Beginner-Friendly Tutorials Aug 2, 2026

Your First Unity Surface Cache GI Preview Evening - URP Keep Hold 2026

Unity Surface Cache GI preview evening for August 2026 - URP enable steps, diffuse-only honesty, temporal stability gates, Monday ritual, and keep/hold vs lightmaps before 6.7.

By GamineAI Team

Your First Unity Surface Cache GI Preview Evening - URP Keep Hold 2026

up movie pixel art thumbnail for Unity Surface Cache GI URP preview evening

If you are searching unity surface cache gi in August 2026, you do not need another “Unity finally has Lumen” rumor thread—and you do not need to confuse Unite Seoul’s Unity 7 roadmap GI marketing with tonight’s alpha. You need a Monday path: open a preview-capable Unity 6.x URP project, enable Surface Cache GI (SCGI) per Unity’s Getting Started guidance, light one small dynamic scene, write honest notes on diffuse-only limits and temporal stability, then decide keep (watch/preview) / hold (ship bake) / rewrite (wrong pipeline) before anyone deletes lightmaps on a production branch because Discord said “realtime GI.”

Surface Cache GI is Unity’s URP realtime indirect diffuse lighting path. Unity’s lighting team invited public preview testing on Discussions ahead of a 6.7 LTS expectation (not a guarantee). GameFromScratch summarizes the design: patches in position+direction space, hardware ray tracing when available, otherwise compute software RT. Weeks after Unite Seoul, Surface Cache GI still shows up in partner decks as a Unity 7 graphics pillar—roadmap language is not “ship tonight on alpha.”

Who this is for: creators evaluating whether to leave baked GI; TDs comparing SCGI to Lumen / Godot SDFGI honesty; leads who need diligence language for preview vs production lighting after Unity 7 / UE 6 calendar noise.
Time: about 60–90 minutes for first enable + smoke + receipt.
Honest limit: preview/alpha maturity; diffuse-only today (pair reflection probes/SSR for specular); temporal flicker reported by community; not a drop-in “delete all lightmaps and ship Steam tomorrow” button.

Non-repetition: This is not the Unity 7 roadmap keep/hold (engine pillars / MCP / CLI). It is not Fast Enter Play Mode 6.6 (Play mode statics). It is not Unity CLI first session. It is not Unreal Lumen Lite UE 5.8 preview (different engine). Same Unity family—URP Surface Cache GI first preview evening.

Why this matters now (August 2026)

  1. Unity publicly invited SCGI testing—and the thread is still the feedback lane. The Surface Cache GI Preview discussion remains the place to learn limits before a hoped-for 6.7 readiness—not invent marketing claims for Steam capsules.
  2. Unity 7 marketing raised the GI stakes; August diligence still separates roadmap from alpha. Unite Seoul lists Surface Cache GI under breakthrough graphics (Unity News). After the on-site Unity 7 keep/hold expand, partners will ask “are you on realtime GI?”—answer with a receipt, not a trailer still.
  3. URP finally has a first-class realtime diffuse GI story. That closes a long SERP gap vs Unreal Lumen and Godot SDFGI—but only if you respect diffuse-only and stability tradeoffs (GameFromScratch).
  4. Engine calendar noise makes FOMO worse this month. UE 6 / Unity 7 diligence waves make it easy to “upgrade lighting” on the wrong branch. SCGI stays a throwaway preview until P1–P6 latch.
  5. SERP is still forum + announce heavy. Indies need a first-evening enable path, keep/hold table, bake fallback, Monday ritual, and receipt—not another “patches are 5D” restatement alone.

Direct answer: enable SCGI on preview/scgi-smoke only; bake/APV remains ship lighting until version + stability gates pass; Unity 7 GI messaging is tracked separately.

Parent Unity search demand stays enormous. The preview thread’s large reply volume shows real community pressure for honest enablement guides.

What Surface Cache GI is (and is not)

Surface What it does Use tonight?
Surface Cache GI (SCGI) Realtime indirect diffuse GI for URP via world-space patches Preview smoke on a throwaway branch
Hardware RT path Uses HW ray tracing when available Prefer for quality; document GPU
Software RT fallback Compute-based RT when HW RT missing Expect cost/quality tradeoffs
Baked lightmaps / APV Production-proven static/dynamic bake paths Keep as ship default until SCGI green
Lumen / SDFGI Other engines’ realtime GI Comparison only—not Unity APIs
Unity 7 marketing GI Future product messaging Do not confuse with tonight’s alpha

Unity frames SCGI for dynamic lights/geometry, procedural content, time-of-day, and avoiding long bake waits—while still asking for community bug reports before release.Preview thread

Keep / hold / rewrite — SCGI preview

Decision Keep Hold Rewrite
Preview branch Throwaway URP project; define + feature enabled; smoke filmed Waiting for 6.7 LTS packaging Enabling on main ship branch first
Ship lighting Bake/APV still owns release SCGI looks good in editor once Deleting lightmaps because Discord said “realtime”
Art direction Diffuse GI + probes/SSR plan written Need specular GI as hero look Promising Lumen parity on URP tonight
Hardware targets Deck/mobile cost measured Only tested on RTX desktop Claiming “runs everywhere” without swRT notes

Print the table. Circle one cell.

Monday ritual (25 minutes after standup)

  1. Paste the Discord checklist below into the channel.
  2. Confirm production lighting owner is still bake/APV (or your written lock)—no silent SCGI on main.
  3. Re-check Unity preview version string vs last week’s receipt (preview drops move).
  4. Separate SCGI smoke from Unity 7 roadmap, FEPM CoreCLR, and Lumen Lite UE 5.8.
  5. Assign preview branch owner or leave Hold in receipt notes.
  6. File gates P1–P6 if not already latched for this planning window.

Creator Discord paste (shareable)

Surface Cache GI (URP) — August keep/hold:
• Preview branch only (preview/scgi-smoke) — never main/ship
• Diffuse GI + probes/SSR plan — not “Lumen parity”
• Ship lighting stays bake/APV until stability + version gates
• Temporal flicker = hold ship claims, not “fix in post”
• Partner line: “evaluating SCGI preview; bake owns release”
Cite Unity Discussions Surface Cache GI Preview + Unity 7 news — ignore “delete lightmaps” thumbs

Beginner path - one evening SCGI smoke

Prerequisites

  • Unity 6.x build that includes Surface Cache GI preview bits (confirm live Getting Started / package notes—versions move).
  • A URP project you can break (not your Steam ship branch).
  • GPU notes: does this machine have hardware RT?
  • 60–90 minutes.

Step 1 - Branch and snapshot

  1. Create git branch preview/scgi-smoke (or duplicate the project).
  2. Screenshot current lighting setup (lightmaps on/off, APV, probes).
  3. Write baseline_fps on a known camera path if you have a simple HUD timer—optional but diligence-friendly.

Step 2 - Enable Surface Cache GI (follow live docs)

Unity’s Getting Started (linked from the preview thread) is authoritative. Typical pattern reported by coverage:

  1. Add the SURFACE_CACHE (or live-documented) scripting define in Player/Project Settings.
  2. Add the Surface Cache Global Illumination renderer feature to your Universal Renderer Data.
  3. Configure volume / radius / quality knobs per Getting Started—start conservative.
  4. Restart Editor if defines require it.

If menus differ, stop and quote the live doc—do not invent menu paths from memory.

Step 3 - Build a tiny dynamic light scene

Minimum scene:

  1. Closed room with a few colored walls.
  2. One movable point light or day/night directional change.
  3. One dynamic prop you can move at runtime (or a simple transform script).
  4. A camera fly that shows indirect bounce on walls.

Observe:

  • Does indirect light update when the light moves?
  • Flicker / ghosting when the camera moves?
  • Cost spike when turning SCGI off vs on?

Step 4 - Honesty checks (diffuse + stability)

Check Pass Fail
Indirect diffuse responds to light change Visible bounce update No change → feature not active
Specular expectations You planned probes/SSR “Looks dull” without specular plan
Temporal stability Acceptable for preview Unshippable flicker → hold ship
Mobile/Deck target Measured or explicitly N/A Assumed free

Community reports often praise performance while flagging temporal stability / flicker—treat that as a gate, not a surprise.Discussions

Step 5 - Keep / hold decision sentence

Examples:

  • Keep (preview): “SCGI enabled on preview branch; bake remains ship lighting until 6.7 + stability green.”
  • Hold: “Flicker/cost fails targets; revisit after next preview drop.”
  • Rewrite: “HDRP/custom pipeline—wrong article path; evaluate differently.”

Common beginner mistakes

  • Enabling SCGI on the release branch first.
  • Deleting lightmaps after one pretty screenshot.
  • Comparing URP SCGI to Lumen specular quality without probes.
  • Ignoring software RT cost on Deck/iGPU.
  • Quoting Unity 7 marketing as if preview is LTS.
  • Skipping a receipt (“it worked on my machine”).

Developer path - gates, comparisons, receipt

P1–P6 keep/hold gates

Gate Question Pass criteria
P1 Is preview source dated (Discussions/Getting Started)? URLs + Unity version recorded
P2 Is enable path completed on a throwaway branch? Define + renderer feature screenshots
P3 Did dynamic light smoke show diffuse GI? Notes + optional video
P4 Are diffuse-only + stability limits written? Diligence sentences filed
P5 Is bake/APV fallback still intact for ship? Branch policy: SCGI ≠ default ship
P6 Is there a receipt? unity_surface_cache_gi_preview_receipt_v1.json filed

Latch scgi_preview_ok only when P1–P6 pass.

Sample receipt JSON

{
  "schema": "unity_surface_cache_gi_preview_receipt_v1",
  "date": "2026-08-02",
  "unity_version": "6000.x.y-preview-or-alpha",
  "pipeline": "URP",
  "branch": "preview/scgi-smoke",
  "august_framing": "post_unite_seoul_unity7_gi_marketing_separated",
  "sources": [
    "https://discussions.unity.com/t/surface-cache-gi-preview/1720494",
    "https://unity.com/news/unity-7-roadmap-revealed-at-unite-seoul"
  ],
  "define_enabled": true,
  "renderer_feature_added": true,
  "hw_ray_tracing": true,
  "dynamic_light_smoke": true,
  "temporal_stability": "acceptable_preview|fail",
  "diffuse_only_acknowledged": true,
  "specular_plan": "reflection_probes",
  "ship_lighting_owner": "baked_lightmaps",
  "decision": "keep_preview_hold_ship",
  "gates": {
    "P1_sources": true,
    "P2_enable": true,
    "P3_smoke": true,
    "P4_limits": true,
    "P5_fallback": true,
    "P6_receipt": true
  },
  "scgi_preview_ok": true,
  "operator": "session-owner-handle"
}

Comparison table - SCGI vs cousins

System Engine Strength Indie watch-out
Surface Cache GI Unity URP Realtime diffuse, broad-device intent Preview; diffuse-only; flicker risk
Baked lightmaps / APV Unity Ship-proven Bake time; static assumptions
Lumen Unreal High-end realtime GI Different engine cost model
SDFGI Godot Open-source realtime GI Different art pipeline
Unity 7 GI messaging Roadmap Future pillar Not tonight’s alpha contract
Lumen Lite (UE 5.8) Unreal preview Epic-side realtime GI smoke Different engine—see Lumen Lite evening

Workflow - August engine-calendar FOMO hygiene

August 2026 is loud: Unity 7 diligence, UE 6 roadmap chatter, and preview lighting threads all show up in the same Slack. Hygiene rules:

  1. One evening, one engine. Do not mix SCGI enable with Unreal Lumen Lite smoke in the same branch night.
  2. Roadmap ≠ preview. Unity 7 GI callouts do not authorize lightmap deletion on a Steam demo branch.
  3. Write the ship owner. If bake owner is blank, SCGI is unauthorized on main—full stop.
  4. Partner decks get the receipt date, not a Discord screenshot of a pretty bounce.

Workflow - when to keep bake as ship lighting

Keep bake/APV as the release path when any are true:

  1. You need specular GI as a marketing hero without a probes/SSR plan.
  2. Temporal stability fails your camera language (cutscenes, slow pans).
  3. Target devices lack HW RT and swRT cost breaks frame budget.
  4. Certification/partner builds forbid preview packages.
  5. Art leads cannot tolerate patch cascade artifacts.

SCGI can still be a preview R&D win on a side branch.

Workflow - Unity 7 vs 6.7 expectation hygiene

Unite Seoul markets Surface Cache GI under Unity 7 graphics.Unity News Discussions aim SCGI readiness around 6.7 expectations without hard guarantees.Dynamic GI roadmap

Partner language:

We are evaluating Surface Cache GI on a preview branch. Ship lighting remains baked until version X and stability gates pass. Unity 7 roadmap GI is tracked separately from this preview.

Pair with your Unity 7 roadmap playbook for MCP/CLI pillars—not lighting smoke.

Workflow - performance notes

Community feedback often calls SCGI surprisingly cheap on capable GPUs while complaining about stability more than raw ms.Preview thread Your job:

  1. Measure your scenes, not Twitter ms quotes.
  2. Test with HW RT on and off if possible.
  3. Log Deck/mobile separately—or mark N/A.
  4. Do not ship “1.5ms everywhere” claims from someone else’s APU test.

Company / diligence language (paste-ready)

Q: Are we shipping Surface Cache GI?
A: Preview decision: keep_preview_hold_ship / hold / rewrite. Unity version: . Branch: . Receipt attached.

Q: Is this the same as Lumen?
A: No. SCGI is Unity URP realtime diffuse GI with different design goals (reach/performance). Specular needs probes/SSR or later Unity work.

Q: Does Unity 7 announcement mean we can delete lightmaps?
A: No. Roadmap messaging ≠ preview maturity. Bake remains ship default until gates pass.

Q: What is the risk?
A: Alpha/preview bugs, temporal instability, diffuse-only look, device cost variance. Mitigated by throwaway branch + bake fallback.

Sample evening timeline (strict)

Minute Task Exit criteria
0–10 Open preview thread + Getting Started; note Unity version P1 notes
10–25 Branch + baseline lighting screenshot Safe rollback
25–45 Enable define + renderer feature P2 screenshots
45–65 Dynamic light smoke + stability notes P3/P4
65–75 Confirm bake fallback still on ship policy P5
75–90 Receipt JSON + partner Q&A scgi_preview_ok only if green

Troubleshooting matrix (night-one)

Symptom Likely cause Next action
No menu / feature Wrong Unity version / missing package Re-read Getting Started for your build
No indirect change Feature not active / wrong renderer Verify feature on active URP renderer
Looks flat Diffuse-only expectation Add probes/SSR; don’t claim Lumen
Heavy flicker Alpha temporal issues Hold ship; file feedback on Discussions
Deck tanks swRT cost Hold target or reduce quality knobs
Team enables on main Process failure Revert; enforce branch policy

Studio ownership checklist

  1. Preview branch owner: ___
  2. Lighting TD: ___
  3. Ship bake owner: ___
  4. Device matrix owner (PC/Deck/mobile): ___
  5. Receipt path: receipts/unity_surface_cache_gi_preview_receipt_v1.json

Quality knobs - what to turn first (preview)

Unity exposes quality/performance tradeoffs (patch size, cascades, sample counts—confirm live Getting Started names). Night-one rule:

  1. Start medium/default knobs—do not max everything on first enable.
  2. If flicker dominates, reduce aggressiveness before declaring SCGI “broken.”
  3. If cost dominates, lower far cascades / radius before blaming URP itself.
  4. Log every knob change in the receipt—otherwise next week’s “it regressed” is unfalsifiable.

Do not publish a Steam trailer claiming “fully dynamic GI” until your art lead signs the stills.

Art lead checklist (paste)

  1. Hero still with SCGI on: ___
  2. Same camera with bake: ___
  3. Specular plan (probes/SSR): ___
  4. Acceptable flicker on slow pan: yes/no
  5. Color grading still matches brand bible: yes/no
  6. Decision: keep_preview / hold_ship / rewrite

Any “no” on ship-critical rows → bake stays king.

HDRP and custom pipelines

This evening is URP Surface Cache GI. If your project is HDRP or a custom SRP:

  1. Do not force URP conversion tonight for FOMO.
  2. File decision: rewrite with pipeline name.
  3. Track Unity’s live docs for whether SCGI expands—do not invent HDRP parity.

Sample Discord brief (paste, then edit)

We enabled Surface Cache GI on branch preview/scgi-smoke only (August 2026). It is Unity URP realtime diffuse GI in preview—not Lumen, not a lightmap deletion party, not Unity 7 “shipped.” Ship lighting stays baked until stability + version gates pass. Owner: ___. Receipt filed.

Device matrix worksheet

Target HW RT? Result Notes
Dev desktop pass/fail
Min-spec PC pass/fail
Steam Deck pass/fail/N/A
Mid mobile pass/fail/N/A

Blank cells that matter for your SKU → hold ship claims.

After tonight - what keep unlocks next sprint

  1. Weekly re-test on newest preview drop.
  2. One hero scene comparison: bake vs SCGI stills for art lead.
  3. Optional feedback post on Unity Discussions with repro steps.
  4. Keep Unity 7 / FEPM / CLI workstreams separate in the roadmap.

Feedback hygiene - how to report upstream without drama

Unity asked for preview feedback. Useful reports include:

  1. Exact Unity version string and URP package versions.
  2. HW RT on/off.
  3. Minimal repro scene (strip art).
  4. Video of flicker or incorrect bounce.
  5. What you expected vs what happened in one sentence.

Useless reports: “looks bad,” “not Lumen,” “fix my ship build.” Preview bandwidth is finite—spend it on falsifiable bugs.

When SCGI is the wrong evening entirely

Sit this playbook out (decision rewrite or schedule later) if:

  1. You ship HDRP-only and cannot spare a URP spike.
  2. Your art bible requires specular GI as the hero without a probes plan.
  3. You are mid-cert with a frozen editor—no preview packages allowed.
  4. The team cannot staff a second lighting owner for bake fallback.

Wrong evenings create fake “we tried realtime GI” myths. Better to wait for a green version gate.

Key takeaways

  • unity surface cache gi is Unity URP’s realtime indirect diffuse GI preview—not automatic Lumen parity.
  • August framing: Unity 7 GI marketing and SCGI preview stay on separate tracks—receipt both.
  • Public testing is invited on Unity Discussions ahead of a 6.7-class expectation.
  • Enable on a throwaway branch; keep bake as ship lighting until gates pass.
  • Plan specular via probes/SSR; SCGI is diffuse-first today.
  • Treat temporal stability as a hard gate.
  • Measure cost on your targets—especially without HW RT.
  • Use P1–P6 and a dated receipt (2026-08-02 class).
  • Run the Monday ritual so FOMO does not silently enable SCGI on main.
  • Do not delete production lightmaps after one pretty screenshot.
  • File bugs upstream with repros—preview exists for that.
  • Compare honestly to Lumen Lite UE 5.8 only as cross-engine diligence—not API copy-paste.

FAQ

What is Unity Surface Cache GI?

A URP feature that computes realtime indirect diffuse lighting using a world-space surface cache of patches, with hardware or software ray tracing backends, aimed at dynamic lighting/geometry without traditional bake waits.Preview GameFromScratch

Is Surface Cache GI ready to ship?

Treat it as preview/alpha until your Unity version and stability gates say otherwise. Unity discusses a 6.7 readiness expectation without hard guarantees.

Does it replace lightmaps tonight?

For production: usually no. Use SCGI on a preview branch; keep bake/APV as the ship path until you pass P1–P6.

How is this different from Lumen?

Different engine and design goals. Unity emphasizes reach/performance and currently focuses on diffuse GI; Lumen is a different Unreal stack with different cost/quality tradeoffs.

How does this relate to the Unity 7 roadmap?

Unity 7 marketing highlights Surface Cache GI as a graphics direction. That does not make today’s preview LTS-ready. See the Unity 7 roadmap playbook for pillar decisions.

What about specular reflections?

Plan reflection probes and/or screen-space techniques. Do not expect full specular GI from SCGI alone today.

What scripting define should I add?

Follow the live Getting Started doc linked from Unity’s preview thread. Community write-ups often mention a SURFACE_CACHE define, but menu names and defines can change between preview drops—quote your doc version in the receipt.

Can I use SCGI with procedurally drawn meshes?

Unity’s preview materials discuss limitations (including some procedurally drawn / indirect draw cases). Verify against current docs before promising SCGI on GPU-driven foliage systems.

Related reading

Official and useful sources