Spatial streaming for high-fidelity 3D

3D render of a collectible character figure in green dragon armour and a dragon-head helmet, against a blue gradient

Stop reviewing the proxy. Review the asset.

To share a heavy 3D asset for review, you decimate it into a proxy or export a turntable to video with a fixed viewport. Either way, the sign-off happens on a copy, not the master.

The full-fidelity asset can't leave the viewport without friction. You can review it in person at the monitor, but the moment it has to travel, every option makes you give something up: a glTF proxy you can navigate but can't trust on fine surface detail, or a turntable video you can trust pixel-for-pixel but can't inspect. Miris streams from your source asset, progressively and at full fidelity, adapting to whatever device is on the other end, so the real asset opens on a phone in the field, an iPad in the studio, or a headset at home.

| THE PROBLEM

Creation isn't the bottleneck anymore. Delivery is.

Creatives are producing high-fidelity, interactive 3D content faster than ever before. The break is everything after creation: getting that asset to anyone who isn't sitting at the workstation. Every path that scales forces a compromise. You decimate the asset so it will travel on the web, or you freeze it in a video that can't be explored. The options that preserve both full fidelity and interactivity, like pixel streaming or remoting into a workstation, don’t scale, because they require a dedicated, active host session for every single viewer. Asynchronous review and feedback for 3D is missing a delivery layer as seamless as video.

| ASSET WEIGHT

Heavy assets can't leave the workstation

Production-grade 3D assets are heavy. Web-delivery best practice targets just a few megabytes per model. That gap is why the master gets decimated into a proxy before it can leave the workstation.

| DELIVERY COST

Cost scales with compute, not consumption

At one GPU instance per concurrent user on a G5, hosting 10,000 simultaneous viewers costs approximately $10,000 per hour. Even at shared-session ratios, the cost scales with sessions, not storage.

| SCALE CEILING

Broadcast scale isn't feasible today

A full-fidelity spatial 3D experience at broadcast scale isn't possible in today's per-session economics.

You already produce high-fidelity 3D. The problem is not creation. It is a delivery model that scales.

| THE solution

One asset. Every surface.

Stream from your source 3D at full fidelity to any device, with no decimating, no per-viewer cloud session, and no tradeoff between fidelity and reach. Progressive spatial streaming drops into the tools your teams already use and the products you're already building, and scales like video. Whether it is going to a room of review stakeholders or a million fans, it is the same interactive asset, at full fidelity, on every surface.

A VR headset, phone, tablet and monitor each displaying the same 3D cube asset

This asset is conditioned once upstream, then streams adaptively to the browser. The client renders locally, so no cloud GPU sits idle waiting per viewer.

The expensive optimization happens once, upstream. Delivery scales like video streaming, not cloud GPU rental.

| Proof

Abstract 3D cube streaming across a gradient field

ONE ASSET, EVERY SURFACE

One source asset. No platform-specific builds.

The same full-fidelity source that renders to the big screen opens in a phone browser, a studio iPad, or a headset. No decimated proxy, no separate web-optimized version, no per-device build to maintain. Miris conditions the asset once and adapts the stream to each device.

Line chart of cost to serve against concurrent viewers: per-viewer GPU sessions rise steeply while CDN delivery stays flat

DELIVERED LIKE VIDEO, AT BILLION-USER SCALE.

Scales like a CDN, not a render farm

Pixel streaming rents a cloud GPU for every viewer, so capacity becomes a hard ceiling and cost climbs with each concurrent session. A successful launch turns into an infrastructure problem. Miris delivers one asset over standard CDNs, cached and bandwidth-based, the same model that lets video reach billions of viewers. Reach scales alongside your audience, on the delivery economics of the web rather than a GPU session per viewer.

Miris runs its pipeline on CoreWeave infrastructure to condition assets once upstream, then streams them over standard CDNs.

| THE INFRASTRUCTURE MOMENT

Creation has outrun delivery

GenAI is helping push out high-fidelity 3D faster than ever, and when creation explodes, distribution has to follow. It happened with video: once millions could make it easily, YouTube, streaming, and global CDNs emerged to move it. 3D is at that moment now.

The Media and Entertainment industry is converging on OpenUSD as the standard for 3D production pipelines. The question is no longer which format to invest resources around. It is how you move those final-pixel-ready assets from the pipeline to the audience, fully interactive.

Miris is built for that delivery layer. Assets are conditioned once, then streamed to any browser or device, with no per-viewer cloud session to provision. The same high-fidelity asset that renders to the big screen is interactive in a web campaign, a virtual press event, or a product page.

Table titled The 3D moment, comparing how creation and delivery scaled for music, video and 3D

Stylized 3D render of a translucent handbag with a blue handle, streamed by Miris

Talk to us Explore the platform

Add 3D to your product, not to your roadmap

Create an account and stream your first asset today, or partner with us to embed spatial streaming in your platform.

Join the beta Partner with us

Careers

Streaming video reshaped media consumption; we're doing the same for 3D. Join a small team solving tough spatial streaming, content delivery, and developer experience challenges.

See open roles