fix(VR): exponential height fog stereo mismatch

[YtzyFvra]

Used Stereo::ConvertFromStereoUV to fix left/right eye mismatch and ghosting when Exponential Height Fog is enabled in VR.

Summary by CodeRabbit

• Bug Fixes
  • Improved stereo/VR rendering calculations to correctly handle per-eye view positioning, enhancing compatibility and accuracy for stereo displays.

[coderabbitai]

📝 Walkthrough

Walkthrough

A shader modification that improves stereo rendering compatibility by converting screen texture coordinates to mono-space UV values before calculating world-space positions in the height fog calculation path of ISSAO compositing.

Changes

Cohort / File(s)	Summary
Stereo UV Conversion package/Shaders/ISSAOComposite.hlsl	In the EXP_HEIGHT_FOG path, added stereo-to-mono UV conversion using Stereo::ConvertFromStereoUV() before computing positionWS, introducing an intermediate monoUV variable for proper stereo rendering support.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~3 minutes

Possibly related PRs

• fix(vr): improve stereo UV handling #1899 — Directly relates to stereo UV handling refactoring and ConvertFromStereoUV usage patterns.
• fix: move ResolveMonoUVForEye outside VR guard #1906 — Touches the same stereo/mono UV conversion functionality across VR code paths.

Suggested reviewers

• alandtse
• doodlum

Poem

🐰 With stereo eyes we hop so high,
Converting UVs toward the sky,
No double vision, clean and bright—
Each eye sees true, the world looks right! ✨

🚥 Pre-merge checks | ✅ 3

✅ Passed checks (3 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Title check	✅ Passed	The title accurately and concisely summarizes the main fix: correcting a stereo mismatch issue in exponential height fog for VR.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[github-actions]

No actionable suggestions for changed features.

[Copilot]

Pull request overview

This PR fixes a stereo mismatch bug in VR mode where the Exponential Height Fog feature in ISSAOComposite.hlsl was incorrectly using the raw stereo UV (spanning [0,0.5] for left eye and [0.5,1.0] for right eye) to reconstruct a world-space position, causing the wrong clip-space X coordinate for the right eye and producing left/right eye mismatch and ghosting artifacts.

Changes:

• Introduces Stereo::ConvertFromStereoUV to convert the packed stereo texture coordinate to a per-eye normalized [0,1] UV before world-space position reconstruction.

[github-actions]

✅ A pre-release build is available for this PR:
Download

[jiayev]

@alandtse could u review this, i got no vr

[alandtse]

Oohh nice. Got a test location? I thought I saw somewhere with this issue but never hunted it down.

[alandtse]

@coderabbitai is there any other places in the shader code that may need this fix?

[coderabbitai]

Tip

For best results, initiate chat on the files or code changes.

🧩 Analysis chain

🏁 Script executed:

#!/bin/bash
# 1. Find all usages of EXP_HEIGHT_FOG in shader files
echo "=== EXP_HEIGHT_FOG usages ==="
rg -n "EXP_HEIGHT_FOG" --type hlsl -C 5

# 2. Find all existing correct usages of Stereo::ConvertFromStereoUV
echo ""
echo "=== Stereo::ConvertFromStereoUV usages ==="
rg -n "ConvertFromStereoUV" --type hlsl -C 5

# 3. Find world-space position reconstruction patterns that use raw TexCoord (potentially missing conversion)
echo ""
echo "=== positionWS reconstruction with TexCoord (may need stereo fix) ==="
rg -n "positionWS" --type hlsl -C 5

# 4. Find inverse projection / screen-to-world using UV without ConvertFromStereoUV in the same block
echo ""
echo "=== InvVP or InvProj usage with TexCoord ==="
rg -nP "InvVP|InvViewProj|InvProjection" --type hlsl -C 5

# 5. Look for eyeIndex usage in shaders (VR-aware paths that might need the fix)
echo ""
echo "=== eyeIndex usage in shaders ==="
rg -n "eyeIndex" --type hlsl -l

Length of output: 370

---

🏁 Script executed:

#!/bin/bash
# 1. Find all usages of EXP_HEIGHT_FOG in shader files
echo "=== EXP_HEIGHT_FOG usages ==="
rg -n "EXP_HEIGHT_FOG" -g "*.hlsl" -C 5

# 2. Find all existing correct usages of Stereo::ConvertFromStereoUV
echo ""
echo "=== Stereo::ConvertFromStereoUV usages ==="
rg -n "ConvertFromStereoUV" -g "*.hlsl" -C 5

# 3. Find world-space position reconstruction patterns that use raw TexCoord
echo ""
echo "=== positionWS reconstruction (all occurrences) ==="
rg -n "positionWS" -g "*.hlsl" -C 4

# 4. Find eyeIndex usage - list files first
echo ""
echo "=== Files with eyeIndex ==="
rg -n "eyeIndex" -g "*.hlsl" -l

# 5. InvVP / InvViewProj usage
echo ""
echo "=== InvVP or InvViewProj usage ==="
rg -nP "InvVP|InvViewProj|InvProjection" -g "*.hlsl" -C 4

Length of output: 61098

---

Good findings so far — now let me dig into two specific candidates before giving a definitive answer.

---

🏁 Script executed:

#!/bin/bash
# Check how worldPosition is defined/used in Effect.hlsl around lines 574 and 649
echo "=== Effect.hlsl worldPosition context (around lines 540-660) ==="
sed -n '540,660p' package/Shaders/Effect.hlsl

# Check ISReflectionsRayTracing.hlsl for finalSampleUV context
echo ""
echo "=== ISReflectionsRayTracing.hlsl - full relevant section around lines 100-170 ==="
sed -n '60,180p' package/Shaders/ISReflectionsRayTracing.hlsl

Length of output: 9441

---

@alandtse: I did a thorough search across all shader files. ISSAOComposite.hlsl appears to be the only location that needed this fix. Here's the breakdown:

---

EXP_HEIGHT_FOG sites — all safe ✅

Every other shader that uses EXP_HEIGHT_FOG passes a vertex-interpolated world position (input.WorldPosition.xyz / input.WPosition.xyz) directly to the fog functions — no screen-space UV reconstruction involved, so no stereo mismatch is possible:

File	World position source
Lighting.hlsl	input.WorldPosition.xyz (interpolated)
Water.hlsl	input.WPosition.xyz (interpolated)
RunGrass.hlsl	input.WorldPosition.xyz (interpolated)
DistantTree.hlsl	input.WorldPosition.xyz (interpolated)
Effect.hlsl	worldPosition (function parameter, caller passes interpolated vertex data)

---

Screen-space positionWS-from-UV reconstruction sites — all already correct ✅

These are the shaders that, like ISSAOComposite.hlsl, reconstruct world position from depth + UV. They all correctly call ConvertFromStereoUV (or the equivalent) before building the NDC position:

File	Stereo UV handling
DeferredCompositeCS.hlsl Line 93	uv = Stereo::ConvertFromStereoUV(uv, eyeIndex) before reconstruction
Particle.hlsl Line 292	uv = Stereo::ConvertFromStereoUV(...) before reconstruction
ISVolumetricLightingGenerateCS.hlsl Line 77	Stereo::ConvertFromStereoUV(normalizedCoordinates, eyeIndex)
Utility.hlsl Line 605	Stereo::ConvertFromStereoUV(depthUv * ..., eyeIndex)
ISReflectionsRayTracing.hlsl Lines 120–122	finalSampleUV is populated by Stereo::ResolveMonoUVForEye(...), which already outputs a per-eye mono UV paired with finalEyeIndex — the mul(CameraViewProjInverse[finalEyeIndex], ...) is correct

---

Conclusion: ISSAOComposite.hlsl was uniquely problematic because it's a post-process pass that reconstructs world position from depth + screen UV, but unlike the other post-process shaders it was not calling ConvertFromStereoUV first. This PR is a complete, self-contained fix.

---

🧠 Learnings used

Learnt from: ThePagi
Repo: doodlum/skyrim-community-shaders PR: 1369
File: src/Features/SnowCover.cpp:515-515
Timestamp: 2025-10-02T14:20:33.454Z
Learning: In the Community Shaders codebase (skyrim-community-shaders repository), hardcoded shader resource slot numbers are used directly in code rather than being defined as named constants. This is the established convention and should not be flagged as an issue.

---

Learnt from: Dlizzio
Repo: doodlum/skyrim-community-shaders PR: 1950
File: package/Shaders/Lighting.hlsl:2008-2012
Timestamp: 2026-03-07T00:53:16.539Z
Learning: In doodlum/skyrim-community-shaders PR `#1950`, the maintainer (Dlizzio) confirmed that per-axis normal reorientation for triplanar-projected normals/detail in package/Shaders/Lighting.hlsl is not required for their use case; the current approach intentionally mixes samples without axis-specific swizzle/sign correction.

---

Learnt from: alandtse
Repo: doodlum/skyrim-community-shaders PR: 1858
File: src/FrameAnnotations.cpp:1022-1024
Timestamp: 2026-02-21T22:01:15.534Z
Learning: In the skyrim-community-shaders repository, for the `REL::Relocate` function, when the VR offset is not explicitly provided as a third argument, it defaults to using the first parameter (SE offset) as the VR offset. This means `REL::Relocate(0x37f, 0)` will use `0x37f` for both SE and VR, with AE getting `0`.

---

Learnt from: alandtse
Repo: doodlum/skyrim-community-shaders PR: 0
File: :0-0
Timestamp: 2025-07-05T05:20:45.823Z
Learning: In the skyrim-community-shaders repository, file deletion error handling improvements that replace existence checks and try-catch blocks with std::filesystem::remove error-code-based approaches are considered bug fixes rather than refactoring, as they address inadequate error handling and misleading log messages.

---

Learnt from: alandtse
Repo: doodlum/skyrim-community-shaders PR: 1880
File: src/Features/VR/InSceneOverlay.cpp:603-604
Timestamp: 2026-02-15T08:13:49.292Z
Learning: The skyrim-community-shaders repository uses OpenVR SDK version 1.0.10, where IVRCompositor::Submit is at vtable index 5 (0-based). Do not suggest changing this to index 6, as that is only correct for newer OpenVR versions (2.x).

[jiayev]

Oohh nice. Got a test location? I thought I saw somewhere with this issue but never hunted it down.

not sure, maybe turn up the fog density and look around?

[alandtse]

Confirmed. Thanks!