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.txtasset explains how to check what you downloaded. SHA-256 ofDLSS5-Feeder-1.17.0-beta.1.zip:6827E3CD076090C44D39391282A0E69CA63D366EC714EA3AA54F9BCCF313FFCE. 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 runpowershell.exe -ExecutionPolicy Bypass -File .\Install-DLSS5Feeder.ps1 -Prerelease. The standaloneAUTOMATIC_INSTALLATION_AVAILABLE.txtasset 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.
wilsjo2's OptiScaler-DLSSNR-PreSR-Multipass is now a supported neural consumer (#126). It sits next to Dagherbou's OptiScaler_DLSSNR, which still works. Nothing else changed since 1.16.0-beta.7. Every binary embeds release commit 07a9242, and the add-ons identify themselves as 1.17.0-beta.1; check line 1 of dlss5-feed.log before reporting a retest.
32-bit games: install
dlss5-feed.addon32andhost64\dlss5-feed-host64.exefrom this zip, as always. The IPC protocol is still v10, so beta.7 and beta.1 halves do talk to each other. But the OptiScaler detection lives in the helper, so the helper must be the new one.
What changed
- #126 — wilsjo2's OptiScaler fork was reported as "not the DLSS-NR fork". The feeder recognised a DLSS-NR build of OptiScaler by its forwarder DLL's name,
nvngx.dll_dlssnr.dll, inside the OptiScaler DLL. wilsjo2's fork has had no forwarder since v0.8.1: it reaches the neural model through the NVIDIA driver instead. The add-ons, the helper, the installer and the verifier now recognise a DLSS-NR build by the model's own file name, which both forks carry and stock OptiScaler does not. The log says which generation is loaded:direct-runtime build (wilsjo2/...)orforwarder build (Dagherbou/...). - The neural-model check waits for wilsjo2's build. wilsjo2's build creates the model on its own schedule, so the feeder now keeps looking for up to 120 frames before it reports the model missing. Dagherbou's build has it by the second frame, and so did wilsjo2's v0.8.8 on the test rig.
- New warnings for three OptiScaler settings that silently stop the pass:
[DlssNr] FinishedPicture=true(wilsjo2) moves the neural pass to the game's Present. For a 32-bit game that is the helper's own window, which the game never sees.[ProcessFilter] TargetProcessNamenaming another exe puts OptiScaler into pass-through: it loads and does nothing. AnOptiScaler.inicopied from another game brings that game's name along.- A leftover
nvngx.dll_dlssnr.dllbeside wilsjo2's build is never loaded. The fork's own notes say to delete it when upgrading.
- Installer: choosing OptiScaler now asks which fork, and wilsjo2 is the default.
-OptiScalerFork wilsjo2|Dagherbouanswers in advance. By default it downloads the fork's latest stable release (wilsjo2 v0.8.3 today);-Prereleasetakes the newest one (v0.8.8). It skips the-rtx40-mfgvariant, leaves wilsjo2'sdocs\tree out of the game folder, and disables a leftover forwarder. - Verifier: knows both generations, requires the forwarder only for Dagherbou's build, checks the process filter, and reads wilsjo2's
OptiScaler.loglines.
Installing wilsjo2's fork by hand is in the README under Alternative: OptiScaler DLSS-NR. Multipass is Passes in OptiScaler's neural-rendering section (Insert).
Verified, and not
Both add-ons and the helper build without warnings; the Vulkan layers and the shader are unchanged from beta.7. The helper's self-test passes 300/300 evaluates at 640x360 on driver 616.92:
| OptiScaler build | Evaluates | Neural model | GPU ms/frame |
|---|---|---|---|
| wilsjo2 v0.8.8 | 300/300 | created by frame 2 | 3.5 |
| Dagherbou v0.2.0 | 300/300 | created by frame 2 | 3.0 |
wilsjo2's v0.8.8 was also reported working in Fable Anniversary, a 32-bit game running through the helper.
Not run here: wilsjo2's fork in a 64-bit game, and wilsjo2's v0.8.3, the release the installer downloads without -Prerelease.