Turn your latest Rhino view into a render you can send a client.
Upload the viewport you have open. You get a photoreal image designed to hold your geometry and your camera — and you can change one area of it afterwards. Not the .3dm file, and no render scene to rebuild.
Model screenshotAirenderExample — a model screenshot, not a Rhino export.
Turn your latest Rhino view into a render you can send a client.
Upload the viewport you have open. You get a photoreal image designed to hold your geometry and your camera — and you can change one area of it afterwards. Not the .3dm file, and no render scene to rebuild.
Model screenshotAirenderExample — a model screenshot, not a Rhino export.
Decide up front how much licence the model gets.
Every render runs at one of three fidelity settings, and the setting is what tells the model whether your massing, openings, roof lines and camera are allowed to move. These three are one source rendered at each setting. These are architectural model-view fidelity examples, not Rhino captures — what they evidence is the setting, not the application.
Your modelBalancedBalanced — asks for the same structure to be preserved while materials, lighting, context and realism are enhanced. The studio default.
Real output from the production pipeline — one example per setting, from the same source image. Generation is non-deterministic, so the same settings run again give a different image.
- Massing and proportions
- Window and door openings
- Roof lines
- Perspective and camera
- Materials and finishes
- Lighting
- Surrounding context
- Photographic realism
In the studio this is one choice — Preserve design, which is Balanced above, or More creative, which is Creative. Faithful is the stricter setting and sits under Advanced, on the full Faithful → Creative slider.
When the client asks for one change, you change one thing.
Everything outside your selection stays exactly as it was. For the most predictable result inside it, select a single surface — one wall, one panel, one roof plane — and keep windows, corners and structural edges out of the box. This selection sits on one concrete panel, edge to edge.
The same two files, enlarged. The window frames either side, the soffit reveal with its downlight above and the shadow line at the base all sit outside the selection — if the new panel had drifted, it would no longer line up with them.
Which Rhino view to capture, if you want to revise it afterwards
The selection is a rectangle. On a near-frontal perspective the facade edges are horizontal and vertical, so the box can sit on them and the change ends where the surface ends. That is the view the pair above is made on, and it is the view this works best against.
On a steep axonometric — the view Rhino opens in, and the one a lot of people capture out of habit — every surface runs diagonally, so no rectangle can bound one of them. We have run that case: outside the box every pixel came back from the parent exactly, and inside it the new material stopped at the box rather than at the roof edge. Capture a perspective camera view, which is the one you would send a client anyway, and this stops being a problem.
What is guaranteed is only ever the outside: the finalizer rebuilds every pixel beyond your selection from the version it came from. Inside the box, “change the material” is a sentence added to the instruction — it asks, it does not lock.
Capture, check, render, compare, revise, version, export — in one session.
One session holds the whole design as it evolves — Source v01, the renders and edits made from it, then Source v02 when you re-capture, and the renders made from that. Nothing replaces anything, so the image you showed the client on Tuesday is still there on Friday, and every render says which capture it came from. What you are uploading is a viewport image — never the .3dm — so the final check is always yours: compare the result against the model it came from before you send it.
- 01
Capture the viewport
In Rhino, frame a perspective view and run ViewCaptureToFile — or just screenshot the viewport. JPG, PNG or WebP up to 10MB. ViewCaptureToFile is worth the extra second: it writes the view without the toolbars, at a size you choose.
- 02
Upload it, and read the pre-flight
Arriving from this page, the Studio measures what it can — the pixel count and the shape of the frame — and prints the five things it cannot see: the grid, the gumball, a live selection, the viewport type and the display mode. It never blocks a render. Continue anyway is a real button.
- 03
Render it
Say what the space is, whether to keep the materials you assigned, and how much licence the model gets. Arriving from here, the design-preserving setting is already on. The result opens beside the original with compare already running.
- 04
Check it against your model
Drag the seam and count the things you know: openings, mullions, the curve of the facade, the camera. Or switch to Overlay and fade the two through each other, which is the one that shows you whether something moved.
- 05
Change one thing
Drag a box around the one area a client asked about, describe the change, and every pixel outside your selection is restored from the render it came from.
- 06
Keep the versions
Each run adds a version under the source it came from — Render 01, Edit 01 — labelled with the settings that produced it. Nothing overwrites anything.
- 07
Update the source when the model moves on
Change the design in Rhino, capture the view again, and add it to the same session with Update source. It becomes Source v02 in the rail, the settings from your last render are carried over, and everything you made from the first capture stays exactly where it was.
- 08
Export
Download the full-resolution, watermark-free file on a paid plan, cleared for proposals and published work. Free previews stay in the Studio.
Five decisions, none of them about AI.
There is no sampler, no step count, no seed and no negative prompt. Each control below is described by what it asks the render to do — because with one exception, that is all any of them are: instructions, not locks. The exception is the last one, and it says so.
Geometry — how much the design may move
Preserve design asks for your massing, openings, roof lines, proportions and camera to be held while materials, light and context change. More creative asks only for the composition and silhouette to stay recognisable. A finer Faithful–Balanced–Creative slider sits under Advanced. It is a request written into the prompt, not a lock on your curves.
Camera — left alone unless you say otherwise
Camera defaults to Auto, which adds nothing to the prompt. That is the right default for a Rhino view: you already set that camera, and we should not quietly assert a different one. Eye-level, aerial and street are there when you want them, and they are descriptions — not a camera being placed in your scene.
Materials — whether what you assigned survives
Two buttons above the render button. “Keep what I modelled” asks the render to match the cladding, roofing, glazing, floor and wall surfaces already in your view instead of substituting a palette; “Apply a style” does the opposite, deliberately. A view that arrives already textured opens on Keep, decided from the pixels. An untextured NURBS study has nothing to keep, so it opens on a style.
Lighting — a description, not a light rig
Golden hour, overcast, midday or dusk, and Auto, which is the default and adds nothing. These are words written into the prompt. Nothing here places a sun, a portal or an IES profile in a scene, because there is no scene — there is an image of your view.
One area at a time, after the render exists
Local Edit is the only control that is not a request: you draw a rectangle, describe the change, and every pixel outside that rectangle is rebuilt from the version it came from. Inside the box it is still an instruction. That is the whole revision loop, and it is why a first render is not the end of the session.
Four ways to finish a Rhino view, by the job each one does.
Rhino already comes with a renderer, so that is the first column after ours rather than a footnote. Nothing below says another tool is worse.
The other three read your live model and this one does not — a real disadvantage, and also the reason it needs no install, no seat and no GPU. The three right-hand columns were checked against rhino3d.com, chaos.com and d5render.com on 25 August 2026 and describe only what those pages state about themselves; check their current pages before making a purchase decision on them.
Enough free to render your own view and fix one thing on it.
- Standard · 2K only — HD and Ultra need a paid plan
- Output is watermarked, up to 2K on the long side
- A render, edit, or enhance uses 1 generation. So one preview is deliberately left over for a retry
- Previews are for judging, not delivering — Download and commercial use both start with a paid plan
200 generations / month
$24/mo
or $199 a year, billed annually
500 generations / month
$49/mo
or $429 a year, billed annually
1,200 generations / month
$99/mo
or $899 a year, billed annually
At the annual Studio rate that is $0.06 a generation — watermark-free and cleared for client work, at whichever quality you pick.
See full pricing →Your Rhino views, prompts and renders are never used to train, fine-tune or improve any model.
You keep every right to the view you upload, and paid plans add a worldwide commercial licence for proposals, competitions and published work.
Ask us to remove your uploads and renders — or close the account — and they are gone within 30 days.
Email us within 7 days of your first subscription payment and we refund it in full, no questions.
What Rhino users ask before they upload.
01Do I upload my .3dm file?
No — and there is no plugin, no Food4Rhino install and no Grasshopper component. You upload an image of the view: ViewCaptureToFile, or a screenshot of the viewport, as a JPG, PNG or WebP up to 10MB. Your .3dm never leaves your machine. That is also why the Rhino version does not matter, and why Windows and Mac are the same — you are sending a picture of a view.
02Does it read my NURBS or my Grasshopper definition?
No, and it is worth being blunt about it: nothing on our side sees your surfaces, your control points, your layers, your blocks or your Grasshopper graph. It sees the pixels of the view you captured. Everything on this page about openings and curves being held is about what is visible in that image — not about the geometry behind it. Change the definition and you have to capture the view again.
03Which Rhino display mode should I capture?
A perspective view in Shaded, Rendered or Arctic. Those give the model enough surface and shadow to read. Wireframe and Ghosted give it almost nothing to work from, and a flat Top or Front view is not what this is for. Frame it the way you want the final image framed — the render keeps your camera, so it cannot recover what is outside the frame. Turn off the grid and the gumball, and deselect before you capture; software UI is ignored, but a selection highlight is just yellow paint on your model as far as the image is concerned.
04Will my curves, openings and camera stay the same?
That is what the Preserve design setting is for: it asks the model to hold your massing, window and door openings, roof lines, proportions and camera angle while materials, lighting and context change. Designed to preserve the visible structure. Always compare the result with your original design. It is an instruction, not a lock on your geometry — which is why the compare slider is on by default on the Studio canvas, and why every example on this page is draggable.
05My model already has materials on it. Will it repaint them?
Only if you ask it to. Materials is a two-button choice above the render button, and a view that arrives already textured opens on “Keep what I modelled”, decided from the image rather than from a question we made you answer. Choosing a style is the deliberate opposite, and the panel says so before you spend anything. An untextured NURBS study has no materials to keep, so that one opens on a style.
06The client wants one thing changed. Do I re-render the whole view?
No. Draw a box around the area on the finished render, describe the change, and only that area is repainted — every pixel outside your selection is rebuilt from the render it came from. The result arrives as a new version under the original, which stays exactly as it was. One caveat worth knowing before you frame the shot: the selection is a rectangle, so it bounds a surface cleanly on a near-frontal view and badly on a steep axonometric where every edge runs diagonally.
07What happens when I change the Rhino model?
There is no live link, so you capture the updated view yourself — and then Update source puts that capture into the SAME session rather than starting a new one. It appears as Source v02 in the version rail, above the renders made from it, and the settings from your last render are carried over so you are not re-picking a style you already chose. Everything made from the first capture stays exactly where it was. Two things are deliberately not carried: an edit rectangle, because a box drawn on one image means nothing on another; and any promise that the second render will look like the first. It will run on the same settings — that is what we can honestly offer, and it is not the same claim as reproducing an image.
08Does this replace Rhino Render, V-Ray, Enscape or D5?
Not for what those are best at, and Rhino Render deserves naming first because you already have it: Rhino ships a Cycles raytracer built in that runs as a real-time display mode and renders from your actual model, materials and lights. V-Ray gives you shot-by-shot art direction and renders straight from Grasshopper; Enscape and D5 give you a navigable real-time scene and animation. All three read your model and this does not — a real disadvantage, and also why it needs no install and no seat. What it does that they are slow at is turn a view you already have into something presentable, and revise one area of it in place. Checked against rhino3d.com, chaos.com and d5render.com on 25 August 2026; check their own pages before making a purchase decision on them.
09What uses a generation?
A render, edit, or enhance uses 1 generation. A render that fails on our side is refunded. Free accounts get 3 previews once you sign in, no card — watermarked, up to 2K on the long side at Standard · 2K only, and they stay in the Studio. That is deliberately enough to run one render and one edit on your own project, with one left over for a retry. See full pricing →
010Can I use the renders in client work?
Yes, on a paid plan. Every paid plan includes a worldwide commercial licence to use your renders in proposals, presentations, competition boards and published work, and you keep every right to the view you uploaded. Free previews are watermarked, cannot be downloaded, and carry no commercial licence — they are for judging the result, not delivering it. Read the terms →
Weighing the category rather than the tool? See AI architectural rendering software.
The Rhino view worth judging this on is yours.
Capture the viewport you have open, render it, and check it against your own model — 3 free previews, no card.
Render my Rhino view