What the research told us​

A recurring theme in the research: the old layout had unclear hierarchy. Tools sat side-by-side regardless of how often people used them, important features (like changing geospatial map settings) were buried deep at the bottom of long panels, whereas precious primary screen real estate was dedicated to the infrequently used AI Assistant features.

Our application has also evolved over the years. New features such as our Asset Library (Add Layer panel) and our new Gallery (cloud storage for your screenshots, images, and video generations) have become popular features but were "bolted on" to the existing user interface instead of having a thoughtful integration with the rest of the app.

We focused this release rethinking our UI from the ground up based on your feedback.

New side panels, organized by what you actually do​

Upon reloading 3DStreet, you'll immediately notice our biggest change -- a redesigned panel system. Now we have the space for high-priority application features such as Geospatial and Gallery, which previously didn't have a dedicated home. This also allowed us to greatly simplify the Layers panel, removing extraneous layers that confused most users such as Viewer Mode. Panels also now span the full height of the window and is split into tabs:

• Layers — the scene graph, with a + button at the bottom for adding new layers.
• Gallery — your captured and generated images and videos, moved out of the right-side floating panel.
• Geospatial — geospatial layers (Google 3D Tiles, OSM, terrain flattening) get their own tab. Research found geolocation was one of the most-used features, so it earned a top-level home instead of a custom row inside the scene graph.

App switcher, scene title, and save status sit at the top. Collapsing the panel keeps the app switcher and title visible so you don't lose your place.

We used the same design styling to renovate our right-hand side-panel with:

• Properties — the component inspector for whatever you've selected.
• Console — the natural-language assistant (formerly "AI Assistant"), migrated into a sidebar tab instead of a floating window. The Console now also streams a live history of every editor command as pills you can rewind or replay — see the Console documentation.

Share, auth, and your plan badge sit at the top.

Toolbars and a one-key way to hide everything​

• A new primary toolbar at the top holds the panel-visibility toggle and the snapshot button.
• The action toolbar moves to the bottom of the viewport, and the camera reset and zoom controls move into it from the lower-right corner.
• Press backtick (`) to hide both panels for a near-full-viewport view — useful when you're capturing screenshots or presenting in a community meeting.

What we removed, and what's coming back​

To ship this cleanly we pulled three things that caused confusion and / or were infrequently used:

• Viewer Mode — research showed it "often went in one direction" and lacked the flexibility users wanted for storytelling. Coming back as part of an upcoming camera controls milestone that will also tackle WASD navigation, zoom, and saved camera views — all top requests from the Cincinnati research.
• AI Assistant Report Generator — buried, hard to discover, barely used. We can revisit structured outputs as the assistant evolves.

What's next​

The roadmap from the April update hasn't shifted — the panel redesign was item 5 on that list, and we're now turning to the items the research pushed to the top: camera controls (with Viewer Mode and saved camera views returning here), multi-select, and grouping / nested layers. Custom asset upload, a tutorial flow, and managed-street parity with the legacy Streetmix importer are also in the queue.

Thanks again to Andrea Maravić and the University of Cincinnati team, and to Daria — and to everyone who's shared feedback on Discord. Keep it coming.

Layer Reordering in the Scene Graph​

This has been one of our most-requested features: you can now drag and drop layers above and below managed streets in the scene graph panel. Previously, layers were locked in their creation order, which made it difficult to organize complex scenes. Now you can reorder elements freely to match your intended layout, making it much easier to manage scenes with multiple streets, intersections, and standalone objects.

Next-Generation AI Models​

We've upgraded nearly every AI model powering the 3DStreet image and video generators. If you haven't tried AI rendering in a while, now is a great time -- quality is noticeably better and video generation is dramatically faster.

Image Models​

• Nano Banana Pro from Google is now the default image model
• Seedream from ByteDance upgraded to 4.5
• Added new Flux 2 Max model from Black Forest Labs

Video Models​

• Veo 3.1 Fast is now the default video model, generating results in roughly 90 seconds instead of the previous 5+ minutes which caused timeouts for some users
• Kling 3 Pro replaces Kling 2.5 for video generation
• LTX-2 Fast continues to be a favorite budget option for users

Variable Pricing and Progress Indicators​

We've introduced variable per-model token pricing for image and video rendering. Now you can see exactly what each image and video generation costs before you run it, allowing you to optimize quality, budget or speed.

Estimated progress indicators are now calibrated from three months of real-world API data across nearly 900 predictions, so they are more representative of average wait times, but not an absolute progress indicator.

Faster App Loading​

We've made a significant improvement to how quickly the editor becomes usable after signing in. Previously, the app waited for multiple cloud calls to complete before showing anything. This could take 2-5 seconds or more on slower connections.

Now the editor uses a cached auth state to immediately load while validating and fetching token count in the background. The result is that the editor feels nearly instant on repeat visits.

We've also added a scene loading indicator so you get visual feedback when loading scenes from the cloud, instead of staring at an empty viewport wondering if anything is happening.

Autosave on Snapshot and Share​

Your scene and camera position are now automatically saved when you take a screenshot or share a link. This means the person on the other end always sees your latest changes from the exact camera angle you intended -- no more forgetting to save before sharing.

Expanded StreetPlan Integration​

Thanks An and Mike for contributing more integration upgrades to view StreetPlan streets in 3DStreet:

• Parking vehicles now render correctly in 3D, including angle parking configurations
• Gutter segments have been added to the asset catalog for more accurate curb-to-road transitions
• Improved street labels and segment names for better visual mapping between StreetPlan designs and 3DStreet scenes
• New asset mappings for multi-story buildings, bikes, rail transit vehicles, and pickup trucks

Streetmix Elevation Upgrades and Error Fixes in Managed Street (beta)​

Streetmix recently changed their handling of segment elevation data, switching from integer levels to metric elevation values. This prompted us to clean up elevation across our Managed Street (beta) street segments. 3DStreet Managed Street (beta) uses an integer "level" system (where each level represents 0.15 meters, roughly 6 inches — essentially the height of a standard curb) rather than a specific height value. We've added conversion logic to maintain compatibility with Streetmix's new schema, and this opens the door to more precise segment heights in 3DStreet in the future. We also fixed some elevation calculation bugs that could cause visual artifacts on segment ground surfaces — thank you to users who reported this, it makes our hearts warm to receive bug reports that prove people use this product.

Improved 3D Gaussian splat Support​

For those working with 3D-scanned environments, we've upgraded our Gaussian splat support:

• Upgraded to the official Spark v2.0.0 preview library, which enables...
• Streaming .rad file support -- splat files can now be loaded via URL parameter, enabling faster loading of large scanned environments

Gaussian splat support is still highly experimental, use with caution or contact us for support.

Quality of Life​

A handful of smaller improvements:

• Exported .3dstreet.json files now use your scene title as the filename
• 4x AI render now shows more statuses including an initial "sending" status while the API call is in progress, so you know it's working when using slower connections or high CPU load
• Menu fade delay removed for snappier UI interactions
• AI function call tracking for better feedback analysis when you rate AI responses. Many of you are rating the LLM responses, thank you.

What's Next​

We are working on the following milestones in Q2 and Q3, roughly in this order.

1. Custom Assets (gltf / glb upload)
2. Camera Controls + Zoom Fixes
3. Managed Street Parity w/Legacy Streetmix / Retire Legacy Importer
4. Tutorial
5. Panel Design Fixes
6. Map Controls Design Fix
7. Grouping
8. Draw on Street

Join our Discord (link below) to share your feedback.

AI Generator

The new AI Generator app is a "generative studio" style interface to access our growing set of AI image and video rendering tools in one streamlined app specialized for urban planning and spatial design. Access AI Generator from the new App Switcher menu by clicking the 3DStreet logo in the upper-left hand corner.

AI Generator for Video​

You can now generate short AI-rendered videos of your street designs in the Video tab of the AI Generator app. This brings your scenes to life in a way that static images can't -- imagine showing a stakeholder a flythrough of a proposed protected bike lane or a reimagined neighborhood main street in motion.

Video generation requires a source image to use as the first frame. Upload your own image or open an existing image from your gallery and click "Create Video" to use as an input source. Rendering AI Video assets uses your monthly AI generation tokens, just like image rendering. Quality, rendering time and token usage varies by chosen AI model.

AI Generator for Images: Modify and Create​

In addition to video, access all our image editing tools in one place with the new 3DStreet AI Generator app. You can now:

• Modify from Input Image -- Upload your own image or select an image from your gallery as an input, then re-render the image while applying stylistic, editorial, geometrical, rendering, or other modifications using state of the art frontier image models from Google and other major providers.
• Create from Prompt -- Generate a new image from a text prompt without a source image.
• Edit and iterate -- Use the gallery to experiment with different styles and prompts on the same scene or input file to learn how they affect the final output

Cloud Gallery with Cross-Device Sync​

All of your rendered images and videos are now stored in your 3DStreet Cloud account, replacing the previous browser-only storage. This means:

• Cross-device access - View your gallery from any device where you're signed in including desktop, laptop, and mobile devices
• Generation metadata - Each gallery item includes generation details such as model and text prompt used to quickly replicate and modify proven workflows
• Scene links - Gallery items generated from 3DStreet Editor now link back to the 3DStreet scene it was generated from for rapid editing

This was a major behind-the-scenes refactor, and it makes the gallery a much more reliable and useful part of the workflow.

Building Variants on Managed Streets​

The street segment sidebar now includes selectable building variants, giving you more control over the look and feel of the buildings along your street. Buildings now use a "fit" placement mode instead of random, so the results are more predictable and intentional. Combined with new building assets added to the model catalog, you have a wider palette for creating realistic streetscapes.

New building assets created with support from StreetPlan.

New 3D Assets: Plants and Cars​

We've added a collection of new models to the asset library:

• New tree and median planter models
• New car models
• New Plants and Fixtures categories in the asset panel for easier browsing

Also new, now you can see 3D model metadata in 3DStreet for some assets: click to select a built-in 3D model placed in your scene, then use the right-hand properties panel to scroll down Model Info. Assets in 3DStreet have varying origins including original commissioned art, existing assets from sources like Sketchfab, and partnerships.

Expanded StreetPlan Integration​

If you use StreetPlan.net for 2D design and want to visualize your work in 3D, we've made significant improvements to how 3DStreet imports and renders designs:

• Expanded object mapping -- Many more StreetPlan object types now map correctly to 3DStreet equivalents
• Fixed orientation/side-flipping -- Elements now render on the correct side of the street
• Improved lane type support -- Better handling of different lane types during import
• Improved vehicle placement -- Vehicles are now positioned more accurately on street segments
• Improved building support -- StreetPlan import now supports more building variants to better match intended styles

Thanks to Mike and the StreetPlan team for their support and help to better connect these systems.

Drag and Drop 3D Models​

You can now drag and drop glTF/GLB files directly into the 3DStreet viewport to import custom 3D models. No need to go through menus -- just drop your file and place it in the scene. Here's a quick video showing what you can make with this feature.

Streetmix Schema 33 with Elevation​

If you use Streetmix.net for your street cross-section design, 3DStreet now supports import Streetmix schema version 33 with elevation values in meters which are then converted into a fixed set of curb height levels for 3DStreet.

In the future we should consider using segment height in meters instead of arbitrary curb height levels, but for now Managed Street segments have user-definable curb height levels at increments of 0.15m (approx 6 inches).

Bollard Buddy: AR Street Safety on Web and iOS​

We've launched Bollard Buddy, a new AR app for placing street safety objects -- bollards, traffic cones, planters, and more -- on real-world surfaces using your device's camera. It's available in two forms:

• Bollard Buddy Web -- Built into the 3DStreet platform as part of our app suite. Access it from the app switcher in the upper-left hand corner to quickly place safety objects in AR and capture photos. This is a demo app with limited functionality using the legacy 8th Wall WebAR SLAM engine.
• Bollard Buddy iOS -- A native iOS app built with SwiftUI and RealityKit for a smooth, device-optimized AR experience. Features include tap-to-place objects, drag and rotate gestures, a measurement line tool, and photo capture that saves to your cloud gallery for further image processing and scene editing. (TestFlight beta available for 3DStreet Pro customers.)

Both versions share the same 3DStreet Cloud gallery, so photos you capture in the iOS app show up when you sign in on the web, and vice versa. Bollard Buddy is a fast, focused way to visualize street safety improvements in context -- snap a photo of an unprotected bike lane, drop in some bollards, and share the before-and-after.

Apple Sign-In​

You can now sign in to 3DStreet with your Apple ID, joining our existing Google and Microsoft sign-in options. If you use multiple providers, your profile will show all of your connected accounts. This is part of our work on the new Bollard Buddy iOS App.

Terrain Flattening for Geospatial Scenes​

When working with Google 3D Tiles in a geolocated scene, existing real-world elements like trees, utility poles, and parked cars can get in the way of your design. The new Terrain Flattening feature lets you flatten a rectangular area of the 3D map tiles, clearing space so your street design takes center stage.

• Enable flattening from the Geospatial Layers panel and adjust the flattening shape to cover your project area
• Remove distracting map elements like trees, cars, and utility poles that overlap with your design
• Position the flattening shape below ground level to design subterranean treatments like underpasses or sunken plazas

This is especially useful for presentations where you want the focus on your proposed design rather than the existing conditions.

GeoJSON Import for Building Envelopes​

3DStreet already supports 2.5D building geometry from OpenStreetMap, rendering extruded building footprints based on OSM data for any geolocated scene. We've now expanded this with a general-purpose GeoJSON loader that can render building envelopes from any GeoJSON source, not just OSM.

This opens up new workflows for importing building envelopes from tools like Cityscaper, a Python-based cityscape forecasting tool that simulates future building development based on zoning scenarios. We recently contributed to Cityscaper to add GeoJSON export with direct 3DStreet integration. Now you can run a zoning simulation, generate a GeoJSON file, and open it directly in 3DStreet from the command line.

Gaussian Splat Rendering with Spark​

We've updated out Gaussian splat visualization library to use the new Spark 3D Gaussian Splatting renderer for THREE.js improving our support for 3D-scanned environments. If you're working with photogrammetry or 3D capture data, splats now render more efficiently in your scenes.

What's Next​

That's a lot of feature talk. To put it all in perspective, we're excited to host a webinar with power user John Boyle from the Greater Philadelphia Bicycle Coalition Thursday February 26 at 1pm Eastern / 10am Pacific.

John will share how he leverages realistic street improvement visualizations to drive change in local communities. Click here for the registration link.

The Challenge​

Communities struggle to imagine proposed street changes, leading to project delays, pushback, and eroded trust between residents and agencies. The UC team, led by Alejandro Lozano Robledo with researchers Mythili Emani, Domagoj Bui, and Andrea Maravic, asked two questions:

1. Is accessible AR technically feasible? Can we deliver on-site experiences using just smartphones, no apps required?
2. Does it actually improve engagement? What happens when communities can walk through future streets before they're built?

After testing at Cincinnati's Short Vine and University Avenue intersection, both answers came back strongly positive.

Technical Analysis: Evaluating AR Platforms​

With numerous AR technologies available, it's difficult to determine which is best suited for community engagement without thorough evaluation. Quick trials don't reveal the nuances of reliability, accessibility, and user experience that emerge in real-world deployment. The UC team conducted a comprehensive analysis of five platforms across 13 test dimensions: Adobe Aero, InCitu, WebAR via Onirix, WebXR via 3DStreet, and 8th Wall. Test dimensions included geo-anchor support, authoring app usability, ability to reproduce model accurately including scale, and more as detailed in the full published report.

Consumer Reports-style comparison of AR platforms evaluated for the UC pilot, excerpt from A Window into Our Future: an Augmented On-Site Experience

Top Platform Choice: 8th Wall​

8th Wall emerged as the top choice for this use case, though not by a landslide:

• No app required for viewing: 8th Wall was the most reliable experience working in mobile browsers on many devices without need for additional apps or specific API support.
• Visual Positioning System (VPS): More reliable than GPS-only anchoring

However the decision wasn't straightforward. Adobe Aero and its AR distribution platform tested even better than 8th Wall, but was recently discontinued by Adobe. The WebXR Device API is a standard supported by a who's who coalition of device and browser vendors, but the research team found inconsistent support even for devices which professed WebXR compatibility.

Honorable Mention: InCitu​

InCitu is a professional-grade platform that leverages native SLAM (Simultaneous Localization and Mapping) features built into modern smartphones. While more expensive than other offerings, it may be particularly appropriate for professional architecture and planning use cases where precision and reliability justify the investment. Learn more about InCitu's capabilities in our joint webinar on AR for street design.

Dishonorable Mention: WebXR Standard​

Two platforms use a web API called "WebXR" -- an open API for augmented and virtual reality on the web: Onirix and 3DStreet's built-in AR Viewer Mode functionality. The research team found insufficient support across devices used by target communities. Especially notable was lack of native support in some models of Samsung and iPhone devices. 3DStreet offers support for iOS devices with a third-party AppClip solution Variant Launch.

Editor's note: As a long-time proponent of the WebXR standard, it pains me to admit that while it's a great solution when it works through its access to native-level SLAM APIs, device and browser support is simply not sufficient enough to be used in consumer applications for the general public to be accessible on a wide range of devices.

Storytelling Innovation: Personas Over Data​

The research team experimented with visual presentation design in addition to testing AR platform technology.

Instead of floating statistics, the team created diverse character avatars -- ie. local business owners, elderly residents, families -- each explaining how changes affect their lives. Speech bubbles combined personal stories with real data: "These streetscape improvements didn't just look better—crime dropped 40%." (Example for experimentation of display and communications, not actual data from any specific site.)

As Luis Alonso from MIT noted, personas could show current problems, then reappear in the redesigned street celebrating improvements—creating emotional narrative arcs that abstract data cannot.

"When I see the person and their speech bubble, I can't stop myself from reading," noted another researcher. "Unlike floating data visualizations, which I ignore."

The User Experience​

Participants scan a QR code (paper prototype, eventually stickers or spray-painted markers), explore the AR scene for 2-3 minutes, walk to different corners to reveal layered information, then complete a survey. Directional markers guide users to optimal viewing positions.

The team documented everything with 360° cameras and the captured content and immersive AR experience can also be leveraged as immersive training materials for future facilitators.

Why This Research Matters​

This work helps to validate that accessible AR can meaningfully improve transportation planning engagement without expensive equipment, proprietary software, or specialized expertise. The framework shifts conversations from loss aversion ("you're taking my parking!") to collaborative opportunity ("what should we gain here?").

For 3DStreet users, the workflow is proven: Create scenes in 3DStreet, export as "AR Ready," deploy through 8th Wall or InCitu, then iterate based on community feedback.

As cities face mounting challenges around climate, housing, and equity, democratizing urban design is an opportunity to make positive change quickly.

Based on research by University of Cincinnati (Alejandro Lozano Robledo, Mythili Emani, Domagoj Bui, Andrea Maravic), MIT City Science Lab (Luis Alonso), and 3DStreet (Kieran Farr). Read more at UC Digital Futures and MIT Media Lab.

Today we made it happen -- for just $5,000 per year, your organization gets Pro access for up to 50 users with a straightforward activation process.

Why Pro Team?​

Over the past few years, I've worked with professionals and educators across a wide range of industries and locations who wanted to use 3DStreet with their teams but found individual subscriptions cumbersome. You told us you needed:

• Simple activation - No lengthy procurement processes or complex contracts
• Domain-based access - Automatic access for anyone with your organization's email domain
• Predictable pricing - One flat rate regardless of how many people use it (up to 50 users)
• Self-service - The ability to activate immediately without waiting for sales calls

Pro Team delivers on all of these.

What's Included​

Pro Team gives your entire organization (up to 50 users) access to all Pro features:

All Pro Features for Your Team​

• Unlimited Geospatial Access - Set locations anywhere in the world with Google 3D Tiles and satellite imagery
• AI Generation Tokens - Each user receives 100 monthly AI generation tokens for photorealistic renderings
• GLB Import/Export - Bring in custom 3D models from Blender and export scenes for professional applications
• iOS and Android AR - Share augmented reality scenes on any device
• Clean Exports - Download snapshots without watermarks for professional presentations
• Custom AI Prompts - Precise control over rendering style and atmosphere

Hassle-Free Administration​

• Domain-Based Access - Anyone with your organization or university email domain gets automatic Pro access
• One-Year License - Simple annual licensing with no auto-renewal surprises
• Setup Support - We'll help get your team onboarded within 2 weeks
• Flexible Payment - Pay by credit card for instant activation, or request an invoice

Perfect for Universities​

If you're a university faculty member or administrator, Pro Team is designed with you in mind. For many institutions, $5,000 is easier to approve than individual student subscriptions, and the domain-based access means students simply sign in with their university email to get Pro features.

Educators: Your students can explore geospatial features, create AI-powered visualizations, and design safer streets with professional tools. The one-year license gives you a full academic year (or more) to integrate 3DStreet into your curriculum.

For K-12 educators: We have a separate classroom sponsorship program specifically designed for primary and secondary schools with even more accessible options.

How to Get Started​

Getting started with Pro Team couldn't be simpler:

Option 1: Instant Activation with Credit Card​

1. Visit our Pro Team purchase page
2. Complete the payment with your credit card
3. Your organization's domain is activated within 2 business days
4. Team members sign in with their organization email to access Pro features

Option 2: Invoice Payment​

1. Schedule a brief demo call where we'll discuss your needs
2. We'll send you an invoice for $5,000
3. Once payment is received, we activate your domain
4. Your team gets Pro access immediately

That's it. No lengthy contracts, no complex procurement processes, no waiting months for approval.

The Fine Print​

We support Google and Microsoft authentication systems "out of the box." Other authentication systems may require setup time or additional fees. 3DStreet user content is public by default for all users. Additional privacy options may be available for a fee. AI generation tokens top-up at the beginning of each month for each user.

Ready to Get Your Team Started?​

Whether you're a university wanting to bring cutting-edge visualization tools into the classroom, or a professional team looking to generate rapidly prototyped visuals, Pro Team makes it simple to get started.

Purchase Pro Team Now (instant activation with credit card)

Schedule a Demo Call (for invoice payment or questions)

View All Pricing Options (compare with individual Pro plans)

Questions?​

Visit our pricing page or join our Discord community where we're happy to answer questions about Pro Team and help you determine if it's the right fit for your organization.

To that end, I'm happy to announce that as of September 1, 2025 we're now offering more in our 3DStreet Pro plan, adding new cloud AI image rendering which incorporates 4 frontier image generation models and AR viewing for Android and iOS devices. Pro users now have these new cutting edge features at no additional cost.

100 AI Generation Tokens Every Month​

Every Pro subscriber now receives 100 AI generation tokens monthly to create photorealistic renderings of their street designs. These tokens unlock our powerful AI rendering pipeline that transforms your 3D scenes into compelling visualizations.

With AI generation, you can:

• Create photorealistic images for presentations and stakeholder meetings
• Generate multiple style variations of the same scene
• Produce images that better convey the feeling and atmosphere of your proposed designs
• Export high-resolution images without watermarks

Choose from 4 leading image-to-image generation models:

• Flux Kontext Pro by Black Forest Labs
• Seedream 4.0 by ByteDance
• Nano Banana by Google Gemini 2.5 Flash Image
• Real Earth Kontext Pro LoRA Model by Ismail Seleit

Pro Annual subscribers get an extra bonus: 840 AI generation tokens on first purchase – that's 7 months worth of tokens upfront to jumpstart your creative projects.

Custom Prompts for AI Rendering​

Pro users can now customize AI generation prompts, giving you precise control over the style and mood of your rendered scenes. Want a sunny morning atmosphere? A rainy evening scene? Dinosaurs and robots with lasers roaming the streets? Simply describe your vision in the prompt field.

Import and Export GLB and Image Files​

Professional workflows demand flexibility, and Pro users can now import and export GLB files -- the industry standard for 3D content -- and import images to create truly custom simulation environments. This opens up powerful new workflows:

Import Custom 3D Models​

• Bring in custom-designed street furniture from Blender or other 3D tools
• Add unique local elements that make your designs more authentic
• Import AI-generated 3D models

Import Image Models​

Pro users can now work with various image formats to enhance their street designs:

• Example of local style signs for speed limits and town welcome signs for Kiritimati Island
• Import equirectangular 360 images for use in geolocated AR visualization
• Use an existing design or plan drawing as an image to use as a reference for designing in 3DStreet

Export for Professional Applications​

• Export your complete scenes as GLB for use in presentations
• Integrate with professional visualization pipelines
• Export "AR Ready" formats with simplified geometry made specifically for third-party Augmented Reality apps

Enhanced AR Support for iOS Devices​

Pro users can now share AR experiences on both iOS and Android devices, dramatically expanding your ability to showcase designs in real-world contexts.

iOS AR Mode (Pro Feature):​

• iOS AR support via Variant Launch
• No app installation required – works directly in Safari
• Share AR scenes via QR codes or direct links

WebXR AR Mode (free feature):​

• Works on Android devices and AR headsets
• More interactive features and controls
• Real-time scene adjustments in AR

Unlimited Geospatial Tokens​

While free users get limited trial geospatial tokens, Pro users enjoy unlimited access to all geospatial features:

• Set scene locations anywhere in the world
• Integration with Google 3D Tiles and satellite imagery
• Generate automatic location descriptions

Clean, Professional Exports​

Pro allows users to download JPEG snapshots without the "Free Community Edition" watermarks, perfect for:

• Professional presentations and reports
• Grant applications and funding proposals
• Community engagement materials
• Social media and marketing content

Pro Annual: Best Value for Professionals​

For professionals committed to transforming street design, Pro Annual offers the best value:

• Save 30% compared to monthly billing
• 840 bonus AI generation tokens on first purchase
• All Pro features for a full year
• Save with Pro Annual

Ready to Go Pro?​

Whether you're creating compelling visualizations for community meetings, designing safer streets, or exploring creative possibilities, 3DStreet Pro gives you the tools to bring your vision to life.

Join our growing community of Pro users who are reimagining streets around the world. Have questions or need help getting started? Visit our Discord community where Pro users share tips, showcase their work, and help each other create amazing street designs.

Already a Pro user? Make sure you're taking full advantage of these features in the 3DStreet app.

3 Free Location Lookups for Everyone​

Every 3DStreet user now starts with 3 free location lookups that unlock geospatial capabilities including:

• Location lookups - Get street addresses, locality, state, and country information
• Elevation data - Access terrain height information for your scenes
• Location description - Create automatic location descriptions including nearby intersections to automatically identify locations of your street projects

These free lookups give you the power to create location-accurate scenes and experiment with geospatial features before deciding if you need the unlimited access that comes with our Pro plan.

Free Intersection Creation​

One of our most requested features - street intersections - is now available for all users! Previously limited to Pro subscribers, you can now:

• Create 4-way 90° intersections with custom sizing for each direction
• Add crosswalks, sidewalks, and traffic control devices
• Build T-intersections and dead-ends
• Experiment with intersection design without any payment requirements

While this current implementation focuses on 90-degree intersections, it provides a solid foundation for exploring intersection design. As one of our most requested features, we've made it free to give everyone access to experiment with these critical street components while we continue developing more advanced intersection capabilities.

Simplified Access - No More 30-Day Trial​

We've also streamlined our onboarding by removing the 30-day trial requirement. With these new free features available upfront, you only need to provide payment information when you're ready to unlock unlimited geospatial access and advanced Pro features.

This change reflects our commitment to making 3DStreet accessible while giving you the freedom to explore what's possible before making any commitment.

Try It Out!​

Ready to explore these new free features? Click here or the "Launch App" button in the upper-right hand corner of this blog to start creating with your 3 free location lookups and intersection capabilities, or join our Discord community to share what you build and provide feedback on what you'd like to see next.

These updates are part of our ongoing mission to democratize street design tools and empower more people to reimagine their neighborhoods. We can't wait to see what you create!

Ever wondered what your neighborhood looked like 100 years ago? With 3DStreet's enhanced AR capabilities and AI tools like World Labs, you can now create immersive augmented reality experiences that let people walk through history. I recently used this workflow to create AR scenes of historic San Francisco for a Green Density walking tour, allowing participants to toggle between past and present views of the same location.

What You'll Create​

This tutorial will walk you through the complete process I used to create AR experiences for my Green Density walking tour, where participants could scan QR codes and see how San Francisco's landscape has evolved over time.

By the end of this tutorial, you'll have:

• Transformed historic photos into 360° panoramic scenes using World Labs
• Created geospatially-positioned AR scenes in 3DStreet
• Generated QR codes for easy sharing with tour participants
• Built an engaging community experience that brings history to life

Step 1: Find and Prepare Your Historic Photo​

Start by finding a historic photograph of your location. In this case I used a photo from early 1900s that I found on a local San Francisco history website.

Sometimes, your historic photo may need improvement. Here's another photo of the same area 50 years earlier in 1882. The visual quality is much poorer and we'll need to do some work to improve it.

Enhancing Source Image Quality (Optional)​

These days there's more than just Photoshop with new AI-powered options to restore source images.

Option 1: Flux Kontext Pro​

As covered in a recent blog post tutorial, I strongly recommend the Black Forest Labs Flux Kontext Pro model for AI powered image editing that retains original geometric dimensions. This is perfect for a case like this where we want to retain topography and structure geometry but reshade nearly everything.

Option 2: Flux Inpainting (Fill)​

• Use tools like the Fill or Inpainting function in Black Forest Labs playground to edit specific areas
• Great for replacing empty areas with historically accurate details
• I used this to better illustrate the historic Precita Creek area

Option 3: OpenAI GPT Image Model​

ChatGPT has image-to-image editing features that create high quality outputs, but the model does not retain accurate geometric information from the source image. This is fine for artistic interpretation -- while it has high quality output, it comes at the expense of frequent hallucination

My Enhanced Precita Creek (Flat) Image​

Step 2: Create a 360° Scene with World Labs​

Once you have your enhanced historic photo, it's time to create a full panoramic environment.

1. Import to World Labs: Upload your modified photo to World Labs (early access required)
2. Generate 3D Scene: The AI will expand your 2D image into a full 360° environment
3. Export Panoramic Image: Download the resulting equirectangular (panoramic) image

This is the output after importing the 2D image and generating the 3D scene using World Labs, then exported as a panoramic 360º equirectangular projected image.

Panoramic Image Enhancement for Topography Matching​

The newly generated panoramic image won't be perfect -- the AI will not be able to match actual topography for the outpainted areas. Luckily, World Labs has a tool for panoramic inpainting that makes it much easier to target specific areas of the horizon to modify the topography to match the actual land area in the location of your historic photo.

This panoramic image has been modified with World Labs inpainting to add hills in the rough location of Twin Peaks as well as the distant bay and a small island representing Yerba Buena.

Step 3: Host Your Panoramic Image​

You'll need your panoramic image(s) to be publicly accessible via URL:

1. Upload to GitHub Pages: Create a new repository and upload your image
2. Alternative Hosting: Use any publicly accessible web hosting service
3. Get Direct URL: Copy the direct link to your image file

I created a GitHub repository for these AR source and output materials.

Step 4: Create Your 3DStreet AR Scene​

Now comes the exciting part - building your AR scene in 3DStreet.

You can also Clone an existing 3DStreet scene that already has the below configuration and simply replace the location and image Src values. Here is a known-working panoramic AR 3DStreet project that you can Save as... to modify for your own project.

Setting Up the Basic Scene​

1. Create New Scene: Start a new project in 3DStreet
2. Set Geospatial Location:
   • Click the location icon and enter the real-world coordinates
   • This anchors your AR experience to the physical location

Adding the Panoramic Skybox​

1. Add Custom Geometry:

   • Start by adding a circle geometry
   • Click "Advanced" and change geometry type to "sphere"
   • Change the Radius value to 150

2. Apply Your Historic Image:

   • With this custom geometry selected, in "Advanced" section
   • Open the Materials tab
   • Select the Src value and replace with the URL of your panoramic image
   • Press enter
   • Change Shader to flat
   • Change Side to back
   • Scroll back up to the common controls and change Scale x value to -1

3. Adjust Orientation:

   • Fine-tune the Y rotation to align the view correctly
   • Use the red axis of orientation indicator in the center of your scene to identify North

Configure AR Viewer Mode​

1. Select Viewer Mode: Click Viewer layer and fine Mode in the right sidepanel
2. Enable WebXR-AR: Set the viewer Mode to webxr-ar
3. iOS Support: Enable iOS support (requires 3DStreet Pro plan)
4. Generate QR Code: Click to generate and download the QR code (You must save or Save as... your scene to generate a QR code)

Step 5: Test Your AR Experience​

Before sharing with others, thoroughly test your AR scene:

1. Visit the Location: Go to the physical location where your scene is anchored
2. Start Recording: Use your device's built-in screen recording
3. Scan QR Code: Use your phone's camera to scan the code
4. Launch AR: Follow the prompts to open the AR experience
5. Orient Yourself: Face the correct direction (typically west for my examples)
6. Activate Scene: Click the play button to begin the AR experience
7. Toggle Visibility: Use the eye icon to fade the historic scene in and out

Step 6: Prepare for Community Engagement​

Print Materials​

• QR Codes: Print multiple copies of your QR codes
• Map Integration: Include codes on tour maps or information sheets
• Instructions: Provide simple step-by-step instructions for participants

Day-of-Event Tips​

• Multiple Copies: Bring plenty of printed QR codes
• Backup Device: Have a dedicated device ready to demonstrate
• Quick Demo: Practice a 20-second demo for hesitant participants
• Troubleshooting: Be prepared to help with QR code scanning issues

Real-World Results​

During my Green Density walking tour, I tested three historic locations:

1. Cesar Chavez (1931) Showing early rails and paved surfaces over buried Precita Creek
2. 26th and Folsom (late 1800s): Revealing the original Precita Creek before it was buried
3. Dolores Park: Displaying the original marshland and watershed view toward the East Bay

The experience was powerful - participants could literally see how San Francisco's landscape and housing density have evolved over time. Most people needed less than 20 seconds to see the full experience, making it perfect to illustrate a stop on a walking tour.

Try it Out!​

Here is a copy of the map that we used that included QR code mapped to the specific locations of the source images.

Click the map to enlarge. You can print it out and go to the physical locations if you're in San Francisco, or just scan the QR codes to see what the views look like. After scanning the QR and loading the page, when you press the play button the default orientation is assumed to be facing West.

Device Compatibility​

• Android: Works with modern devices using Chrome or Edge browsers that support WebXR
• iOS: Available through enhanced compatibility features for Pro subscribers, using Variant Launch as a polyfill for iOS.

Tips for Success​

• Start Simple: Begin with one location before scaling to multiple sites
• Test Early: Visit the location and test the AR experience before your event
• Backup Plans: Have printed photos as fallbacks if technology fails
• Engage Participants: Encourage people to share their reactions and discoveries

Wrapping Up​

Creating AR experiences with historic imagery opens up incredible opportunities for community engagement and education. By combining AI-enhanced historic photos, World Labs' 360° generation, and 3DStreet's AR viewer mode, you can create experiences that truly bring the past to life.

This workflow bridges the gap between historical preservation and modern technology, creating memorable experiences that help people understand how their neighborhoods have changed over time. Whether you're planning a community walking tour, creating educational content, or simply exploring local history, these tools make it easier than ever to share the stories embedded in our urban landscapes.

Ready to create your own historic AR experience? Start by finding a compelling photo from your area's past, and let technology help you bring that history into the present.

Did you know that you can realistic rendering and animations of your street scenes using free or low cost AI tools? Transform your basic 3D street designs into stunning photorealistic images and dynamic animations that will captivate stakeholders and bring your vision to life with this step-by-step guide.

In this guide, we'll walk through the complete workflow to take a 3DStreet scene from simple 3D rendering to more realistic images and animations using Black Forest Labs (BFL) and Google's Veo3 Models.

Step 1: Create Your 3D Street Scene​

First, let's create a 3D street scene like normal in 3DStreet. Whether you're starting from scratch or using a user-generated scene, make sure your layout captures the key elements of your street design vision.

Once you're satisfied with your scene composition, it's time to export it for AI enhancement.

Step 2: Export a Clean JPEG​

Download a clean, high-resolution JPEG from your 3DStreet scene using the share button. This will serve as the foundation for our AI enhancement process.

Capture the scene from an angle that best showcases your street design, as this will be the basis for all subsequent enhancements. Make sure to turn off the 3DStreet Logo and the Scene Title (this is a Pro feature) to create a clean export.

Step 3: Import to Black Forest Labs for Photorealistic Enhancement​

Next, we'll import your screenshot into the Black Forest Labs Playground to add photorealistic shading and lighting. This AI-powered tool will transform your basic 3D render into a more realistically shaded image.

Upload your JPEG and for the Flux Kontext [pro] Model to work its magic. You can use a basic prompt like "add photorealistic shading and lighting" or add more details for specific styling changes if desired.

You'll get a couple different example outputs to choose from, each with its own interpretation of how to make your scene more photorealistic.

Output Before & After Comparison​

Step 4: Use the Fill Tool for Structural Changes​

Another powerful feature from Black Forest Labs is the Fill tool, which allows you to modify specific parts of your scene. For example, you can change the structure of buildings to have something more realistic looking for suburban single-family homes.

Once you have a source image loaded from a previous enhancement, or directly from 3DStreet, and you're viewing it in full screen from the BFL Playground, click the three-dots in the lower-right-hand corner and then click "Fill".

In Fill mode, click and drag to create a mask (or outlined area) to indicate the area in which you'd like to replace with new imagery.

Then, enter a prompt to automatically fill the area with whatever you desire, such as:

• Replacing generic buildings with more realistic architectural styles
• Add detailed facades and realistic building materials
• Modify landscaping and environmental elements
• Enhance street infrastructure with realistic details

In this example we used the prompt, "replace with suburban single family homes":

Now we've created a customized, high quality rendering ready for sharing with stakeholders.

Step 5: Create Animation with Google's Veo​

We can take it one step further, by turning this enhanced static image into a dynamic animation! Take your photorealistic screenshot, right-click to download it, and load it into Google's Veo through Google Labs Flow.

Open up the Flow tool, choose "Frames to Video", then upload your enhanced image, then choose that image as the first frame, then enter a prompt such as "Animate this suburban street scene."

The system will take about a minute to generate your animation options.

Step 6: Choose Your Animation Style​

After processing, Google's Veo will show you a couple different animation options. Each will have its own interpretation of how to bring your street scene to life with:

• Subtle camera movements
• Environmental animations (trees swaying, etc.)
• Lighting changes
• Atmospheric effects

You can choose which ones you like best, or sometimes you'll need to adjust your prompt slightly to get an output that perfectly matches your vision.

Tips for Success​

• Start with a clear, well-composed 3DStreet scene - The quality of your final animation depends heavily on your initial composition
• Experiment with different prompts - Small changes in your description can yield dramatically different results
• Be patient with processing times - AI rendering can take a few minutes, but the results are worth the wait
• Save multiple versions - Keep copies of your work at each stage in case you want to try different approaches

The Results​

And there you go! You've successfully transformed a basic 3DStreet scene into a photorealistic, animated visualization that can effectively communicate your street design vision to stakeholders, community members, or anyone who needs to understand your proposed improvements.

This workflow opens up exciting possibilities for urban designers, planners, and advocates who want to create compelling visualizations without needing advanced 3D animation skills.

Video Tutorial​

Here's a video version of this tutorial:

The new AI Assistant is here, and it's transforming how we create and modify street designs in 3DStreet! To help you hit the ground running with this powerful new feature, we've compiled a list of 10 essential prompts that showcase its capabilities and will help you optimize your design workflow right away.

Getting Started with AI Assistant Prompts​

Following our recent announcement of the AI Assistant beta, we've already seen users discovering creative ways to leverage this tool. These prompts range from simple shortcuts to complex design tasks – all made accessible through natural language commands.

Let's dive into the top 10 prompts you should try today:

1. Make a Street​

The most fundamental prompt is also one of the most powerful. Simply click the "Create Street" button shortcut or type something like:

Create a street with two drive lanes, two sidewalks, and two bike lanes

This instantly generates the foundation of your scene, and you can easily modify these parameters before generation. It's the perfect way to kickstart any project without manually placing individual street elements.

2. Take Snapshots​

Documentation and presentation are crucial parts of the design process. With the "Take Snapshots" button shortcut, the AI Assistant will capture three views of your scene:

• Your current perspective view
• A bird's eye view from above
• A head-on view of your street

These images export directly as PNGs to your computer, making it effortless to share your vision or include in presentations.

3. Generate Report​

Need to communicate the impact and purpose of your design? The "Generate Report" button shortcut prompts a form where you can specify:

• Current conditions
• Problem statements
• Proposed solutions

The AI Assistant then creates a report and generates snapshots highlighting how your design addresses the specified challenges -– useful for stakeholder presentations or community engagement.

4. Insert Objects​

Populating your scene is simple with commands like:

Add three benches along the north sidewalk

Place more trucks on the street

The AI understands these requests but sometimes still requires manual placement adjustments. We're still learning how to improve its spatial understanding, so please use the thumbs up or down button on the response depending on how accurate the AI is.

5. Rotate Elements​

Fine-tuning your scene is easy with rotation commands:

Rotate the food truck 45 degrees clockwise

Turn the bicycle rack to face the street

This is an alternative to manually adjusting rotation values in degrees.

6. Generate Ideas​

The AI Assistant isn't just for execution – it's an ideation partner too. Try prompts like:

What elements could I add to make this street more pedestrian-friendly?

Suggest ways to improve bicycle safety at this intersection

The assistant responds with context-aware suggestions based on urban design principles and your current scene.

7. Clone Objects​

Quickly populate your scene with multiple similar elements:

Clone this car three times along the parking lane

Duplicate this tree every 20 feet along the sidewalk

This dramatically speeds up the process of adding repeating elements like street trees, parking spots, or seating areas.

8. Resize Objects​

Adjust dimensions easily with natural language:

Make the sidewalk 50% wider

Resize the bike lane to 6 feet wide

This tool is particularly useful when adjusting street dimensions to meet specific standards or when testing different space allocations.

9. Change Materials​

Transform the look and feel of your scene instantly:

Change the road surface to concrete

Update the bike lane material to green asphalt

This capability allows for quick material studies without navigating through material menus.

10. Change Objects​

Need to swap elements in your scene? Simply tell the AI what you want to replace and with what:

Replace the sedan with a food truck

Change the maple trees to palm trees

This flexibility makes iterating through design alternatives seamless and efficient.

Unlocking New Possibilities​

These prompts are just the beginning of what's possible with the AI Assistant. As you become more familiar with its capabilities, you'll discover your own workflow shortcuts and command combinations that make street design faster and more intuitive than ever before.

The beauty of the AI Assistant is its ability to understand natural language, so don't be afraid to experiment with different phrasings and commands. The system is designed to interpret your intent, even if you don't use the exact wording shown in our examples.

Provide Feedback​

New Built-in Feedback Mechanism​

Since the release of this feature we've added a feedback mechanism to the AI Assistant. After a text and/or function call response, you can click the thumbs up or down button to rate the response.

Pressing either thumbs up or down will rate your conversation for review and will help us improve the AI Assistant over time.

Share your Discoveries​

Have you found an innovative way to use the AI Assistant? We'd love to hear about it! Join our Discord community (link in the footer) and share your favorite prompts and workflows. Your insights could help fellow designers unlock even more potential from this powerful tool.

For more information on the Console, check out our documentation and stay tuned for future updates as we continue to enhance this feature based on your feedback.

Happy designing!

Today we're excited to announce the beta release of the 3DStreet AI Assistant, a new way to create and modify street designs using simple natural language. Just describe the street scene you want to create, and watch as your vision comes to life – no complex 3D editing required.*

Empowering Everyone to Design Better Streets​

At 3DStreet, our mission has always been to provide tools that empower anyone to rapidly prototype visions for improving our shared environment, starting with streets. The introduction of our AI Assistant marks a significant step toward making street design more accessible to everyone – from urban planning professionals to community advocates who may not be familiar with traditional 3D editing tools.

With the new AI Assistant, you can:

• Create complete street scenes from scratch using natural language descriptions
• Modify existing designs by simply explaining what you want to change
• Automate repetitive tasks that would normally require multiple manual steps
• Explore design alternatives quickly without needing to master complex editing tools

How It Works​

Using the AI Assistant is simple. Just click on the new "Assistant" icon at the bottom of the 3DStreet interface and start describing what you'd like to create or modify.

For example:

• "Create a two-way street with protected bike lanes and mature trees on both sides"
• "Add a mid-block crosswalk with a pedestrian refuge island"
• "Change the bike lanes to be at sidewalk level with planters as separation"
• "Add outdoor seating areas in front of the cafes on the north side"

The AI Assistant interprets your requests and performs the necessary operations to bring your descriptions to life, handling all the technical details behind the scenes.

Why We Built This​

We've heard from many users – especially those new to 3D design tools – that while they have clear visions for better streets, translating those ideas into 3D visualizations can be challenging. Creating even simple scenes often requires learning multiple interface elements and performing several discrete actions.

When we analyzed user behavior, we found it difficult to understand user intent through traditional analytics. Is someone moving around the scene because they're confused, happy with what they made, or looking for something specific? It's hard to tell.

By providing a natural language interface, we can:

1. Break down barriers to entry for non-technical users
2. Better understand user needs through their natural language requests
3. Provide an alternative UI for operations that can be challenging or repetitive when using the existing interface

Perhaps most importantly, it allows people to focus on the "what" of their street design rather than the "how" of the software.

Under the Hood​

The 3DStreet AI Assistant is powered by Google's Gemini LLM via its Vertex AI API and our custom React-based chat interface. The most useful part of the Assistant is its ability to call tools that perform actions in the 3DStreet Editor. These tools include entity placement, entity updates, street segment creation and modification, camera control, geospatial location updates and project information memory. When a tool is called, the interface will display a yellow notification while pending and resolve into a green or red notification depending on the result of the function call. You can click the function call notification to see the details of the function call.

Building the AI Assistant has helped us improve the underlying 3DStreet codebase. By creating clear function calls for the AI to use, we've had to improve our code structure and interfaces for a number of underlying features, especially street editing through our new Managed Street Format.

The Fine Print: Beta Status and Availability​

Beta Status​

The AI Assistant is currently in beta testing, which means it's still evolving and may occasionally (often?) misinterpret requests or behave unexpectedly. During this early release period, the AI Assistant is available free of charge to all 3DStreet users with some caveats below.

We're actively collecting feedback to improve the experience and would love to hear about your interactions – both successful and challenging! Please share them in our Discord server or other social media (links at the bottom of this page).

Availability​

During this early release period, the AI Assistant is available free of charge to all 3DStreet users, but you must have an account and be signed in. We reserve the right to set daily caps for free users if usage exceeds our budget limits. We reserve the right to log chat conversations to improve the AI Assistant and our street design tools. Access to this feature is subject to the terms of our Terms of Service.

What's Next?​

We're just getting started with AI integration in 3DStreet. Tell us what you'd like to see as we continue to develop this feature! Some ideas we've had so far include:

• Better guidance for users wishing to create scenes from scratch, such as a Q&A flow for street geometry
• Support for more specialized street design elements
• Better leveraging of existing street data from geospatial sources to assist LLM with street geometry and placement of street elements
• Integration with local LLM APIs for users who prefer to run AI locally
• Expanded automation capabilities for professional workflows

FAQs​

Q: Is this a distraction from fixing other features?​

A: Improved street geometry editing – especially better intersections support – continues to rank highly on our roadmap. Some users also share frustrations with elements of the 3DStreet user interface, especially new users. Providing simple interfaces for navigating and editing detailed scenes over a large 3D space is challenging. We hope the AI Assistant complements ongoing feature development by making existing features more accessible while we continue to develop more advanced capabilities. And sometimes developing UI for a new spatial feature is harder than developing the underlying code, making the AI Assistant a great option to introduce more experimental features.

Q: Is this replacing the current editing interface?​

A: Not at all! The AI Assistant is an additional option that works alongside our traditional editing tools. Power users can continue using the full interface for precise control, while switching to the AI Assistant when it's more convenient.

Q: What if the AI Assistant doesn't understand what I want?​

A: The technology is still evolving, and some requests may require clarification. You can always switch back to manual editing if needed. Your feedback helps us improve the system's comprehension over time.

Q: Why did you choose the Gemini LLM model?​

A: We've tested a few different models and found gemini-2.5-flash to be the most effective for this use-case. Gemini appears to perform well with large context tasks, and our system prompt combined with a user's scene graph and other scene data elements can drive up context length quickly.

Try It Today​

The AI Assistant beta is available now in the 3DStreet app. Just look for the "Assistant" icon in the bottom of your screen to get started. We can't wait to see what you create!

The future of urban planning is happening right on our streets. In a recent webinar featuring 3DStreet founder Kieran Farr and InCitu's Nick, we explored how augmented reality (AR) is transforming community engagement with street design proposals.

Read our written adapation or skip to the original webinar recording to learn more.

The Challenge: Bridging the Communication Gap​

Urban planning has long struggled with a fundamental problem: spatial communication is extremely difficult. When engineers present striping drawings or two-dimensional plans to community members, you can see the disconnect in people's eyes. Small business owners, residents, and stakeholders often can't visualize how proposed changes will impact their daily lives.

This information asymmetry between professionals and communities leads to predictable outcomes: projects lose political support, get watered down, face delays, or are canceled entirely.

Three AR Breakthroughs​

1. AR as a Rapid Prototyping Feedback Loop​

The game-changer isn't using AR to display finished designs—it's using AR for rapid iteration at the earliest stages. Instead of spending significant time in AutoCAD before getting community feedback, planners can now create quick prototypes, test them on-site, and iterate based on real-world constraints.

2. Converting Prototypes into Outreach Materials​

Once refined through iterative AR testing, those same prototypes become powerful communication assets and can serve as the foundation for detailed design work in traditional tools like AutoCAD.

3. Democratizing Design: Anyone Can Do This​

These tools are accessible to both professionals and amateurs. Community advocates, planning staff, and engaged residents can all participate in the design process.

Real-World Case Studies​

Valencia Street Bike Lane Reconfiguration​

A community advocate organized visualizations for the conversion of Valencia Street from center-running to curbside bike lanes.

The AR perspective the new configuration from the actual perspective of a crossing pedestrian was recorded to video to provide content for education and advocacy.

Queens Boulevard Improvements​

A quick sample project created by Nick Kaufmann, a community manager at InCitu, to demonstrate how quickly he can create an example scene showing a sidewalk widening and protected cycle track improvement project.

Buffalo Pedestrian Bridge​

A pedestrian bridge connecting an underserved neighborhood to park amenities cut off by highways. The AR visualization told a dual story—both the bridge architecture and the narrative of neighborhood reconnection. The project was completed in 2024.

Columbus Bus Rapid Transit​

Central Ohio Transit Authority integrated AR into actual bus tours along proposed routes. The technology showed not just transit infrastructure, but walkable neighborhood development that transit investment could encourage. Officials used AR's people-occlusion feature for authentic social media content.

The Hayes Promenade, San Francisco​

Community advocate Andrew Seigner with Hayes Promenade proposed converting a two-lane road with retail and parking into a single-direction lane with expanded pedestrian space. He "created an initial 2D design in Streetmix, imported it into 3DStreet, built it out in 3D space, and then use the AR Ready export option. This resulted in a complete augmented reality demo, where users can walk down our street with their phones, experiencing the proposal in first-person."

Best Practices for AR Community Engagement​

Make It a Group Experience​

AR isn't about individual visualization—it's about unlocking conversations. Successful implementations bring diverse groups together, often reaching stakeholders who wouldn't attend traditional civic meetings.

Meet People Where They Are​

Take AR proposals to farmers markets, street festivals, and existing community events. Mobile AR's portability makes it possible to engage communities in familiar, comfortable settings.

Mix AR with Other Tools​

The most effective campaigns combine AR with traditional outreach methods, web-based 3D maps, and other communication tools. In Columbus, embedding AR into a real bus tour created an immersive "trip to the future" that made abstract planning concepts tangible.

The Emotional Power of Spatial Storytelling​

Data alone doesn't change minds. When engaging with community members, statistics "roll off them like oil and water." The response to change is often emotional—people may intellectually understand that street safety is important, but they feel the loss of a parking space more acutely.

AR's power lies in turning fear of loss into collaborative opportunity. Instead of focusing on what's being taken away, these tools help communities envision positive alternatives: "You may be losing a parking space, but here's what you gain—and what would you like to gain?"

When people can see five new restaurant tables or beneficial rain gardens that prevent flooding, they start thinking: "I have agency to decide what could be there." This shift from loss aversion to collaborative visioning is where real change happens.

Getting Started​

Both 3DStreet and InCitu offer free tiers for experimentation. The basic workflow is simple:

1. Create a scene in 3DStreet with your target location
2. Add and modify street elements using drag-and-drop tools
3. Export as an AR Ready model
4. Import to InCitu for location-based AR deployment
5. Test and iterate on-site, gathering community feedback

The most important insight from practitioners? Start imperfect and iterate quickly. The goal isn't award-winning renderings—it's making spatial concepts tangible enough for meaningful community engagement.

As cities face mounting challenges around housing, climate change, and equity, tools that democratize urban design and accelerate community engagement aren't just nice to have—they're essential for building sustainable, inclusive neighborhoods.

Webinar Recording​

Ready to experiment with AR for your next community engagement project? Contact us to get started.

Have you ever found yourself designing a street scene and wishing you had that perfect 3D model to bring your vision to life? Whether it's a unique piece of street furniture, a local landmark, or even a toy car, traditional 3D modeling can be time-consuming and requires specialized skills. But what if you could turn any object into a 3D model with just a couple of photos and some AI magic? Let's explore how to do exactly that!

The Power of AI-Generated 3D Models​

Until recently, creating custom 3D models required extensive knowledge of 3D modeling software and hours of work. Now, thanks to advances in AI technology, tools like Microsoft's TRELLIS can transform simple photographs into usable 3D models in minutes. This opens up exciting possibilities for urban designers, planners, and anyone interested in creating more imaginative and personalized street scenes.

Step-by-Step Guide to Creating Your 3D Model​

1. Capture Your Reference Images​

The quality of your 3D model starts with good reference photos. Here's what you need:

• Take 1-2 clear photos of your object from different angles
• Ensure good lighting and contrast
• A neutral background works best
• Include both front and side views when possible

2. Generate Your 3D Model with TRELLIS​

Head over to TRELLIS on Hugging Face Spaces to start the conversion process.

1. Upload your reference photos -- make sure to click on the "Multiple Images" tab if you have more than one
2. Click "Generate" and wait for the AI to work its magic. Review the preview to ensure it captures your object well.
3. If the preview looks good, click "Extract GLB" to generate a GLB version of the model.
4. Click "Download GLB" to download your 3D model

If the first attempt doesn't look quite right, consider using more than 1 picture. Try for soft lighting instead of bright specular reflections and ensure the object is in focus and entirely in frame.

You can also try Hunyuan3D-2, which allows you to generate 3D models directly from text descriptions.

3. Validate Your Model​

Before proceeding, it's helpful to verify your model:

1. Visit gltf-viewer.donmccurdy.com
2. Drag and drop your GLB file into the viewer
3. Confirm the model loads correctly and looks as expected

4. Optimize for Web Use​

Your initial model might be larger than necessary. Here's how to optimize it using OptimizeGLB.com:

1. Upload your GLB file. Enable Draco compression (but avoid mesh simplification)
2. Click "Process All Files"
3. Right-click the "Download" button, and
4. Click "Copy Link Address"

This step not only reduces file size but also gives you a hosted URL for your model.

5. Import into 3DStreet​

Finally, let's bring your model into 3DStreet. Open the 3DStreet App, and then:

1. Click "Add Layer"
2. Choose the "Custom Layers" tab
3. Select "GLTF Model from URL"
4. Paste your OptimizeGLB link URL
5. Adjust scale, position and rotation as needed after adding to the scene

See our documentation on importing custom 3D models for more details on this step. A 3DStreet Pro plan is required to use this feature.

Tips for Success​

• Start with simple objects that have clear shapes and details
• Take photos in good lighting conditions
• Use the validation step to catch any issues early
• Consider the final size and scale in your street scene
• Keep backup copies of your original files

Share Your Creations!​

We'd love to see what you create with this workflow! Share your models and scenes with the 3DStreet community on our Discord channel or tag us on social media.

Looking Ahead​

The intersection of AI and 3D modeling is evolving rapidly. Tools like TRELLIS are just the beginning, we're also seeing huge improvements in live style transfer to convert the low poly and low fidelity 3D tiles imagery into higher fidelity and more detailed 3D scenes.

Here's an example of using live image to image style transfer with KREA.ai, a user-friendly AI generative creation and manipulation service.

Remember, whether you're adding unique street furniture, local landmarks, or playful elements to your scenes, AI-powered 3D modeling can help bring your vision to life with unprecedented ease.

Want to see this process in action? Check out our video tutorial below:

Happy modeling! 🚀

Map Layer Blending​

One of the most exciting new features we've been working on is adding a number of custom blending options for 3D map layers. To enable this, use the new Blending option in the Geospatial Layers panel, then choose your desired Blend Mode.

Environment Settings​

A common request is changing background and colorization of the environment in 3DStreet scenes. This is now possible with the new Environment Settings panel. Choose from presets including day, night, sunny morning, cloudy afternoon, and more. Or choose the color preset to select a custom background color.

Combining Blending and Environment Settings​

By combining blending and custom background colors you can create a huge range of creative interpretations for your scenes. Here's an example using the Dark blending mode and adjusting a custom background color:

Focus Animation Adjustment​

When you double-click on an object in the viewport or in the layers panel, you'll notice a brief focus animation as the camera zooms into the focused object. The new Focus Animation Speed selector allows you to adjust the speed of the animation which can make it more suitable for creating quick videos and sharing. The focus animation is a simple way to add fly through effects to your scenes when sharing live or when recording with screen capture software. Click Viewer > Focus animation > Speed > slow to slow down the animation.

Combining Everything Together​

Combining all of these features together can create a wide range of creative options for your scenes.

Try it out!​

Click the "Launch App" button in the upper-right hand corner to play with these features (Pro plan required for Geospatial integrations) or join our discord below to provide more feedback on these features and what else you'd like to see in the future.

A Capital City in Transition​

Dodoma, located about 400 kilometers inland from Tanzania's largest coastal city Dar es Salaam, is experiencing unprecedented growth as it embraces its role as the national capital. Formally declared the capital in 1973 and home to the National Parliament since 1995, the city has seen accelerated development since 2016 when the government began actively relocating institutions from Dar es Salaam.

Growing Pains on Major Corridors​

The city's main arteries, particularly the Dodoma-Morogoro Road, currently reflect the growing pains of this urban transformation. The road serves multiple functions as it traverses urban, rural and regional landscapes, requiring careful design consideration to accommodate various user demands while maintaining principles of sustainability and connectivity.

Today, this vital corridor operates as a two-lane arterial road accommodating a diverse mix of traffic:

• Motorcycles and tuk-tuks (bajaji)
• Private cars and vans
• Light and heavy commercial vehicles
• Public transportation buses

While the current infrastructure technically supports both walking and cycling, it predominantly caters to motorized travel - creating challenges for sustainable urban mobility.

From Concept to Digital Twin​

Our team worked with key stakeholders including Tanzania National Roads Agency (TANROADS), Tanzania Rural and Urban Roads Agency (TARURA), Land Transport Regulatory Authority (LATRA) and Dodoma City Council (DCC) to explore treatments that would better support regional infrastructure needs.

As analysts entering the workforce, our background in civil engineering was very limited compared to some of our tenured colleagues and professional partners, making it challenging to convert traditional engineering drawings into visualizations that we could then leverage for our work with stakeholders. That's when we discovered 3DStreet, which allowed us to rapidly prototype different scenarios from the foundational engineering work of our professional staff and partners.

Using 3DStreet's prototyping capabilities, we created digital twins of both existing conditions and proposed changes:

• A six-lane dual carriageway with 2 dedicated public transport lanes for busses
• Two-lane service roads on both sides for local access
• Wide, shaded walkways and cycleways for non-motorized transport
• Integration with adjacent land use to support sustainable urban development

The proposed design creates dedicated space for all modes while improving safety and connectivity.

Bringing the Vision to Life​

We found the ability to instantly transform cross-sections designed in Streetmix to interactive 3D visualizations with satellite map layers in 3DStreet to be transformative to understanding the true impacts and benefits of the proposed project in context with the existing environment.

Also, using a combination of 3DStreet and Mapillary’s street view imagery to create a before and after first-person perspective was particularly impactful - it helped both our team and our local partners understand how the proposed changes would affect the actual experience of using the street. Being able to visualize the separation between modes and the scale of proposed facilities made the benefits much more tangible.

By combining all the elements of 3D GIS data and Street elements from 3DStreet combined with the use of real-world ground truth from Mapillary imagery, we were able to create a very effective video visualization of the project background and potential future scenarios.

Looking Ahead​

As our internship draws to a close, we're excited to see Dodoma continue its trajectory of growth aided by our World Bank colleagues investing in Tanzania’s future. We’re proud of the work we’ve done to create digital twin representations of these projects to better engage local stakeholders, and hope that the team is able to leverage our work to envision the future of our country’s sustainable transportation infrastructure.

Read More​

Read more information about the project on the World Bank website

Disclaimer: The views expressed in this blog post do not necessarily reflect the views of the World Bank Group.

An Island at the Crossroads​

Located in the central Pacific Ocean, Kiritimati Island faces unique challenges as climate change reshapes the region. While the island's current population numbers only a few thousand, it may soon need to accommodate more residents from Kiribati’s other island which is increasingly vulnerable to rising sea levels.

Planning Ahead for Change​

The island's main road network currently reflects its small population and historic development patterns. The existing infrastructure is basic, with limited provisions for pedestrians, cyclists, or public transport. As we look toward a future where Kiritimati could play a crucial role in regional climate migration, these roads will need to evolve to serve a larger, more diverse population.

The current road infrastructure supports:

• Private vehicles, motorcycles, and cargo trucks
• Limited pedestrian movement
• Basic access to key facilities and settlements
• Connection between main population centers like London Town and Banana

From Reality to Digital Twin​

Working with key stakeholders including local government authorities and the International Road Assessment Program, the team explored design solutions that would create safer, more inclusive streets while preparing for potential population growth.

Using 3DStreet's visualization capabilities, we created digital twins of both current conditions and proposed improvements across three critical areas that represent the diverse needs of transportation infrastructure on the island.

Three areas of focus for the Kiritimati project

Reimagining Street Space​

Proposed designs incorporate several key features to support sustainable mobility:

• Wider, shaded walkways to protect pedestrians from the tropical sun
• Slow speed zones in urban areas that safely mix vehicles, motorcycles, and bicycles
• Dedicated bus bays to support future public transport services
• Enhanced safety features based on iRAP assessment recommendations

Making the Case for Change​

By combining Mapillary's street-level imagery with 3DStreet's visualization tools, we were able to create compelling before-and-after comparisons that helped stakeholders understand the tangible benefits of the proposed changes. This approach was particularly valuable given the unique context of planning for both current needs and potential future scenarios.

The visualization process helped demonstrate how relatively modest infrastructure improvements could significantly enhance safety and accessibility while building climate resilience. These visualizations are proving instrumental in making the case for infrastructure funding that considers not just today's needs, but tomorrow's challenges.

Stretching the Capabilities of 3DStreet​

The Kiritimati project pushed our platform in exciting new directions. While 3DStreet was originally designed for visualizing individual street segments, the World Bank team's requirements encouraged us to significantly expand our capabilities. The project's scope required handling multiple kilometers of roadway across three distinct sections, driving us to optimize our rendering pipeline while maintaining detail.

Creating locally appropriate assets and smooth fly-through animations that transition between current and proposed conditions required new approaches to both asset development and camera control. These technical challenges led to innovations in scene management and rendering that have since become standard features in our platform, benefiting visualization projects worldwide.

Looking Forward​

As climate change continues to impact Pacific island nations, the importance of forward-thinking infrastructure planning becomes increasingly critical. Through this project, we're demonstrating how digital twin technology can help communities like Kiritimati prepare for complex demographic and environmental changes while improving quality of life for current residents.

The visualizations we've created are helping stakeholders understand how streets can be transformed from basic transport corridors into vibrant, safe spaces that support sustainable mobility for all users. As we continue to refine these designs, we're excited to see how this approach could serve as a model for other islands facing similar challenges across the Pacific region.

Read More​

Read the press release from International Road Assessment Program

Disclaimer: The views expressed in this blog post do not necessarily reflect the views of the World Bank Group.

In these cases, our new integration with Open Street Map (OSM) offers new dimensions to our mapping capabilities.

A screenshot of the new 3DStreet OSM integration showing San Francisco City Hall.

Introducing Open Street Map Integration​

3DStreet now offers integration with Open Street Map (OSM) as a new Geospatial Layer map source. This integration provides 2D political map ground tiles and 2.5D buildings through OSM GeoJSON buildings data. This feature is now available for all Geospatial Pro Plan users in addition to the existing map data providers Google Maps 3D Tiles and Mapbox 2D Satellite Maps.

Click Geospatial Layer > Map Source > osm3d to use the new OSM data source.

The Advantage of Open Source Maps​

Integrating OSM with 3DStreet continues our commitment to open-source collaboration. A key advantage of the OSM approach is that it allows you to leverage the existing OSM community and resources, ensuring that the data remains up-to-date and accessible to all. The 3DStreet integration today is "read only" and relies on existing OSM data, but users wishing to update map data can do so through a variety of tools from the OSM ecosystem.

More than just tiles - GeoJSON buildings​

In addition to Open Street Map 2D ground tiles, the new OSM integration provides basic building geometry through GeoJSON OSM data sources. The basic building footprints provide just enough context in most cases for locals to understand the location of a proposed project, while not distracting from the focus on the proposed project itself.

This GeoJSON parser also opens the door to future integration of alternative geospatial data sources, although today this only supports closed polygons for buildings.

A-Frame Component for OSM Integration​

Similar to our other integrations, we've chosen to "build in the open" by contributing improvements to an existing A-Frame OSM4VR component designed by anonymous community member ctrlw.

You can view the original A-Frame component by ctrlw here https://github.com/ctrlw/osm4vr and check out the 3DStreet version here which adds webpack support to better handle dependencies as well as custom features to allow for integration with the 3DStreet Editor app. Like all parts of 3DStreet, we welcome community contributions to this component and others.

What's Next?​

Looking ahead, we're excited about the potential to support custom tiles. In addition to the political map, we aim to move beyond this basic OSM implementation and support private servers and alternate tile types. Our ultimate goal is to allow users to specify which tile type they prefer, providing even greater flexibility and customization.

Stay tuned for more updates as we continue to enhance 3DStreet's mapping capabilities and explore new ways to empower our users to reimagine public spaces.

You can help us shape the future of the 3DStreet OSM integration by providing your feedback through email, GitHub issues, or our public Discord server (link in footer).

Introduction​

I got involved with this project through my ongoing work with Dr. Marc Schlossberg at the University of Oregon, where I’ve been assisting him in exploring how Augmented Reality can improve street safety in Eugene, Oregon. This experience has given me the opportunity to merge my passion for design with city planning and street safety, topics I never thought I’d be involved in.

This whole experience has been transformative for me, especially in sharpening my 3D modeling skills. As a product designer, I’m used to working with highly detailed models, often with enormous file sizes. But for the models I created to function on mobile AR platforms, I had to rewire how I approached 3D design. When Dr. Schlossberg initially told me to create models with the smallest polycount possible —- without losing their identity —- I was at a bit of a loss on where to begin.

Working with Dr. Schlossberg and team of graphics professionals from 3DStreet like Kieran Farr and GnanaSai Reddy pushed me to expand my knowledge of low-poly model design and the associated pipeline for preparing 3D assets for high performance real-time XR applications -- an area I knew little about before this project. I’m incredibly grateful for the opportunity to collaborate with such experienced individuals, and I’m proud of how much I’ve learned throughout the process.

Pictures and descriptions of 3D Models​

Traffic calming bumps: This object is versatile and low profile, used traffic control, it is effective at alerting drivers of low speed areas ahead

Corner island: This object was created in order to provide safer turns for cyclists in city intersections while still allowing a safe place for pedestrians to stand, it gives bikers full protection from cars throughout the entire bike lane

Parklet: These objects are a great use of space for extra seating around restaurants, parks, and rest areas in cities

Picnic table, USPS Mailbox, Bin, Residential Mailbox: Variations of these objects are common in nearly every city around the world, it seemed necessary to add them to the library of 3DStreet for added realism to city scenes

Conclusion​

Working alongside Dr. Schlossberg, Kieran Farr, and the team at 3DStreet has been an incredible experience, and I’m truly grateful for the opportunity to learn from such talented professionals. As I look ahead, I’m excited to embark on my next adventure in the design world—ideally with a design consultancy in Portland, Oregon—where I hope to collaborate with another group of creative, driven designers and continue growing both personally and professionally.

Use these traffic safety objects now!​

Use these models now to create your own 3DStreet scene by launching the 3DStreet app, then click + to add a new layer, and find these new models in both the Sidewalk Props and Dividers & Traffic Control categories.

New Microsoft Login Option​

Users can now sign in to 3DStreet using their Microsoft accounts. This feature is available alongside our existing Google sign-in option, providing more flexibility for our users.

To use the new Microsoft login:​

1. Click on the "Sign in" button in the 3DStreet editor.
2. In the sign-in modal, you'll now see two options: Google and Microsoft.
3. Click on the Microsoft button to proceed with Microsoft authentication.

This addition makes it easier for users in organizations that primarily use Microsoft services to access 3DStreet seamlessly.

Preview of the new Microsoft login flow:​

Look for Publisher Identity Verified​

3DStreet LLC is a Microsoft Verified Partner. You should see a blue checkmark next to our name and clicking shows the following "App info":

Administrator Guide: Enabling 3DStreet for Your Organization​

For organizations using Microsoft Entra ID (formerly Azure AD), administrators may need to grant consent for 3DStreet to access user data. Here's how to do it:

1. Sign in to the Microsoft Entra admin center as a Global Administrator.
2. Navigate to "Identity" > "Applications" > "Enterprise applications".
3. Search for "3DStreet" in the list of applications.
4. Select the 3DStreet application.
5. In the left-hand menu, click on "Permissions".
6. Click on the "Grant admin consent for [Your Organization]" button.
7. Review the permissions requested by 3DStreet and click "Accept" if you agree.

By granting admin consent, you're allowing all users in your organization to use 3DStreet with their Microsoft accounts without individual consent prompts.

Benefits of Admin Consent​

Granting admin consent offers several advantages:

• Streamlined access for all users in your organization
• Reduced individual consent prompts, improving user experience
• Centralized control over application permissions

For more detailed information on the admin consent process, please refer to the Microsoft documentation on application consent experience.

Feedback and Support​

We're excited to bring this new login option to our users and make 3DStreet more accessible for organizations using Microsoft services. If you encounter any issues or have feedback about the Microsoft login process, please don't hesitate to reach out to our support team.

Happy designing with 3DStreet!

In this post I'll share how I got involved in 3DStreet, the features of the Intersection component, and what's coming next!

I started my career working for Cesium on CesiumJS, helping hundreds of developers build bespoke applications around global-scale geospatial data. It was through this open-source 3D work that I met Kieran from 3DStreet and shared his vision of a world where everyone was empowered to reimagine their neighborhood.

I was inspired to help contribute as an open-source developer and use my background in web-based 3D rendering and full stack engineering to bring the project forward.

Many 3DStreet users have been asking for support for visualizing intersections in their neighborhoods. It turns out, intersections are hard! So we decided to start with something simple: the goal for this initial component was to create a simple yet versatile interface for connecting multiple streets. The component includes options for adding curbs, sidewalks, crosswalks, stop signs, and traffic signals. We made this available in the 3DStreet API for other developers in June 2022, but it wasn’t until now – 2 years later – that the user interface and editor capabilities of 3DStreet have finally caught up to make this available for people to use without writing code.

Here’s a quick overview of the capabilities:

Create 4-way 90º intersections with custom sizing for each direction​

A custom intersection example with painted crosswalks along two sides of the long crossing distance, such as a mid-block crosswalk across an arterial where it meets an alleyway.

Customization Features​

• Supports optional dead-ends and t-intersections
• Supports optional traffic control devices: traffic signals, stop signs, or none
• Supports optional pedestrian features: crosswalks, sidewalks (for t/dead-end), or none

An intersection example using nearly every feature.

What's next for 3DStreet Intersections?​

We call this release Early Access for a reason -- 3DStreet Intersections need more work and we need your help to prioritize what needs improvement. Do we need better geometry? Curb ramps? Better interface to edit the intersection? We'd appreciate ideas and suggestions on how to help you visualize your perfect intersections.

What’s next for Sam?​

I’m honored to be a contributor to the first iteration of a critical component of 3DStreet. Since my 3DStreet contribution, my career has propelled me to new roles including autonomous vehicle data visualization and a new announcement coming soon -- follow me on LinkedIn for the latest! I’m excited to see how people use this component to change the world!