jlrouzies-fr/DLSS5-Feeder v1.16.0-beta.6

1.16.0-beta.6 - in-game panel: real cursor, modifiers, cast state to the helper

on GitHub

Careful of fake copies. The only download for this project is this releases page. Fake websites and lookalike GitHub repositories exist (#88, #115); this project has no executable installer. The standalone CAREFUL_FAKE_MALICIOUS_FEEDER.txt asset explains how to check what you downloaded. SHA-256 of DLSS5-Feeder-1.16.0-beta.6.zip: 0F9D9CDA66E05A0F21CCABBD778694B826C8C80402CC62046AEDEA6D157BDCAD. The genuine zips are scanned on VirusTotal, for instance the 1.16.0-beta.2 zip; look this release's hash up there yourself.

There is a one-command installer. Download Install-DLSS5Feeder.ps1, put it beside the game's executable, and run powershell.exe -ExecutionPolicy Bypass -File .\Install-DLSS5Feeder.ps1 -Prerelease. The standalone AUTOMATIC_INSTALLATION_AVAILABLE.txt asset explains every option.

Does not work with your game? I can't test every game reported in the issues: I don't own most of them, and I don't have the time to try them all. Read this README section. It gives a prompt for an AI coding agent that reads your logs against the source and opens a pull request if it finds a legitimate fix.

Four follow-up fixes on top of beta.5 for the in-game DLSS 5 panel — all of them from one reporter's retest of beta.5 on #118 — plus a new diagnostic for NGX failures that only happen inside the game (#120). Every binary identifies itself as 1.16.0-beta.6 and embeds release commit 2470210; check line 1 of dlss5-feed.log before reporting a retest.

⚠️ 32-bit games: update both halves. The IPC protocol moved to v10 in this build. A beta.5 add-on will not talk to this helper, or the other way round — install dlss5-feed.addon32 and host64\dlss5-feed-host64.exe from this zip. They refuse a mismatch rather than misparse it, and say so in the log.

Fixes

• #118 — the in-game panel could not be clicked in games that don't give ReShade their mouse messages. Measured in Castlevania: Lords of Shadow 2, which reads the mouse another way: ReShade's tracked cursor position never moved, so the panel never saw the pointer, while the real cursor moved freely the whole time. When ReShade's position has been frozen for 60 frames and the real cursor is somewhere else, the panel now follows the real cursor and the real mouse buttons instead — a game that hides moves from ReShade hides clicks too, so the cursor alone would have fixed nothing. It switches back the moment ReShade starts tracking again, and logs which one is in use.
  The wheel is still unavailable in those games, deliberately. It never reaches the game window at all, and the only thing that would see it — a raw-input sink — is registered per process, so claiming it would take away the game's own mouse input. That is not a trade worth a scroll wheel.
• #118 — the helper opened ReShade's overlay even when told not to. [INPUT] KeyOverlay=0,0,0,0 in host64\ReShade.ini parsed to 0, which the old guard read as "unparsable, keep the default", so Home was posted anyway. An explicit zero now means unbound: nothing is posted, and the helper window stops advertising a key that does nothing. Separately, with OptiScaler as the only neural consumer beside the helper, the startup key press is skipped on its own — OptiScaler's menu is its own window on Insert, not a ReShade page. Home by hand, and the add-on's "Toggle ReShade in host" button, still work.
• #118 — the panel's toggle key ignored modifiers. Binding Home also fired on Alt+Home and Alt+Shift+Home, so it collided with other tools' hotkeys. New config value cast_mods (1 = Alt, 2 = Ctrl, 4 = Shift, added together); hold the modifiers while you press the key under "Set key" and the combination is bound.
  This changes existing behaviour: the match is now exact in both directions, so with cast_mods=0 the bare key toggles the panel and the key pressed with a modifier held does not.
• #118 — a neural consumer drawing its own UI in the helper window couldn't tell whether the panel was on screen, and was reduced to tailing dlss5-feed.log to find out. The helper is now told (IPC v10), logs it with the destination rectangle and scale, and republishes it to its own window as RegisterWindowMessage("DLSS5_FEED_CAST") — wParam 1 while the panel is on screen, lParam the scale it is drawn at ×1000. Documented in the README.
• #120 — an NGX failure inside the game now names the module responsible. When NVSDK_NGX_D3D12_Init raised an exception, the log said only that "another module hooking NGX in this process faulted during init" — which named nothing and was a guess written before any case like this existed. It now records the module the fault came from and the chain of modules the call was made through. 0x80000003 is also called by its proper name, EXCEPTION_BREAKPOINT: an int 3 that was executed, not a memory fault, and not something this add-on ever issues.
  This is a diagnostic, not a fix. If your game fails this way, that new line is what the issue needs.

Also in this build: the README's "64-bit DirectX 9/11/12 → use renodx-dlss on its own" guidance now says what it always presumed — that the game calls DLSS itself. In a game that makes no DLSS calls there is nothing for that add-on to hook, and it waits forever with every counter at zero while looking healthy in its log (#120).

All beta.1 to beta.5 fixes are included.

Verified, and not

The two add-ons, the helper and both Vulkan layers build clean; the helper's self-test passes 300/300 evaluates here. The helper's side of the new IPC message, the overlay-key changes and the unbound-key case were exercised against the real helper over the pipe with a v10 test client.

The mouse and button fallback has not been run in an affected game — it needs a title that hides the mouse from ReShade, and I don't have one. If you were waiting on #118, please say whether the panel is genuinely clickable now.