The rack tells the truth. Open a legacy Crestron DigitalMedia system in a Greenwich estate and the architecture shows itself fast: a central chassis, long home runs, spare card slots treated like insurance, and every future room change already a little expensive.

That is why so many old DM systems stay in service. They work. The trouble starts when the house stops matching the original drawing set. A study wants USB-C. A smart TV app needs audio back on the system. The same Kaleidescape source now needs to land in more than one place without a pile of local boxes. At that point the question is not whether the house needs 4K. It is whether the video system is still a fixed switcher, or whether it is ready to behave like a network. On Crestron's current DM NVX platform, especially the DM-NVX-384 and DM-NVX-385 family, that difference is now very clear.[1][2]

The real difference is topology, not resolution

A legacy DM matrix is strong at one thing

A traditional DM matrix is a hardware crosspoint. If the source list is fixed, the display list is fixed, and the rack room is not changing, it can run for years with very little drama. Press a button on a Crestron TSW-1080, route Source 3 to Display 7, and the chassis does exactly what it was built to do.

What it does not like is change. Add another display and you are back in chassis math: outputs, cards, cable path, power, rack space, heat, and whether the original design ever left room for that zone. Add a local source at the display end and the work gets awkward fast. The whole system assumes the rack is the center of the world.

Anthony Grimani put the broader shift well in Residential Systems this spring: matrix switchers and point-to-point wiring made sense in a world of fixed layouts, but current residential projects want flexibility and scalability without invasive rework.[7] That is exactly where a legacy matrix starts to feel old.

DM NVX turns routing into a network problem

DM NVX changes the conversation because the switcher stops being a single box. The system becomes encoders, decoders, multicast planning, and control logic. A source can live in the rack, behind a display, or in a study cabinet. A display can subscribe to that source because the route exists on the network, not because a fixed hardware output was landed there a decade ago. If the endpoint count justifies it, DM NVX Director gives you matrix behavior back as software-managed routing instead of a fixed crosspoint.[3]

That is why a DM NVX upgrade is usually more useful than a one-for-one matrix replacement. The question stops being how many inputs and outputs the chassis has left. The question becomes where sources should live, which rooms need local inputs, and how routing should behave when the house changes later.

What current DM NVX hardware actually buys you

The DM-NVX-385 is not a generic endpoint

The DM-NVX-385 makes the upgrade argument easy to understand because it solves several residential problems in one device. Crestron's current manual shows two HDMI inputs, two USB-C inputs, 4K60 4:4:4 over standard gigabit Ethernet, support for 5K wide and super-wide resolutions, H.264 streaming, eARC, AES67 audio embedding and de-embedding, a 4K60 scaler, and 7.1 surround support with downmixing.[3] One endpoint can do work that used to sprawl across switchers, extenders, extractors, and adapters.

Crestron's May 5, 2026 DM NVX documentation update matters for the same reason. It shows the 38X platform still moving, not just staying alive.[1] In current guidance, the DM-NVX-384(C) and DM-NVX-385(C) family is where Crestron has focused the features that actually change residential design decisions.

Firmware 7.4 changed the audio and utility side of the system

Crestron's current getting-started guidance for DM NVX lays out the important 7.4 additions on the 384 and 385 family: ARC and eARC support, BTS multichannel audio-over-IP, and H.264 streaming.[2] Those sound like line items until you apply them room by room.

ARC and eARC matter because display-native apps are now real sources. If a Samsung app on a bedroom display is what the owner actually uses, the system needs a clean path to return that audio to the rest of the house. DM NVX now gives you that path.[2]

BTS matters because multichannel audio no longer needs to sit outside the video distribution conversation. H.264 matters because not every screen in a house needs the full main stream. A service station, preview panel, or secondary touch interface often wants visibility, not hero-room performance. Crestron also notes that the 38X family can combine up to six DM NVX video streams into one output in fixed Multiview layouts, which is useful in game-day or monitoring scenarios without adding another layer of hardware.[2][3]

DM NAX makes the upgrade more than a video conversation

Crestron's February 2, 2026 DM NAX announcement made something explicit that matters in residential work too: DM NAX uses DM NVX for audio-over-IP transport, adds AES67 interoperability, and auto-discovers devices on the network.[4] The practical takeaway is simple. A DM NVX upgrade is not only about replacing a video switcher. It is also a chance to clean up how stereo downmix, media-room surround, and distributed audio handoff actually work.

That is a real advantage on the DM-NVX-385. The same endpoint that moves the video can also pass or downmix 7.1 formats, including Dolby Atmos and DTS:X, and distribute stereo and multichannel behavior more intelligently than the older matrix world usually allowed.[3]

What changes in a Greenwich estate

Sources stay central, rooms stop being fixed

The biggest gain is not cosmetic. It is operational. Good sources can stay where they belong: in the rack, cooled properly, power managed properly, and easy to service. Kaleidescape, Apple TV, a cable or satellite box, or a game console do not need to be duplicated room by room just because the old matrix ran out of elegant options.

What changes is the destination end. A great room display can take the game from the rack. The study can pull the same source later. An outdoor screen can mirror it without relying on every TV to behave like its own island. If the owner wants the display's internal app instead, eARC gives that app a path back into the system.[2]

That is what owners usually feel first, even if they never see the rack: the house behaves like one AV system instead of a collection of unrelated televisions.

Local inputs stop becoming side projects

The older the project, the more likely it is that every request for a laptop input left scar tissue behind it: odd wall plates, local switchers, USB workarounds, or notes in the rack about which room can and cannot do what.

The DM-NVX-385 is useful precisely because it cleans that up. Its two HDMI and two USB-C inputs create a real 4x1 input switcher at the endpoint, and the USB-C side can carry USB 2.0 data back through the system.[3] A study monitor, flex office, or lounge display can support a modern laptop cleanly without turning the room into its own miniature AV build.

In a residence with a Crestron CP4-R or DIN-AP4-R at the center, those local inputs can be treated like real system sources. They live inside the same routing logic as the rack sources instead of sitting outside the design as a tolerated exception.

Lutron scenes stop running into video limits

A luxury residence never experiences AV in isolation. Video lives with lighting, shading, climate, and access. That is where fixed matrix architecture starts to feel dated, because the rest of the house is already scene-based.

A Lutron HomeWorks QSX system with Palladiom keypads, Palladiom shades, or Sivoia QS shades is built around state changes. DM NVX fits that way of thinking better than a legacy matrix does. An Entertain scene can call the correct video route, move the shades, shift Ketra or dimmed lighting to the right level, and leave the rack serving sources instead of explaining why a room cannot quite do what the scene implies.

The network is now part of the switcher

The baseline is still enterprise network discipline

DM NVX does not remove engineering. It moves engineering to the network. Crestron's current AV-over-IP network design guidance is blunt about the minimums: a 1 Gbps port for each endpoint, a nonblocking backplane, Layer 3 switching, IGMPv2 snooping, an IGMP querier, and fast-leave on the switching side.[8]

The same guidance says to assume 1 Gbps of bandwidth per DM NVX device across inter-switch uplinks and to design for nonblocking behavior all the way from edge to core.[8] This is where mediocre upgrades go wrong. The endpoints are priced correctly, but the network is treated like leftover infrastructure.

Crestron also spells out the multicast planning. On the DM-NVX-384 and DM-NVX-385 models, Multiview and H.264 each consume additional multicast addresses, so address planning has to be done early if the system is going to stay readable and expandable.[2][8]

2026 UniFi tools finally help at the AV layer

This is one reason UniFi Enterprise switching has become more relevant on residential AV jobs. Ubiquiti's April 15, 2026 EAV Switching announcement puts PTP-based shared timing, latency correction across hops, and sub-microsecond synchronization on the table, with Dante, AES67, and SMPTE ST 2110 compatibility called out directly.[5] The switch market is finally treating timing as an AV issue, not just an IT issue.

The June 25, 2026 UniFi Network 10.5 release matters for a different reason. Test & Confirm and automatic rollback reduce the risk of making live network changes on occupied properties.[6] When the AV system rides the network, safer change control is not an administrative detail. It is part of serviceability.

That is also why primary video paths should stay hardwired. A UniFi U7 Pro Outdoor can be the right tool for terrace coverage. It is not the backbone for distributed 4K video.

What to lock before you order gear

The best DM NVX upgrades are usually decided before the first endpoint ships.

Count destinations by behavior, not just by room. A media room display, a study monitor with USB-C, and an outdoor display are three different endpoint problems.
Decide where local inputs are truly needed. The DM-NVX-385 earns its keep when HDMI and USB-C at the room end prevent a pile of smaller fixes later.[3]
Separate the audio brief early. If one room needs full multichannel while the rest of the house wants stereo distribution, design that path up front.[2][3][4]
Reserve multicast space generously. On 384 and 385 deployments, Multiview and H.264 are not free from an address-planning standpoint.[2][8]
Size the switching fabric for simultaneous use, not average use. Crestron's guidance is explicit about uplink math.[8]
Decide how the control layer will be presented. A CP4-R with Crestron Home OS may be enough for many estates; some projects still need deeper custom logic.
Tie video behavior to Lutron HomeWorks QSX scenes only after the route logic is clearly defined.

When keeping the legacy matrix is the right call

Not every system needs to be converted. If the house has a stable source list, a stable destination count, no demand for local USB-C or display-app audio return, and no renovation pressure, a healthy DM matrix can still be the right answer to keep.

The turning point is usually when every small change starts creating a workaround: another local input box, another temporary audio extractor, another room that sort of shares the source if the rack stays patched a certain way. Once the system is being preserved by exceptions, the hardware is already telling you the architecture is done.

The decision

A legacy DM matrix is a piece of hardware. DM NVX is an architecture.

If the estate only needs the old hardware to keep doing the same job, keep it. If the owners want the house to route video, audio, USB, and control with the same freedom they already expect from Lutron scenes and modern network infrastructure, stop shopping matrix sizes and start designing the AV network instead.

Crestron DM NVX vs Legacy DM Matrix: 4K AV-over-IP