Skip to content

View as MarkdownLLMs: see llms.txt

XR Tour

What's D5 RENDER XR?

D5 RENDER XR combines D5 rendering with 3D Gaussian Splatting (3DGS), turning a set of multi-angle renders into a shareable 3D web experience. Render the scene from multiple viewpoints, upload the images for cloud training, then edit and share the resulting online project.

image

D5 RENDER XR workflow

3DGS, in brief

Rather than building a conventional polygon model, 3DGS represents a scene with many tiny, soft ellipsoids. Each one stores its position, colour, size and transparency; together, they form a detailed 3D view that can be rendered smoothly in the browser. This information can be stored in a .ply or .splat file.

image

Illustration of a 3D object represented by many Gaussian ellipsoids

Before you start — basic 3DGS limitations

  • It cannot reconstruct what was never captured. Areas hidden from every camera angle — such as occluded backsides or interiors — have no reliable data and may be missing or incomplete.

  • Keep the scene static while shooting. Moving people, vehicles, animated objects, water or changing vegetation can create ghosting, blur or artifacts in the result.

  • Reflective, transparent and very thin objects need extra care. They often provide inconsistent visual information, so use more angles and review the preview before rendering.

Why use it?

  • Detailed: retains the look and lighting of the original renders.

  • Fast to view: supports fluid online exploration without conventional modelling.

  • Open workflow: export the source file for use in other 3D tools.

Workflow

Workflow · Step 1 — Create an XR project & shoot with templates

  1. Select TOUR / XR TOUR in the upper-right corner to create an XR TOUR project file. This file stores the shooting-template settings and the file storage location.

image

Create XR project

  1. In the XR project, click Create Template at the top-right of the left panel, then click a shooting position in the scene to initialize the template.

image

Select template

Advanced Tip — Template types & when to use them

1. Hemisphere template — the most widely applicable type, ideal for quickly capturing a comprehensive bird's-eye view of an entire building.

2. Cube template — comprises six faces and two camera orientations. You can also exclude the models contained within certain faces, enabling more detailed capture of specific angles of the building.

3. Sphere template — suitable for 3D objects that need a full 360° display of every detail, such as product photography.

image

Hemisphere, cube and sphere shooting templates shown together

  1. After initializing the template, use the gizmos to adjust its position and angle, use the Parameter Panel to adjust the size, and use Preview to check the final shooting result and fine-tune it. The final render matches the Preview.

image

Adjust template

Initialise templates, editing adjustments and previews.mp4

70MB

DownloadOpen

  1. A single template has limits — for example, it can't capture both a close-up and a distant view of an object at the same time. You can therefore add multiple templates to improve coverage and capture more detail.

image

Multiple camera templates

Multiple camera templates for display.mp4

92MB

DownloadOpen

  1. After editing the templates, tick the ones you want to render in the render window, then confirm the image count and output resolution before clicking Render. Both directly affect the final training quality: more valid images provide denser viewing-angle coverage, while higher-resolution images retain finer material, edge and lighting details.

image

Render window

Initiate Multi-Template Rendering.mp4

3MB

DownloadOpen

  1. When rendering finishes, all local files are stored at the path you specified, and a local HTML file is generated for viewing the result. For enterprises or devices with strict data-security policies that forbid uploads, the entire set of local files can be copied offline for multi-angle presentation of the design.

image

Local files

Local file browsing.mp4

55MB

DownloadOpen

Workflow · Step 2 — Upload to the cloud & wait for training

  1. After rendering, open the Render Tour window, choose Share , and upload the local images to the cloud Personal Projects space.

image

Share & upload

  1. Once uploaded, a task with the status Queued is automatically created in the web backend. Please be patient until training completes.

image

Queued status

  1. When training finishes, the status updates automatically (you'll be notified by email). You can then click Edit to open the online editor, review the training result and share the link.

image

Training done

Workflow · Step 3 — Publish

Edit & preview

When training is complete, click Edit to open the online editor. Use Preview at any time to check the project from a visitor's perspective before publishing.

image

Open the Web Editor

image

Open and edit an XR project

Hotspot editing

The online interactive editor supports two types of hotspots for adding information and interactive content to a 3D scene. Media Hotspots can combine video (via an iframe link), images (up to 20 files, 100 MB each; JPG, JPEG, GIF, PNG, WebP or BMP), hyperlinks and text (up to 1,000 characters). Web Hotspots can open a URL or embedded iframe page. You can customise the hotspot colour, size, name, label position and text size, then use Preview to check how the content opens when a visitor clicks its icon in the scene.

image

Edit a media hotspot

image

A hotspot can also save the current view. In the shared link, clicking it first moves the camera smoothly to the saved viewpoint, then opens the hotspot pop-up or web page.

image

Save current view on a hotspot

Scene management

The interface now introduces the concept of Scenes. You can manage multiple viewpoints directly in the browser — reorder and rename them — and switch between these viewpoints inside the shared link, which makes it easy to build guided, fixed-point presentations.

image

Scene management in editor

image

Switching scenes in the shared link

Scene settings

Use Scene Settings to adjust the project's presentation. The background options include a Skybox rendering mode, allowing you to add a sky background for a more realistic and immersive result.

image

Switch the Skybox background

Comment function

Comments are part of the Web Editor's collaboration toolkit. You can add or view comments on the XR tour's viewing page. The comment function is enabled by default and can be disabled from the list on the right side of the editor. Users must be logged into a D5 account to comment — anonymous visitor comments are not supported.

image

Comment function

Comments support:

  • Text: up to 500 characters, supports @ mentioning team members (within a team only).

  • Image: up to 20 MB, screenshot paste supported.

  • Camera angle: a comment can carry a camera angle; clicking it switches the view to that angle automatically.

Share

When all edits are done, click Share to generate a shareable link that displays your 3D project online. You can communicate with teammates or external collaborators through comments.

Mobile performance optimisation

For XR projects that use a .sog source file, Share Settings include a Mobile Performance Optimisation switch. It is enabled by default and creates a compressed .ssog version so the project can load and render more reliably on mobile hardware. Because this is lossy compression, the mobile version may contain less detail than the original.

  • When sharing: enabling the option starts compression and upload. A progress window is displayed; processing may take several minutes depending on model size and network conditions. Keep the page open until it finishes.

  • After editing: if you crop the SOG model and click Save Changes , the compressed mobile file must be regenerated and uploaded. Re-enabling optimisation also triggers this process when the source model has changed.

  • One link for every device: the share link remains the same. Phones, tablets and VR headsets automatically use the .ssog version, while desktop computers use the original .sog version.

image

Publish a share link

Export PLY File Workflow

Beyond D5's online XR tour service, you can download the .ply source file from the project list and import it into tools such as Cesium, Spline, Webflow, or Google AI Studio for further presentation and development.

Once training is complete, go to your Personal Projects centre, open the relevant project's menu, and select Download PLY to save the source file locally.

image

Download PLY

For detailed, step-by-step integration guides (Cesium, Spline & Webflow, Google AI Studio / Gemini), see the full D5 XR Open Workflow document →

Create an XR project from local files

In MySpace , you can create an XR project from existing 3DGS files, or reconstruct one from photos or video captured locally.

Upload an existing 3DGS file

image

Upload a PLY or SOG file to create an XR project

  1. Select Create → XR Tour → Upload 3DGS file.

  2. Upload one local .ply or .sog file (up to 800 MB), then enter the required project name and upload a cover image.

  3. After the upload succeeds, the XR project is created automatically. Continue with editing, sharing, and publishing as usual; when sharing a PLY file, choose whether to compress it first.

Reconstruct from photos or video

image

Upload photos or video to reconstruct an XR project

  1. Select Create → XR Tour → Create 3DGS from images/video , then enter the project name and optional description.

  2. Upload one MP4 or MOV video (up to 30 minutes; total source material up to 10 GB), or 50–2,000 PNG/JPG images (total up to 10 GB). Capture the same static scene from multiple angles and avoid moving subjects.

  3. Optionally upload camera-pose JSON and an initial sparse PLY exported after local camera solving. These files can shorten reconstruction time; without them, the project first waits for camera solving before entering 3DGS training.

  4. Start reconstruction and wait for the task to finish. The completed XR project supports the same editing, sharing, browsing, and PLY/SOG download workflow as other XR projects.

D5 RENDER XR Best Practices

Below we walk through how to choose the right camera templates and shooting parameters for a medium scene (villa) and a large scene (school). The current XR 3DGS training strategy is geared toward medium and large scenes; very small projects may produce less satisfactory reconstruction results.

Project name

Template count

Image count

Training quality

Online link

Garden villa

5

967

Default

View project

Campus aerial view

6

950 images

Default

View project

Case A · Medium scene — Villa

First we analyse the composition. There is a single main building at the centre, with three areas of "negative space" between the building and the background (surrounding trees). Based on this, five camera templates are created: two hemisphere templates for the main structure, and three cube templates for details within the surrounding spaces.

image

Best practice overview

Sandwich technique for the main structure:

  1. Use a hemisphere template with a larger radius while zooming in the lens focal length in Preview. Adjust the viewpoint so that most of the main structure occupies the centre of the frame. This template captures the overall detail of the main structure.

image

Large-radius hemisphere template for the main building

  1. The previous template lacked context (vegetation), so add another hemisphere template. There's no need to pull the camera too close — capture the outline of the whole scene.

image

Hemisphere outline

Use cube templates for the three negative spaces:

  1. Create one cube template each for the front elevation, rear courtyard, and garage.

image

Cube templates for scene detail

  1. For the principal elevation, keep four faces of the cube template so it fits more flexibly into the spaces between buildings.

image

Cube 4 faces

Case B · Large scene — School

This school project includes the main campus, sports fields, annex buildings, the entrance, and the surrounding landscape. Use a large template for the overall context, then add focused templates for the areas that need clearer architectural and landscape detail.

image

School project zoning

Cover the whole campus first, then refine key areas:

  1. Create a large hemisphere template for the overall project. Include the school, surrounding hills, shoreline, and approach roads to establish the complete site context.

image

Overall project template

  1. Use a box template for the main campus , covering the principal teaching buildings and the larger sports field. This gives the core of the school sufficient, high-density coverage.

image

Main campus template

  1. Add a hemisphere template to supplement the main campus , concentrating on the building massing and landscape around the core campus that the box template does not fully cover.

image

Supplementary main campus template

  1. Create a separate hemisphere template for the small playground so the field, its track, and the adjacent buildings retain clear detail.

image

Small playground template

  1. Use another hemisphere template for the annex teaching buildings. Keep the surrounding trees and paths in frame to preserve their relationship with the main campus.

image

Annex building template

  1. Finish with a box template for the main entrance. Focus on the gate, entrance plaza, and facade so this key arrival view remains sharp in the final XR tour.

image

Main entrance template

Troubleshooting — Why are training results sometimes poor?

If your 3DGS reconstruction turns out blurry or full of artifacts, it usually comes down to how the scene was shot. The three points below are the most common causes and how to avoid them.

  1. Set an appropriate [viewpoint offset value] for your project's dimensions, especially with the cube template. As shown below, without adjusting the offset, cameras in the lower part of the cube end up shooting an upward-facing angle — which contributes little useful detail.

image

Viewpoint offset

  1. Check for large numbers of invalid shots — cameras inside objects or below the horizon. A large quantity of such invalid shots will seriously degrade the training result.

image

Invalid shots

  1. For large projects , use multiple templates to capture different areas separately, then submit them together for training.

image

Large projects

D5 Help Center