SketchUp, Revit and AutoCAD Files for 3D Rendering

Quick answer: Architectural rendering studios work from SketchUp, Revit and AutoCAD files every day, but the three hand over very different things. SketchUp gives you surfaces, Revit gives you parametric building objects, AutoCAD often gives you flat linework with no height at all. What you send decides how the project starts.

This guide is about one service: architectural rendering, the work of turning a design that already exists in a CAD or BIM file into a finished image. It is not about drafting, and it is not about modeling a building from scratch. It is about the handoff, the moment your file leaves your office and lands in a rendering studio, and what happens in the hours after that.

That handoff is where most rendering projects quietly lose time. Not in the lighting, not in the materials, not in the revisions. In the first afternoon, when someone opens the file you sent and finds out that the roof is a placeholder, the window frames are 2D symbols, or the whole model sits ninety thousand feet from the origin because it was georeferenced once and never moved back. None of that is unusual. All of it is fixable. It just needs to be found on day one instead of day four.

What file formats do rendering studios actually need?

A rendering studio needs geometry with real three dimensional volume, materials or at least material intent, and a clear sense of scale. Native application files are ideal because they carry the most information. SKP, RVT, DWG, 3DM and PLN all work directly. Exchange formats such as FBX, OBJ and IFC work too, with some predictable loss.

The practical hierarchy runs like this. Native files preserve everything the authoring software knows: layers, components, families, materials, units, and the relationships between objects. Exchange formats preserve geometry and usually materials, but flatten the intelligence. A Revit wall becomes a set of surfaces. A SketchUp component becomes a mesh. Neither is a problem for rendering itself, because the render engine only cares about surfaces anyway. It becomes a problem when something needs to change, because the studio then has to rebuild rather than edit.

The second thing that matters is units. A model authored in millimeters and exported without a unit declaration can arrive a thousand times too large or too small. Most studios catch this in the first minute, but it is worth stating your units in the email rather than leaving them to be inferred from the geometry.

The third thing is what is missing. Almost every model arrives incomplete in some way, and that is normal. Site context is missing, or the interior is empty, or the landscape is a flat plane. The studio needs to know which absences are intentional and which are oversights, because those get handled completely differently. An intentionally empty interior means the shot is exterior only. An accidentally empty interior means the glass will read as black holes in a dusk render, and someone has to decide whether to model it properly or treat the glazing so it does not matter.

Sending SketchUp files for rendering

SketchUp is the most common file a rendering studio receives from architects and designers in the United States, and it is usually the easiest to work with. The geometry is direct, the components are readable, and the model almost always sits at a sensible scale. It also carries the most predictable set of problems.

The first is reversed faces. SketchUp surfaces have a front and a back, and a face that reads correctly in the viewport can render as a hole or a black patch once real lighting is applied. Fixing reversed faces on a large model is tedious rather than difficult, but it is time that comes out of the schedule. Running a quick check in monochrome style before you send the file, where front faces read white and back faces read blue, catches most of it in a few minutes.

The second is component hygiene. A model where the window is a single component used ninety times is a model where the render team can swap the frame material once and update every window. A model where each window was copied and exploded is ninety separate objects. Both render. Only one of them survives a revision request without a full afternoon of selection work.

The third is imported geometry. SketchUp warehouse assets vary enormously in quality. A sofa downloaded from an open library can carry a hundred thousand polygons and no usable material assignment, and a dozen of those will make a file slow to open and slower to render. Most studios strip them and substitute their own library assets, which is usually the right call, but it changes what the final image looks like compared to what you saw in the viewport. If a specific piece of furniture is actually specified, say so, because otherwise it will be treated as a placeholder.

If you are sending SketchUp, purge unused components before export, save at the version you actually authored in rather than an older one, and include any texture images that are linked rather than embedded. Purging alone routinely cuts file size by half or more, and it removes the abandoned schemes that would otherwise raise questions.

Sending Revit files for rendering

Revit files carry more information than any other common source, and that is both the advantage and the difficulty. A Revit model knows what a wall is, what it is made of, and how it meets the slab. It also carries every consultant discipline, every view, every sheet, and often several linked models, most of which have nothing to do with the image you want.

The usual approach is not to render the Revit file directly but to export a clean subset. A 3D view set up specifically for visualization, with the disciplines you want visible and the ones you do not want hidden, exported to FBX, gives a rendering studio something far more workable than the full central model. It also means you control what is included rather than leaving that judgment to someone who has not been in your project meetings.

Two Revit specific issues come up repeatedly. The first is level of detail. Revit families are often modeled at a schematic level that reads fine in plan and section but falls apart in a close exterior view. Window frames with no depth, handrails as single extrusions, curtain wall mullions with square profiles. None of that is wrong for documentation. All of it is visible in a rendering at eye level, and someone has to decide whether to upgrade the families or compose the shot to avoid them.

The second is coordinates. Revit projects that have been coordinated against a survey often sit at very large distances from the internal origin. Export that directly and you get precision artifacts, geometry that shakes in the viewport, and surfaces that flicker in animation. Exporting relative to the internal project origin rather than shared coordinates avoids the whole category of problem, and it costs nothing to do.

If your project is documented in Revit and you also need the drawings themselves prepared or cleaned up before the visualization stage, that is a separate service. Our architectural drafting services page covers that side of the work, and the two are frequently sequenced together.

Sending AutoCAD files for rendering

AutoCAD is the format where the biggest misunderstanding lives. A DWG can be a fully modeled three dimensional building, or it can be a set of 2D plans and elevations with no third dimension at all. Both are called AutoCAD files. Only one of them can be rendered without building a model first.

This distinction changes the price and the schedule more than any other single factor in a rendering project. If you send a 3D DWG, a studio is preparing an existing model. If you send 2D linework, a studio is building a model from your drawings, which is a different scope of work, usually measured in days rather than hours. Neither is better. They are simply not the same job, and quoting one as the other is how rendering projects go wrong before they start.

If what you have is 2D, that is entirely workable and very common. Plans, elevations, sections and a roof plan are enough to build from, provided they agree with each other. Where they disagree, and they often do, someone has to make a decision. The studio will either ask you or make an assumption. Asking is slower and correct. Assuming is faster and sometimes wrong, and the cost of being wrong lands at the review stage rather than the modeling stage.

Some practical notes for DWG handoff. Bind external references rather than sending them separately, because a file with unresolved xrefs opens empty. Include the layer names you use for walls, glazing and roof, since a rendering team reading someone else's layer scheme cold will spend real time on it. And say which drawing is the current one. A folder with four revisions and no indication of which is live is a guess waiting to happen.

Rhino, ArchiCAD, 3ds Max and everything else

Beyond the three most common applications, rendering studios regularly receive Rhino, ArchiCAD, 3ds Max, Blender, Vectorworks and SolidWorks files. All of them are workable. The considerations are broadly the same, with a few application specific notes worth knowing before you export.

Rhino models tend to arrive as NURBS surfaces, which are mathematically precise but must be converted to meshes for rendering. The conversion settings matter. Too coarse and curved surfaces show faceting along the silhouette. Too fine and the file becomes unmanageable. A studio will usually handle this, but if you are exporting yourself, err toward finer meshing on anything the camera will see up close and coarser on background geometry.

ArchiCAD behaves much like Revit. The building model is intelligent, the export to a neutral format is straightforward, and the same advice about isolating a visualization view applies. IFC is a reasonable exchange path when a native file is not available, though it carries less material information than most people expect it to.

3ds Max files are the only case where you may be handing over a model that is already render ready. If so, say which renderer it was set up for. A scene built for one engine does not carry its lighting and materials cleanly into another, and a studio that opens it expecting one setup and finding another loses the time twice, once discovering the mismatch and once rebuilding.

Blender and Vectorworks both export reliably to FBX and OBJ. SolidWorks and other mechanical formats come up for product and fabrication work, where the geometry is usually excellent and the material assignment is usually absent entirely.

How should you prepare a model before sending it?

Preparation is not about making the model perfect. It is about making it legible. Purge unused elements, isolate the geometry you actually want rendered, confirm the model sits near the origin at real world scale, and write a short note saying what is finished, what is placeholder, and what is missing on purpose.

That note is worth more than any amount of cleanup. A rendering studio opening a file cold has to reverse engineer your intentions from the geometry, and it will get some of them wrong. Three sentences saying that the landscape is indicative, that the north facade is the one that matters, and that the podium material is still under review will save a full revision round.

Beyond the note, a short checklist covers most of what helps. Confirm your units and state them. Remove other buildings, alternate schemes and abandoned iterations from the file, or at least put them on clearly named layers that can be switched off. Include any linked textures. Provide a floor plan or elevation as a reference even when the 3D model is complete, because a flat drawing resolves questions about dimensions that a model does not answer quickly. And send the material palette if one exists, in whatever form it exists, including a photograph of a physical sample board.

None of this needs to be polished. A phone photo of a sample board pinned to a wall is genuinely useful. A finish schedule is better. An empty material field on every surface means the studio is choosing your materials for you, and the first draft will reflect that rather than your intent. If you want to see how the same source information carries through into interior work specifically, our interior rendering services page describes what that stage needs.

What happens when the file is incomplete?

Incomplete files are the normal case, not the exception. A rendering studio expects to fill gaps, and most of that gap filling is routine. The question is not whether the model is complete but whether the incompleteness is understood before work starts rather than discovered halfway through the first draft.

Here is how the common gaps are usually handled. Missing site context gets replaced with generic massing or a photographic backplate, depending on whether the location is real and identifiable. Missing interiors behind glass get either a simple furnished box or a glazing treatment reflective enough that the interior never reads. Missing landscape gets built from library assets appropriate to the region and the season. Missing roof detail on a building that is only ever seen from eye level gets ignored entirely, because nothing above the parapet will appear in the frame.

The gaps that cause real trouble are the ones that affect the composition. A facade material that has not been selected, when the facade is the subject of the image. A site slope that is not modeled, when the building sits on a hillside and the entry sequence is the point of the view. Those cannot be filled by assumption without a real chance of producing an image that has to be redone from the start.

The way through is to identify them at the beginning. A studio reviewing your file before quoting should tell you which absences it can absorb and which ones it needs a decision on. If nobody raises any questions at all about a genuinely incomplete model, that is worth noticing, because the questions do not disappear. They just get asked later, when the answer is more expensive.

How much time does file preparation add to a project?

File preparation is usually a small fraction of a rendering project, but it varies more than any other stage. A clean SketchUp model with organized components can be render ready in under an hour. A 2D DWG set that has to be modeled from scratch is days of work before a single frame is lit.

That spread is why studios ask to see the file before quoting rather than pricing from a description alone. Two projects described identically as a three image exterior package can differ several times over in the preparation stage depending entirely on what arrives. It is also why the honest answer to what a rendering costs always begins with a question about the source file. The variables that move that number are set out on our 3D rendering cost page.

The other timing factor is revision structure. A well organized source model makes revisions cheap, because changing a material or moving a component is a contained edit. A model that arrived as an undifferentiated mesh makes the same revision expensive, because there is nothing discrete to select. If you expect the design to keep moving during the rendering process, and it usually does, the organization of the file you send has a direct effect on what those later changes cost.

The last consideration is sequencing. Sending a model at the exact moment it is finished feels efficient, but it front loads every unresolved decision into the rendering stage. Sending it slightly earlier, with the open items flagged, lets the preparation and the remaining design decisions run in parallel instead of end to end.

Frequently asked questions

Can you render directly from a SketchUp file?

Yes. SKP is one of the most straightforward sources for architectural rendering, and no conversion step is required. The usual preparation is checking for reversed faces, purging unused components, and replacing heavy imported assets with cleaner library equivalents. Most organized SketchUp models are render ready within about an hour of arriving.

Do I need to export my Revit model or can I send the RVT?

Either works, but an exported FBX from a purpose built 3D view is usually better. It lets you control which disciplines and elements are included, avoids sending linked consultant models that nobody needs, and sidesteps the coordinate precision issues that appear when a survey coordinated project gets exported at its shared location.

What if I only have 2D AutoCAD drawings?

That is workable and very common, but it is a different scope than rendering an existing model. The building has to be modeled from your plans, elevations and sections first, which typically adds days rather than hours. It also means any disagreements between drawings need resolving, either by you or by an assumption.

What file size is too large to send?

Size matters far less than organization. A purged and well structured model of several hundred megabytes is easier to work with than a fifty megabyte file full of exploded geometry and duplicate iterations. For transfer, any standard file sharing link works. Email attachment limits are usually the only practical constraint worth planning around.

Do you need my materials and finishes specified in the model?

Not necessarily inside the model, but the information needs to exist somewhere. A finish schedule, a sample board photograph, or a list of product references all work. If no material information is provided at all, the studio selects on your behalf, and the first draft will reflect those choices rather than yours.

Can you work from a model built in software you do not use?

Usually yes, through a neutral exchange format. FBX and OBJ carry geometry and most material assignments from nearly any authoring application. The loss is in editability rather than appearance. Geometry that arrives as a flat mesh renders identically but is harder to modify if the design changes mid project.