Mouse cursor disappears on client when using virtual display

[AhmadRafiq90]

Environment

• Apollo version: 0.4.6
• Host OS: Windows 11 24H2
• Client: Moonlight 6.1.0 on Windows 10
• GPU: AMD Radeon RX 9060XT 16gb AMD Adrenalin 26.2.2
• Monitor setup: Single

Issue
After launching the virtual display in Apollo, the mouse cursor is not visible inside the Moonlight client. However, mouse movements from the client are still registered on the host system.

Steps to reproduce

1. Start Apollo on host
2. Connect via Moonlight
3. Launch virtual display
4. Begin streaming

Expected behavior
Cursor should be visible and usable inside the Moonlight stream.

Actual behavior
Cursor is invisible in the client window, but moving the mouse still moves the cursor on the host.

[ClassicOldSong]

Rare case and not yet figured out why.

Do you have any physical mouse connected to the host? If not try connect one.

[AhmadRafiq90]

Yes, I do have a physical mouse connected to the host

[ClassicOldSong]

Then try Ctrl+Alt+Shift+N on the client

If cursor still doesn't show up then it might be something else blocking the cursor.

[AlexHgC]

Hey, i have this issue since 1 or 2 months now, didn't had before. The mouse sometimes appear 0.1sec when moving it fast. I'm using virtual display. Ctrl Alt shift N does nothing. My gf have no cursor issue in his PC with the same client. If you need, i can try some things for you !

Edit : Tried to uninstall Apollo and virtual driver, fresh installing with default settings, just using virtual display, same issue.

[AlexHgC]

I have tried lot of things, nothing works, and seems to be a Windows Host issue. I'm running Windows 11 Experimental.

Tried to switch m'y client from Windows 11 to Kubuntu, same issue, no cursor showing, so not a Moonlight issue.

Tried to switch from virtual display to real display, same issue, not a virtual display issue.

Tried to switch from apollo to sunshine, still same issue. Not an apollo issue.

If you want, i can try some things and share logs to help.

[AhmadRafiq90]

I think the dev doesn't revisit old issues. Maybe try making a new issue with your findings

[ClassicOldSong]

3 month isn't old. And this is usually user issues, like they have some weird security apps installed or something, they can affect how hardware cursor being captured. Since both Sunshine and Apollo, virtual and real display all have this issue, it's not Apollo's problem.

[AlexHgC]

3 month isn't old. And this is usually user issues, like they have some weird security apps installed or something, they can affect how hardware cursor being captured. Since both Sunshine and Apollo, virtual and real display all have this issue, it's not Apollo's problem.

Well, i have some news. Sunshine and Apollo have the same issue, but it's working fine when using Vibepollo. So i think there is definitely something changed in this fork

[AhmadRafiq90]

Sorry, apologies if my comment was inappropriate. Also, it works for me with real display. The issue only occurs with virtual display.

[Jasonvdmeijden]

same issue any solutions?

[dimkKy]

I have the same issue while streaming win11 in hyperV to win11 host. For me it replicates on vibepollo as well. Radeon 780m

[kcoxeff-oss]

I'm having the same issue (I'm not on a virtual display or headless) the only thing that has worked is turning on magnify idk why it works

[awolden]

Adding a data point from an Nvidia setup. Since it's cross-vendor with OP's AMD, it reads more like Windows/capture than a GPU driver thing.

Setup: Apollo v0.4.7-alpha.1 (service mode), Windows 11 <build>, RTX 5070 Ti. Physical display is a Samsung G95SC at 5120x1440@120 HDR. Client is a Steam Deck at 1280x800x90. SudoVDA virtual display is the only active display during streams (ensure_only_display).

Symptom: cursor is completely absent from the stream. Input is fine, clicks land and position tracks. Occasionally it flickers in for a frame or two, then it's gone again.

What I've ruled out:

• MPO: OverlayTestMode=5 + reboot, no change
• Cursor suppression: EnableCursorSuppression=0, no change on the virtual display
• Driver state: driver restart (Win+Ctrl+Shift+B), quitting Apollo, disabling the SudoVDA adapter, no change
• Input source: moving the physical host mouse doesn't reveal it either, so it doesn't seem tied to the input source on the virtual display

Where it seems to point (capture API):

• Desktop Duplication (DDA): cursor missing on the SudoVDA display. DDA does capture the cursor fine on a physical display. I saw that when virtual display creation failed and Apollo fell back to capturing the monitor.
• WGC: cursor renders perfectly. I had to stop the service and run Apollo as a desktop process with WGC forced to test this. WGC black-screens under the service context, which is the usual reason forcing WGC looks broken (see Windows Graphics Capture in Service #1038).

My read, and I'm less sure on this part: on the SudoVDA output, DDA doesn't appear to be delivering the cursor, not in the frames and not as pointer metadata, while WGC composites it itself. That would line up with the IddCx hardware-cursor path: if the virtual display driver advertises hardware cursor support, the cursor may never land in the desktop image and DDA's pointer metadata stays empty. The VirtualDrivers VDD project ran into something in this class and added a hardware/software cursor toggle (VirtualDrivers/Virtual-Display-Driver#54, #25). I can't confirm that's the exact mechanism here, just that the symptoms match.

Possible directions, not sure which fits Apollo best:

• WGC-in-service (Windows Graphics Capture in Service #1038) makes it work
• a SudoVDA option to disable hardware cursor (filed: <link>)
• an Apollo-side fallback that composites via GetCursorInfo when DDA pointer metadata is absent

Workaround if anyone's stuck: stop the Apollo service, run sunshine.exe directly as your logged-in user, force WGC. Cursor works. Tradeoff is you lose login/UAC screen capture.

One more note: on an extended desktop (virtual + physical both active) there's a separate thing where client-injected input triggers normal Windows cursor suppression (cursor hides for injected/touch-like input, comes back on physical mouse movement). Easy to conflate with this bug, but I think it's unrelated.