image, which is a different input path from writing those attributes out as a prompt. Designers who already assemble mood boards feed that artefact directly instead of translating it into words. Prompts that work can be saved and reapplied, holding a consistent house style across projects.
Still renders can be extended into short clips. A five-second clip costs five credits and a ten-second clip costs nine, so motion work draws down a monthly allowance far faster than stills do. Six camera motions are available — dolly forward, pan left, pan right, orbit left, orbit right and crane up — and the clip inherits its source image's aspect ratio, so a vertical render yields a vertical video for social placements and a widescreen render something suited to a presentation.
Output quality is bounded directly by the input: a soft or noisy source produces a soft or noisy video, so upscaling before animating is the recommended order.
Any render can be upscaled to 4K for print and client proposals. Renders and source images live together inside a project, and generating alternatives does not overwrite earlier work, so a project accumulates comparable options rather than a single current state.
The documented workflow has three steps, short enough to describe completely.
Render Mode is the entry point when the visual does not exist yet. An elevation goes in, materials are described or shown, an image comes back — the path for a first concept pass, for testing two cladding options, or for producing something a client can react to before detailed design begins.
Edit Mode handles the second half. Once a render is broadly correct, changing one element beats regenerating from the drawing with a modified prompt, which drifts. A client who likes the massing but wants a darker roof is an Edit Mode request.
Video Mode works best as a finishing step rather than an exploration tool, for reasons that come down to cost. Clips consume credits several times faster than stills and their quality is capped by the source. Settle the still first, upscale if needed, then animate; five-second clips test which motion suits a subject before committing nine credits.
Motion choice is not arbitrary. Orbits read well on standalone buildings where circling reveals new faces. Push-ins suit entrances and interiors with depth to travel through. The direction field is for mood, light and pacing — the source image already describes the building.
Three groups are named directly: architects and designers iterating on concepts, builders and developers presenting to clients, and homeowners who want to see a house before it exists. Those are different jobs, served with different intensity.
For practices the value sits in the middle of the design process. Six material studies cost minutes instead of an afternoon, and a photorealistic image reaches a non-specialist in a way a line drawing does not. Anything requiring dimensional accuracy or verified daylight analysis does not benefit, and the vendor says so itself.
Residential builders appear prominently in the testimonials and the fit is easy to see: elevations are the working currency of that business, and turning them into images that sell a home is a recurring need. The team feature matters more here than for solo practitioners, though its billing model has consequences covered below.
The third group is unusual for a professional tool. Homeowners working from plan elevations before construction get the same capability without learning rendering software. Credit plans start well above casual-use pricing, so this fits someone mid-build rather than someone browsing.
Anyone needing dimensional fidelity, anyone producing construction-stage visuals where specifications must match exactly, and anyone whose need is post-processing existing CGI rather than generating from drawings. The category has tools built for those jobs.
Everything runs in the browser through the app subdomain, and no desktop or mobile build is listed anywhere on the site. No plugin exists for Revit, SketchUp, Rhino or any other CAD or BIM environment, so models leave the authoring tool as exported files and return as images — structural, rather than a missing feature.
Team members are added by email address from the profile menu, and the arrangement has a shape worth understanding before inviting anyone.
Every render a teammate runs is charged to the owner's credits and storage, never to their own balance. There is no seat limit, which sounds generous until read from the other direction: unlimited people can spend one person's monthly allowance. Five designers on one Studio plan share 800 credits, they do not multiply them.
Members must already hold an ArchRender account before being added, so onboarding a collaborator means they sign up first.
Credits are the unit that actually limits output, so the plan question is really how many renders and clips a month of work consumes. The three tiers separate on allowance, project count and storage rather than features — model import and 4K upscaling appear on all three.
| Plan | Discounted price | List price | Credits per month | Projects | Storage | Support |
|---|---|---|---|---|---|---|
| Basic | $29 / month | $39 | 150 | 5 | 5 GB | Basic |
| Pro | $49 / month | $69 | 300 | 20 | 25 GB | Priority |
| Studio | $119 / month | $139 | 800 | Unlimited | 100 GB | Priority |
The discounted column reflects a limited-time 30% promotion; annual billing takes a further 20% off the displayed rate. Both are promotional rather than standing prices, so the list column is the safer budgeting basis.
Credit consumption is documented only for video: five credits for five seconds, nine for ten. Still-render cost per image is not published, which makes precise capacity planning impossible from the public pages. Directionally, 150 credits on Basic buys thirty five-second clips and nothing else, so any plan involving regular video work should assume the allowance goes quickly.
Failed generations do not consume credits — the platform refunds them automatically, and will not start one at all if the balance is short.
Adjacent tools solve neighbouring problems differently, and the vendor's framing is the clearest guide to when one fits better. The vendor positions the product as a complement to precision rendering tools rather than a replacement for them. If the requirement is AI rendering living inside Revit or SketchUp, BIM-plugin tools address that and ArchRender does not. If it is post-processing existing V-Ray or Lumion output, that is a different category again. What is distinctive here is the elevation-drawing input path combined with project-based organisation.
Credit purchases made through the recurring subscription are final and non-refundable except where law requires otherwise, which makes the first month a real commitment rather than a trial. The subscription can be cancelled at any time so exposure is bounded to one billing period, but no trial window undoes a purchase. Evaluate on the free tier, because the paid step carries no refund path.
Video Mode is still labelled Beta, with motion presets and output quality described as work in progress. Features in that state can change, and building a client deliverable around one carries the risk that it behaves differently next quarter.
The terms provide the service "as is" without warranties of any kind, do not guarantee it is uninterrupted, secure or error-free, and state AI results may vary and should be reviewed before use. That last clause reads as instruction rather than boilerplate.
Two product-specific constraints deserve attention. Because the published format list ends in an unrecognised extension and is stated on only one page, the public site alone cannot tell you which model files are actually accepted. And on teams the billing arrangement is a limitation as much as a feature: because usage is charged to the owner, a shared plan has no per-person guardrail and one collaborator can exhaust the month for everyone.
The vendor's own scope statement should be read as a limitation rather than modesty. Precision work is out of scope by design.
The privacy policy sets out what is collected and what rights users have. Uploaded drawings and models stay yours, and the licence granted to the platform is scoped to the processing needed to return a render. Collected information includes account details, uploaded images and anything supplied to support.
Access, correction and deletion rights are offered along with the right to object to or restrict processing, exercised by contacting the operator rather than through a self-service control. The service is declared off-limits to children under 13, and no personal data is knowingly collected from them.
The omissions matter too: nothing states whether uploaded drawings train models, no retention period is given, and no sub-processors are named. For a practice under client confidentiality obligations, those are the questions to raise before uploading anything sensitive.
Elevation drawings, sketches, photographs and 3D models. The home page lists OBJ, FBX and GPL as model formats, while the official blog lists OBJ, FBX and GLB. Because the two official pages disagree on the third format, confirm inside the application before planning around a specific export.
The workflow is built to avoid it. There is no lighting rig and no scene to assemble — you supply a drawing or model, set season, time of day and environment, then describe or show the materials. Experience still helps in judging the result, since the terms note that AI output varies and should be reviewed before use.
Each plan carries a monthly allowance: 150 on Basic, 300 on Pro, 800 on Studio. Video costs are published — five credits for five seconds, nine for ten. Per-image cost for still renders is not published. Failed generations are refunded automatically, and none can start without sufficient balance.
Yes. Video Mode works from a single source image — usually one of your renders, though any architectural image uploads fine. Choose a duration, pick one of six camera motions, optionally add a short direction covering mood and light, and an MP4 lands in the project the source image belongs to. The feature is in Beta.
They will. Credits and storage always come from the owner's balance, never the member's. There is no per-member limit and no cap on member count, so a shared plan needs an informal usage agreement rather than a technical control.
No. The product runs in the browser and the site lists no desktop build, mobile app or CAD/BIM plugin. Models are exported and uploaded, images come back to be downloaded — there is no live link with the modelling software.
The terms state purchases are final and non-refundable unless local law requires otherwise. Cancellation is available at any time and stops future billing but does not recover what has been paid. Evaluating on the free tier first is the practical answer.
No, and the vendor says so. It is positioned as a complement to precision rendering tools, aimed at early and middle-stage work where speed and clarity matter more than dimensional fidelity. Anything measured against a real specification belongs in a conventional pipeline.
