Columbarium operators split into two camps on online memorial services: one camp believes it is the future, the other believes it is a gimmick.

Both are right, because this service does not solve "paying respects". It solves "not being able to come". Working away from home, ill health, bad weather, schedules that never line up — those are its real user scenarios. Once you see that, you have a criterion for whether to build it and to what extent.
The first is an online memorial hall. A page with photographs, a life story, messages. Lowest cost, almost pure software. Its function is to gather memory into one place, but it has no physical anchor — it exists only while the phone is open, and disappears when it closes.
The second is audiovisual within the hall. Big screens, projection, background music. This is the form most likely to become noise, because a columbarium is not an exhibition hall; rolling imagery interrupts everyone's quiet.
The third is livestreamed proxy visitation. Staff visit on the family's behalf and the family watches online. This is the heaviest of the three: it needs network, camera positions, procedure, privacy rules, and a site willing to be filmed.
Has this investment reduced the regret of "I could not come that time"?
If a family could previously manage three visits a year, and with the online service it becomes two online plus one in person, that is effective. If they would have come three times anyway, the investment is decoration.
By that criterion the priority order is clear: first build an online memorial hall (so memory has somewhere to live), then proxy visitation (so absence has a substitute), and only then audiovisual inside the hall (so the experience is richer). Do it in the reverse order and you will spend the most money for the least effect.
This is the most overlooked link.
A purely online memorial hall will be used by the young and not by the elders — not because they cannot tap, but because they do not know where to tap. So the effective approach is to turn the QR code into a physical object: a 30 × 30 mm etched plaque mounted beside the niche or on the memorial wall. A family member stands at the niche, scans once, and is inside.
A QR plaque on a niche door panel costs under ten renminbi, but it pins something virtual into reality. Every digital effort in a columbarium has to land on small objects like this, or it is built on air.
First, the network in the columbarium. Thick walls and metal frames attenuate signal badly. Test before you build: mobile signal inside the hall should be no worse than -95 dBm, or provide a wireless access point per two hundred niches If the signal does not meet that, the stream will stutter.
Second, a camera position in front of the niche. You need a fixed position that does not occupy the aisle. I suggest reserving a 1.5 m × 1.5 m spot at the end of each zone, with power and fill light (3000 K — do not use cold white). If a niche aisle is only 1.2 m wide the camera can only stand in the aisle, and then it will certainly block people.
Third, fill light. Columbarium illuminance is low, and a phone camera will produce grey, flat footage. Add one 3000 K soft light and the image changes immediately, without glare.

The first: never film anyone else's niche. A livestream camera must be confined to the range of a single niche. Another family's name appearing in frame is a management incident.
The second: no "AI resurrection". Generating conversations from a deceased person's photographs and voice fails ethically, and it violates the most basic expectation a family has — they want remembrance, not impersonation. What a columbarium can do is organise real images, real recordings and real words. Not manufacture a new version of someone.
If you must have it, do only two things: wayfinding (which zone is where, what ceremonies are on today) and ritual notices (what starts at what time, what the sequence is).
Hold volume below 55 dB, zone the playback, and make it independently switchable. The most precious thing in a columbarium is quiet. No audiovisual should become background noise.
Q: Do we have to develop the online memorial hall ourselves?
No. There are plenty of mature templates. The point is not features but content maintenance — who uploads, who moderates messages, how long the deceased's information is retained. Without named owners for those three, even the best system becomes an empty page within two years.
Q: What if the elders cannot use it?
Do two things: make the QR plaque a physical object at the niche, and print an A5 diagram for the entrance (three steps, done). They do not need a tutorial. They need to be able to complete it without asking anyone.
Q: Is it reasonable to charge for livestreamed visitation?
Yes, but state the service content precisely: who goes, what they do, how long it takes, how long the video is retained. Columbarium services suffer most from vagueness, and vagueness ends up as dispute.

Online memorial service is not an upgrade to a columbarium niche It is a patch — a patch over the gap called "I could not come". Patched well, the distance between a family and their dead shortens. Patched badly, it is a television hung in a funeral hall. Before you build it, ask one question: whose problem does it solve? If you can answer, build it. If you cannot, start with that small QR plaque.