AI rendering from any 3D software, or from a photo: one image is enough

AutoCAD, Blender, 3ds Max, Rhino, Cedreo, Twinmotion, or a photo of an already furnished room: if your tool exports an image, RenderThink turns it into a photorealistic render in about a minute, faithful to the volumes and materials of the view. No proprietary format, no plugin to install.

Try it free

14 days, no commitment

Median 1K render, under 30 seconds: 28 s
Ready for large-format printing: 8K
Free trial, no commitment: 14 days

If your software exports an image, it already works with RenderThink.

An image export as PNG, JPG or WebP

RenderThink does not read 3D files: it starts from an image. Export the view from your software, at the highest resolution possible and preferably PNG, or photograph the room, then drop the file into the Studio, in the browser or in the desktop app. It is the same workflow as from Winner Design, Archicad or Revit; SketchUp, InSitu and KitchenDraw also have a plugin that captures the displayed view.

Export or photograph the view
One image per view, camera at eye level, materials already applied in the model, no dimensions or annotations. PNG, JPG or WebP, 10 MB and 8,192 px at most. A smartphone photo of a furnished room works too.
Drop it in and start the render
Drag the image into the project, pick the resolution, from 1K draft to 8K, and start several views at once. Each render comes back in about a minute, the computation runs on our servers.
Edit on the render
Designate a zone with the brush or smart click and change a material, a colour or a product; add, replace or remove an object. The rest of the image does not move. A decor mood applies to the whole room.
Present through a link
Gather renders, before-and-after comparisons and the mood board in a presentation shared by secure link, with a password and expiry if you wish. The client opens it on any screen, without an account.

What you do from an image

All features

Preparing the image before sending

Resolution and format: PNG rather than JPEG, at least 3,000 px wide when the software allows it. Import: 10 MB and 8,192 px at most. A sharp base gives a sharp render.
Framing: Camera at eye level, straight verticals, no steep top-down angle: proportions stay those the client will see on site.
Readable light: No burnt-out areas and no blocked-up shadows: the render cannot restore what it cannot see. A slightly flat export beats a contrasty render.
A clean image: No dimensions, annotations, grid or visible section plane: everything drawn on the image ends up in the render.

Frequently asked questions

Which software works with RenderThink’s AI rendering?

Any software that exports an image: AutoCAD, Blender, 3ds Max, Rhino, Cedreo, Twinmotion, Archicad, Revit, Winner Design, SketchUp, InSitu, KitchenDraw, and any other. RenderThink reads a PNG, JPG or WebP image, never a proprietary 3D file. SketchUp, InSitu and KitchenDraw also have a plugin that captures the displayed view without an export.

Can I make a 3D render from a simple image or a photo?

Yes, that is the principle. The export of a 3D view or the photo of an already furnished room, even taken with a smartphone, becomes a photorealistic render in about a minute: volumes, proportions and perspective stay those of the image, light and materials are computed. An empty room is not furnished: the render dresses what is already there.

Do I need to install anything or have a graphics card?

No. RenderThink runs in the browser, and the desktop app is optional. The computation runs on our servers: an office laptop is enough, no render engine, no light or material settings to adjust. The only skill expected is exporting an image from your tool.

Does the render respect the exported view?

Yes. The render starts from your image, not from a description: volumes, visible dimensions, layout and perspective are those of the view. The materials visible in the image guide the result; to change one afterwards, brush editing works on the render, zone by zone, without touching the rest.

How long does a render take, and how many credits?

From click to result, the median times measured in production in September 2026 are about 30 seconds in 1K, 40 seconds in 2K, 50 seconds in 4K and just over a minute in 8K, upscale included. Each render, edit or decoration uses credits included in your subscription, more in 4K and 8K; presentations and teamwork use none. The free 14-day trial includes credits.

Are my images used to train the model?

No. Your images are used to produce your renders and nothing else: they are not used to train any AI model. They are stored in Europe, in a space of your own, and you can delete them at any time.