How to Map and Simulate an LED Facade on a Building

Turn a 3D building model into a working media facade: import the geometry, map the LED areas, preview the sparse LEDs in the scene view, patch the Art-Net pixel mapping, and feed it from the sequencer. This is the end-to-end route for skyscraper facades, pixel tubes, and architectural LED installations where the LEDs sit far apart and the gaps are part of the look.

Before you start

• The building exported from your 3D application, with texture coordinates that already describe how each piece of content wraps the geometry, modelled in real-world units
• The full pixel grid of each element, including the blank positions created by the physical LED spacing
• The Art-Net universe numbers and controller IP addresses your LED drivers expect
• The wiring order of each LED strand: where it starts and which way it runs
• The physical LED size in millimetres, and the viewing distances the facade is judged from
• A network interface on the server that reaches the LED controllers

Steps

1. Decide the division by content, then export

Divide the building by content, not by geometry. If one video runs around the entire building, keep it as one model — and make sure the exported model already carries the texture coordinates for that wrap, so the content is visualized correctly. If several special facade elements each show different content, export those elements as separate models. Splitting the export is about how the elements are authored, not a limit on placement: one imported model can host several independent content areas through several mappings (step 3), so split when an element needs to be its own object — its own Art-Net output, its own LED settings, its own transform. A model file always imports as exactly one surface object — every mesh inside it is merged and only one material is read — so the files you export decide the surfaces you get. Model in real-world units; the import keeps the file's scale. Geometry that shows content but carries no physical LEDs is not a reason to split a file: you exclude those faces later with LED Faces.

2. Import the models as displays

In the Scene Objects panel, press + (Add scene object) and choose Display ▸ Import model, then pick the file in the Load 3d model dialog. Repeat for each model file you exported. Importing under Display matters: the LED simulation exists only on displays. If you already imported the building as a projection surface, select it and tick Surface As Display in Surface Properties — the object is swapped for a display version and any LED settings survive the conversion.

3. Set up the mappings over the LED areas

Content placement on a facade is flexible, and it is worth understanding before you start clicking. A surface carries as many mappings as you like, and each one covers only the region its gizmo encloses: content lands where that mapping's UVs fall inside 0..1 and the layer is transparent everywhere else, so the rest of the building stays free for other mappings' content. Because every sequencer video channel picks both its surface and its mapping, several channels can play on one building at the same time — a wrap across the whole facade on one mapping, a special element on another — composited by layer order, blend mode, and keying. Keep the two per-area mechanisms apart: mappings place content by region, while LED Faces (step 4) selects which faces carry LEDs.

Select the surface. An imported model arrives with a Custom mapping — the texture coordinates stored in the model file, used exactly as exported.

You are not stuck with what the model brings. Every mapping in the list can be edited, and each row offers set as default (*), edit (E), duplicate (D), and delete (−). While a mapping is open for editing you work with Apply, Cancel, and Reset, so trying something is always reversible; Show Grid draws the mapping's grid on the surface and View Mapping gizmo detaches the gizmo so you can see and move it. Name, Flip U, Flip V, UV Offset Mode (Center or Border), UV Scaling, and UV Offset work on every mapping, imported ones included — so an imported layout can still be flipped, scaled, and nudged onto the content. If the model's coordinates are wrong or missing, or you simply want a different layout, create an analytical mapping and place it yourself, duplicate an existing mapping and adjust the copy, or correct the imported one in place.

To add a content area, pick a type in the Mapping section — Planar, Cubic, Cylindrical, Spherical, Dome, Frustum Projection, or Linear Fisheye to Equirectangular — and press Create. Repeat for each area. LEDs are only generated where the mesh UVs fall inside 0..1, so mapping coverage decides what lights up.

Place an analytical mapping with its gizmo — position, rotation, and scale — plus Mapping Size. Fit snaps it to the surface's bounding box along the axis chosen with the Axis presets (+X, -X, +Y, -Y, +Z, -Z) and is offered for the enclosing types: Planar, Cubic, Cylindrical, Spherical, and Dome. These gizmo-driven controls, together with the type-specific options, are the only ones a Custom mapping does not provide — its layout comes from the model instead. Fully Covering and Repeating are read-only diagnostics that Backstage derives for you, not switches to set.

To reuse a layout authored in your 3D application for a further content area, press Import Mapping and select a model file with the same vertex and index count — only the texture coordinates are taken, and the surface gains another UV set. Add one per content frame; each video channel then picks which mapping it plays into, so different content on the same geometry never forces you to split the model.

4. Switch on the LED simulation in LED Facade mode

With the element selected, tick Enable in LED Simulation (UI) — the section sits below Split & Rasterization — and set Mode to LED Facade. Facade mode draws one dot per LED with the real gaps between them, which is what you need to judge a building; LED Wall mode is for dense panels. Set LED Size to the physical LED size in millimetres. Leave Mask Source on None (all pixels) for now — you point it at the Art-Net mapping in step 8, once that mapping exists.

If a wall shows LEDs on geometry that carries none — edge returns, or the inner shell of a double-sided model — press Pick faces (3D) under LED Faces and click faces in the scene view to toggle them (green carries LEDs). Esc or right-click ends the mode, and the counter below shows how many faces carry LEDs.

5. Add one Art-Net output per element at the full frame resolution

Think of the facade as a complete video frame at full resolution. The mapped faces are one continuous canvas, exactly as if every position on the wall were a pixel. The Art-Net data is then nothing more than the pixels of that frame which are backed by a real LED on the wall — and choosing them is what the DMX pixel mapping does in step 6. The positions in between, the blanks left by the physical LED spacing (elements are usually mounted with gaps along one axis), carry no LED, so their pixels are simply never picked. They still exist in the canvas, and they earn their keep twice over: the full-resolution frame is what the scene view simulates and what the content creator authors against, so the content keeps its true geometry and proportions, and it is what lets you shift content by whole pixels on site.

In the Outputs panel, expand Artnet and press + once per element. Select each new output and set Width and Height in Artnet Output Properties to that full frame: the whole pixel grid of the wall or element, blank positions included. The DMX mapping's X and Y coordinates address exactly this grid, and the LED simulation's Art-Net mask shares the same grid 1:1. Then set Bind to interface to the NIC that reaches the LED controllers.

Because the canvas is the full grid, you can line the content up with the LEDs as they were actually installed by sliding it whole pixels — no re-authoring and no re-cutting the model. Nudge UV Offset on the mapping (with UV Scaling, Flip U, and Flip V), which works on imported mappings too, or change a mapping entry's start X / Y in the DMX mapping so that strand samples a different column or row. For a planar mapping on a plane surface there is additionally a pixel-coordinate placement mode: tick Pixel Values and position the mapping with Total Canvas Size and Sub-Canvas Size in pixels — useful for flat LED elements, though not available on imported facade geometry.

When the whole building is one surface but its walls need separate outputs, route several Art-Net outputs to the same display and use Split & Rasterization → Split (Matrix, or Custom rectangles) to carve the display's pixel canvas into one slot per output.

6. Patch the universes and pixel runs

In the DMX Mapping section, press Add Universe, set its number and the controller IP, then press Show Mapping Window to patch visually. Left-click the first pixel of a strand and drag along the row or column in the direction the strand is physically wired — the selection snaps to that axis — then right-click inside it and choose Add pixel mapping. The start pixel, run length, direction, and next free DMX address are filled in for you. Work strand by strand and add a universe whenever the next strand needs one; 512 channels hold about 170 RGB pixels.

7. Route each display to its Art-Net output

Open Video Routing and check the cell where the element's row meets its Art-Net output column. Routing is what makes the output send DMX at all, and it is also what makes the simulated LEDs sample the composited output content — the exact pipeline the DMX values come from. The status line in LED Simulation (UI) confirms this by reporting Colors: composited output content.

8. Drive the LED positions from the Art-Net mapping

Go back to LED Simulation (UI) and set Mask Source to Art-Net Output. The scene view now simulates exactly the pixels you patched, and it follows the mapping as you keep editing it. Leave the Art-Net Output combo on Auto — it resolves to the output the display is routed to. Reading Auto (not connected) means the routing in step 7 is missing.

9. Play content from the sequencer

Add a Sequence with a Video Channel and select the channel. In its Destination list, tick the facade element and choose which UV mapping the channel plays into. One channel can feed several elements — tick each one and set its mapping. Add a further channel per content area and point it at that area's mapping: several channels can run on the same building at once, and they composite by layer order, blend mode, and keying. Elements outside the sequence's scene group are greyed out. Place your asset as an event and play: the Art-Net output picks up the composited content, the DMX goes out, and the scene view shows the facade running.

10. Verify the wiring before handover

In the DMX Mapping window, right-click a mapped pixel and use Locate to blink a pixel, row, or column on the physical LEDs — the same pixels blink in the 3D scene view, so you can compare model and building side by side. Then tick Live data to colour the grid with the RGB values actually leaving the server. Walk the facade strand by strand this way; a strand patched in the wrong direction shows up immediately.

Options explained

LED Simulation (UI)

Art-Net output and DMX mapping

Common Mistakes

• Assuming a second content area needs a second model file — one surface holds as many mappings as you need. Add a mapping (or an imported UV set) per content area and play each from its own video channel. Re-export as separate models only when the elements must be separate objects, with their own Art-Net output, LED settings, or transform.

• Trying to restrict content to selected faces — content is placed by region, not by face. Size and position the mapping so its 0..1 area covers only the area that should show it; everywhere else that layer is transparent. LED Faces is a different mechanism: it decides which faces carry LEDs.

• The content does not wrap the building correctly — the model's texture coordinates do not match that content. Flip, scale, or offset the imported mapping in place, duplicate it and adjust the copy, create an analytical mapping for that area, add a corrected UV set with Import Mapping, or fix the UVs in your 3D application and re-import.

• No LED Simulation (UI) section in the Inspector — the object is still a projection surface. Tick Surface As Display in Surface Properties, or re-import under Display ▸ Import model.

• Sizing the Art-Net output to the LED count — the output is the full frame, so leaving the blank positions out shrinks the grid and every patched coordinate lands on the wrong LED. Set Width and Height to the complete pixel grid of the wall, then re-patch.

• Re-authoring content because the LEDs sit half a module off — the installed position rarely matches the drawing exactly, and it does not have to. Slide the content whole pixels with UV Offset, or move a strand's start X / Y in the DMX mapping.

• The Art-Net Output combo reads "Auto (not connected)" — the display is not routed to an Art-Net output yet. Route it in Video Routing, or select the output explicitly to keep previewing.

• One wall stays dark while the rest run — no mapping covers that wall inside 0..1, or all its faces are excluded from carrying LEDs. Read the mask, LED, and UV range counters in the status line, extend or add a mapping until it covers the wall, and check the face count with Pick faces (3D).

• LEDs appear inside the building or on edge returns — double-shell models carry inner faces that also receive content. Exclude them with Pick faces (3D), or start from None and click only the outer faces.

• The face selection resets after a model update — the selection belongs to the current mesh, so re-importing or replacing the model clears it. Finish the geometry before spending time on face picking.

Tips

• Share one model with the content team. They need it anyway to see how the content must be built, and Backstage plays into the same coordinates — so the wrap they design is the wrap you get. One mesh can carry several UV sets, one per media frame, added with Import Mapping and chosen per video channel.

• Author against the full frame, sample with the mapping. Keep the content at the wall's full resolution and let the DMX mapping pick the lit positions: the proportions stay true, the scene view shows what the audience will see, and the LED layout can change without touching the content.

• Duplicate before you experiment. Copy a mapping, adjust the copy, and compare the two on the building; Apply, Cancel, and Reset make every edit reversible, and Show Grid puts the mapping's grid on the surface so you can see exactly where it lands.

• Build the facade as layers, not as pieces. Give each content area its own mapping and its own video channel on one surface, then use layer order, blend mode, and keying to combine them — a full-building wrap underneath, highlights and special elements on top, all without touching the geometry.

• Patch by dragging in the direction the strand runs. The selection snaps to the row or column you drag along, and Add pixel mapping then inherits the start pixel, length, direction, and next free address — one drag per strand, no typing.

Related

How to Preview LED Content in the Scene View (LED Simulation) — every LED Simulation option in detail, including LED Wall mode and the colour-source status lines.

How to Set Up ArtNet and sACN Pixel Output — Art-Net and sACN fundamentals, CSV mapping import/export, and the Node Editor encode/decode route.

DMX Mapping Window — full reference for the visual pixel-mapping editor: grid, minimap, navigation, context menu, and Live data.

Route Projectors and Displays to Outputs (Video Routing) — how the routing matrix connects displays to outputs, which is what makes the DMX flow.

Play a Video on a Segmented LED Screen — the dense-wall counterpart: splitting one Display across several LED processors.

Map Content to an Irregular Shape — mapping gizmo and warping techniques for non-rectangular geometry.