RustRaptor/docs/hd-full-review.md

Complete-original HD review

Independent review started 2026-10-10, extended 2026-10-11. This is an ongoing acceptance record, separate from the generation jobs and registration coordinator. Full completion requires all 5,753 standalone source picture IDs to have reviewed, registered remasters and their retained rendering paths to work. Candidate images, queue entries, original fallback and a successful PNG decode do not establish completion.

Inventory checked independently

A strict picture decoder read every one of the 6,449 original catalog entries and checked the recorded byte lengths. Its 5,753 picture IDs, source SHA-256 hashes and logical canvases match the maintained completion ledger and the previous independent Rust pixel audit. 4,768 pictures are outside sprites/. Every picture has remaster_required: true; no unused enemy, directional, terrain or miscellaneous picture has been excluded.

The disjoint queues contain 659 jobs and 5,714 source targets in this snapshot. Their union with the 46 registered pictures covers all 5,753 IDs exactly; seven registered dark-player pictures remain in their frozen queue. Every job target's source hash and canvas match the independent inventory. The original unused WORM direction and DSHP picture families remain explicitly required.

All 4,535 terrain pictures remain required, including the 734 destroyed-only IDs absent from intact-map screenshots. Original missions reference 3,483 intact and 736 destroyed IDs, a union of 4,217. The complete target inventory includes the terrain pictures not referenced by those missions.

The 303 AGX delta frames and four bitmap fonts have explicit preservation policies: classic original cinematics and sharply scaled native glyphs. These exceptions are separate from picture coverage and are never counted as HD replacements.

Snapshot ledger SHA-256: 3b024a1a1dd342afabfb0f3829f1933769869f6a8798059826e796b3f7e34fb4. The ignored evidence is .reference/hd-full-review/inventory.json; later ledger updates need a fresh snapshot rather than treating this dated count as current.

Registered pack reviewed

Manifest SHA-256 aa2b823acce157f0b49ee2f44bad133cefeca0304905038f004b59b4b6c22835 supplies 46 frames in 16 complete families. The actual Rust loader decoded each at all four presentation scales, producing 184 correctly sized indexed sprites with no diagnostics. Every source hash, logical canvas and zero origin passed. All seven source/1×/4× contact pages and six separate native-resolution 2×/3× pages were inspected, including every bank pose and every animation frame.

Reviewed family	Original IDs	Frames	Assessment
Light player banking	1011a–10120	7	Bank ordering, materials, canopy, wing tips and engines remain readable.
Dark player banking	20045–2004b	7	Distinct dark lighting preserves the approved bank geometry and canvas.
Ground targets TARGT1/2/3	10138–1013a	3	Source orientation and centered mechanism preserved; full opaque canvases intentional.
Ground target TARGT6	1013d–10141	5	Complete opening/firing sequence aligned on the same square structure.
Fighters SHIP01G1 / SHIP27G1	102f8, 10330	2	Source directions, canopy and gunmetal/blue or neutral material cues retained.
Industrial boss	10306	1	Full source footprint and left/right machinery retained.
Plasma and enemy-cannon projectiles	101cf–101d0, 101d6–101d7	4	Readable indexed projectiles; Plasma's known 1× tail shortening remains documented.
Energy flash/rings	101ae–101b9	12	Complete expansion sequence aligned; known narrower first flash and thin rings remain documented.
Forward, Plasma, Dumb Missile, Air Missile equipment	102b0–102b2, 102b5, 102bf	5	Logical inventory canvases retained; loss of original dark opaque outlines remains documented rather than called identical transparency.

This pack is partial: 5,707 required pictures still have no registered remaster in this snapshot. Static fighter approval does not cover other old enemies, bosses, directional variants or terrain. A generated menu background does not count as remastering the original portraits, HUD and retained menu pictures.

Evidence: .reference/hd-full-review/snapshot-aa2b823a/ contains the exact manifest snapshot, fresh actual Rust 1×–4× indexed outputs, runtime-qa.json and every contact-sheet page. The loader library is current for asset decoding and HD loading. Desktop rendering changed after that library was built, so this gallery does not establish those newer runtime routes.

A subsequent registered snapshot, 63 frames, has manifest SHA-256 33ec0a782a852ffef36880432d3e8d472f18a39e9a31d545fe47eb0256bb0dfd. Fresh actual Rust validation passes all 252 scale outputs without diagnostics. Every prior 46 frame's indexed output is byte-identical to the previous reviewed snapshot; the 16 canonical terrain frames and Super Shield pickup match their independently reviewed isolated outputs. The newly added terrain's native 2× and 3× outputs were also inspected. The picture inventory still has 5,690 unregistered targets at this snapshot. Evidence: .reference/hd-full-review/snapshot-33ec0a78/review-comparison.json.

Palette animation defect found and corrected: static equipment highlights can select native cycling/fire/flash indices 240–255 despite their original source having none. The PNG source and shape can be valid while the runtime asset flashes unintentionally. The coordinator reported this in Micro Missile/Missile Pods/Mega Bomb candidates and requested an engine correction. Independent pre-fix review found 24 of 63 registered pictures affected across 86 scale outputs. Source-bound palette classes now constrain ordinary art to indices 0–239, water/fire to their own cycling group, and flash masks to exact original labels. Fresh independent post-fix galleries pass all 63 registered pictures and ten additional equipment frames at scales 1–4 (292 outputs) with no newly introduced dynamic index. All 24 changed registered frames and all ten equipment frames were visually re-inspected without a shape or readability regression. Evidence: .reference/hd-full-review/snapshot-63-palette-audit-fixed/ and equipment-palette-fixed/.

Isolated candidate reviews

Candidate	Candidate JSON SHA-256	Actual Rust outputs	Status
Terrain 10360–1036f	ad6062f325e730bdb71d22190ff1589f5d86f6c69c62ed94e8caacfdd9ce955c	16 × four scales	All source/1×/4× rows inspected; 48 contexts, 755 cardinal pairs and four native scene captures independently pass. Approved for partial-pack registration.
Terrain 2010d–2011c initial	74d3a5e8cc32569c849c8baeee3a47d7c265e238c5d00b5f84a98a52198e8527	16 × four scales	Rejected: 20111 invented a tree canopy; corrected candidate reviewed below.
TARGT5 v2 1013c	e7af1db9fb30f995eaf2c2c360f86c952401bac874087b4f4c462487a9e9ca8a	One × four scales	Structure preserved and prior orange speckles removed; darker soil needs neighbor review.
TARG10 v1 10161	44a2720e6b6164cb96888a024b5fc068c4abe6e331cff54e3389c721890c4eb4	One × four scales	Cannon pose preserved; orange border speckles and soil color need correction or contextual approval.
Super Shield pickup 102d0	adf4b4823aee9bade62b755b170bc0393b73d84c02653e51827109016b3d97ea	One × four scales	Independently approved; golden S glyph, pointed silhouette, transparency and source canvas preserved.

All five isolated packs passed source hash, canvas, origin, PNG/crop and nonempty indexed validation without diagnostics. That result does not approve their artistry. TARGT5's outer three logical pixels change from source mean RGB 124/76/63 to indexed 1× 74/32/28. TARG10 changes from 104/65/55 to 4× 92/40/30 and introduces bright orange border pixels (1.63% of that band; source zero). These measurements explain the requested ground-neighbor check; they are not universal numeric art gates.

Evidence is under .reference/hd-full-review/terrain-10360-1036f/, terrain-2010d-2011c/, held-targt5/, held-targ10/ and equipment-102d0/. Only source inspection and preview extraction were performed; no candidate pixels or global registration were changed.

Subsequent complete-family approvals, 2026-10-11

Fresh source-bound Rust galleries and complete sequence inspection approve the following additions. Every frame passes all four scales without loader diagnostics or newly introduced palette indices 240–255. Native 2×/3× views were separately inspected for the new equipment, fighters/bosses and firing player.

Family	Candidate JSON SHA-256	Frames	Independent finding
Micro Missile pair	469d467a03864420d21dbe8d70e89ff503b9c241ea791401d179a63003d60711	2	Source glyphs, paired payload and footprint retained.
Missile Pods pickup	7584fc2ec68033e66c75a185f7ea3f5e781a0eab187c64b165879fc12f104bd5	1	Pod arrangement and logical equipment canvas retained.
Mega Bomb pickup	4f9c2542ea6d16ead3a1a571a171afff15b13bcb1d9490ba83602067a0d8c672	1	Source radiation emblem readable; palette defect corrected by the loader.
Additional equipment 102d5–102da	Six hashes retained in equipment-palette-hold/candidates.json	6	Complete isolated definitions, source material and pickup footprints preserved.
Fighter 1030a	f98cd3c1b58628f15c506271cc6308007d4fae04618b746c344799c09cb74a98	1	Swept source wings and long nose/tail remain legible at 1×.
Fighter 102f9	d5c8e2dcf3f91806752e40487c4473168c9fe9fc353e05d52bfb2ee1abe7153f	1	Red industrial upper panels and separated lower angled engines retained.
Boss 10308	646f5b42172d6be5bb19c39256807f33fdf6e697565bbacf1f5ae07242b59538	1	Curved shoulders and full central machinery preserve source arrangement.
Boss 1030f	f2043a5722023e57342fedab8b021998fa9234bdce5c09ee47262c02ee2f2915	1	Large mechanical silhouette, center shield and lower pods stay aligned.
Boss 10307	55e0c67363e7684b66e25393a1cf6bcebacd8cafab4fc3f0c0d91d08a4aaacbf	1	Approved with disclosed dark-perimeter interpretation; three upper towers, center weapon and bottom pods retained.
TARGT5 v3	629a75b87ef64d01350520207911f2c2a992bb70727913b1b7d489462f820755	1	Earlier soil hold resolved; subdued brown footprint and mechanism preserved.
Firing player 2004c–20052	8462634f071b1b32b4e6814e0e220c0c6fac59ec202ba2cccfa28e03512e04cf	7	Entire bank sequence approved; corrected straight-frame contour checked explicitly.
MiniGun equipment cycle 102b6–102b9	4ab17e410808ca03616551fff76f36e5505b16083f2090bd21c443b7b6b91663	4	All four barrel/vent rotation phases preserve source order.
Dumb Missile 101ba–101bc	c49d40e194870c18d1d70c01c498bc8f327b2ce720017fdf8e4798487d005d72	3	Complete short/long orange exhaust phases; exact logical alpha masks.
Pods/Air Missile 101bd–101be	f434f793e58fa63ffe46f09bdd0be747e0bb7f6a31558feb6e1e96081c526cba	2	Red nose, paired fins and two blue exhaust phases preserved.
Ground Missile 101bf–101c0	0eb6ec3fa8d6b8702d82ba3326c19e5b93451901337475452220655aa2780cb2	2	Green projectile material, exact mask and phase geometry retained.

Boss 10307 has logical alpha IoU 0.8293. Its replacement removes 373 original opaque perimeter pixels (252 very dark) and adds 23; bounds move by at most one logical pixel. The full silhouette remains readable and unclipped, but the alpha contour is an interpretation rather than identical. Evidence: fighters-bosses-cohort/runtime/10307-alpha-difference.png. Original collision dimensions remain authoritative.

TARGT5's source soil-band RGB is about 124/76/63; v3 is 109/61/50 at 1× and 108/61/50 at 4×, substantially closer than rejected v2's 74/32/28. No bright orange gravel or accidental animated labels remain.

Straight firing frame 2004f now keeps (19,8) opaque and (11,12)/(11,13) transparent, with (11,14) still opaque. Broad mask scores alone previously missed those contour errors. The seven source/1×/4× rows and native 2×/3× sequence pass independently. FPLAYER's coordinator-owned job wrapper must record its own manifest hash; the hash above identifies the independently reviewed diagnostic proposal and exact same source/artwork.

Evidence is in ignored .reference/hd-full-review/equipment-palette-fixed/, fighters-bosses-cohort/, targt5-v3/, fplayer-current/, dumb-missile-3/ and new-projectile-equipment/. These approvals do not imply all items are already globally registered.

Industrial terrain review, 2026-10-11

The complete 32-picture industrial cohort passes source arrangement and exact indexed loader checks at scales 1–4. Independent visual review inspected all eight source/1×/4× pages and every one of the 51 exact composition pages: 300ed–300fc includes 456 cardinal pairs, 96 intact/target contexts and 48 state-partner swaps; 300fd–3010c includes 404 pairs, 90 contexts and 45 swaps. Wall/roof corners, horizontal/vertical pipes, pit margins, original damage placement, opacity and source palette classes remain correct. Reverse overlay order is identical at every tested scale.

Runtime texture hold: repeated narrow grain ridges appear across the plain yard at HD/HD joins and pale strips cross roof surfaces. Exact case 72 visibly adds ridges at presentation x=128 and x=256 absent from the original Scale4x yard. This is a renderer/composition finding; no generated artwork was edited. Both cohorts remain held for a corrected composition proof before calling the terrain seam-free.

Hash-bound candidates: 9524b51b512e6cc02cd5f4f343c6391b022b912b1caf8d19e1cc7201fdd1628a and 0206f86be19d8ccc3eceeb3893191ac29fb628bfd8df03a919cf07b5e4759bde. Evidence and every inspected page hash: .reference/hd-full-review/terrain-set3-cohort/context-review.json; full original/composed comparison: case-0072-full-pair.png.

Foreign destroyed-state partners retain original fallback in these isolated packs. Their swaps prove that composition selects and joins the correct source state; those partners are not accepted HD debris and remain required in their own remaster jobs.

First terrain family's contextual acceptance

Candidate 10360–1036f passed independent inspection of all 48 actual 3×3 contexts and all 755 cardinal pairs across 24 review pages: 170 HD/HD joins and 585 HD/original joins. Trees, dirt/grass boundaries, adjoining native coastlines, roads, structures and existing destroyed neighbors keep their source arrangement. No tile hole, crossing-material leakage or low-resolution interior edge grid was found. Texture contrast at the temporary HD/original boundary remains visible, so this approval is for the current partial pack and does not establish finished full-map consistency.

Four actual native scene captures show phase 0, phase 31, intact terrain and settled destruction. The engineer's runtime assertions compare cached and fresh composition at every one of 32 scroll phases, preserve simulation/pilot/script/RNG state across draw calls, and invalidate the prepared layer after mod hash or load-order changes. The final release-profile isolated warmed scene averaged 2.969 ms; its prepared 320×268 strip is bounded to 2.62 MiB plus a 64 KiB key and the separate 2 MiB transition lookup. The final native cache proof SHA-256 is 67bf62d4bb82606971d4a4c40aef46642b6f9730e2c9a5f34c2de6d054aa8076. These scene results are distinct from the PNG gallery check.

This family has no declared destroyed-state partners. Its destruction capture exercises another native tile's intact/settled cache transition; it does not approve generated HD debris. Those 734 destroyed-only targets remain required in their own jobs.

Runtime proof SHA-256: c00e11d26e2a49a6b0f2e156581a7a7b1fc217f922a34e6b70ed7dbd5b95a0de. Independent evidence is .reference/hd-full-review/terrain-10360-1036f/context-review.json, its 24 cardinal pages and three 48-context contact parts. Original native captures are under .reference/terrain-hd/terrain-10360-1036f/.

Corrected second terrain family, 2026-10-11

Candidate SHA-256 e65fe2737df0f02746cd5024decd6d00e6aeffa0574566e24b7ab803c592e3fb replaces only 20111 with separately generated 20111-v2.png; the remaining fifteen raw atlas cells stay unchanged. The invented canopy is gone and its original olive grass/brown dirt arrangement is preserved. All sixteen fresh source/1×/4× rows, 48 refreshed actual contexts and 519 cardinal pairs were independently inspected. Neighboring original trees, roads, channels, coastlines and intact/damaged structures retain their arrangement; no material leakage, opacity hole or invented tree was found.

The source artwork is approved, separately from the same compositor texture gate affecting industrial terrain: fine repeated grain bands remain visible at some HD/HD joins. Do not describe this partial terrain as fully seam-free before the composition correction passes. This family has no declared destroyed-state partners and does not approve new HD debris.

The runtime proof JSON hash remains 68726a272d094c520203a303dcb57b755fec8974da48a4b7f6b35e2b2da3af0f because that older schema records placements rather than artwork digests. The independent review snapshot additionally binds the exact corrected candidate and every refreshed PNG hash, avoiding reliance on the unchanged proof JSON alone: .reference/hd-full-review/terrain-2010d-2011c-corrected/context-proof/proof-snapshot.json. Its three context and seventeen cardinal pages were all inspected.

Additional effects approved, 2026-10-11

Complete blue sparks 1019c–101a4 (nine frames; candidate aaae1833221c1bb2ec7bc93bf03609d89a26bfb822e580e358b9ada6f92a7c8b) and Death Ray hit 101f1–101f4 (four frames; 56b317e51be2d9c973f6e82f1fd5310a81935b03b468ef3e14e9a3f9717235ed) pass fresh exact Rust checks at every scale. All source/1×/4× rows and native 2×/3× sequences were inspected. Muted lavender, sparse disconnected fragments, phase order and cross/corner/star shapes stay source-faithful; ray-hit logical alpha masks are exact. No new reserved cycling/flash index is introduced. Evidence: .reference/hd-full-review/sparks-and-ray-hit/.

Shield, muzzle, power and explosion families with contour/phase cautions remain separate holds; this approval does not apply to those sibling effects.

Further effects reviewed, 2026-10-11

Independent review of 43 complete effect frames passes fresh actual Rust 1–4× loading (172 outputs, no new reserved palette class). All eleven source/1×/4× pages and eight native 2×/3× pages were inspected. Three families are approved: Platoon 10213–1021c ten frames (b52c4302d8410a66c27e7ebae1ea9fcd93b12753f31fc7ad1c23464d74a49980), AirBlast 102a0–102af sixteen (e4611cd22e202e7a8d2d9a8fc0e3c79f1306258ef21e5825dc4fb0297b9df486) and PickupGlow 10225–10228 four (48a26513a296269043c0bce1c4211fb74aa8cb6f66e9df716c338331ddb01e86). Source fragment positions/green debris, moving hot patches, spoke order, shrinking ring and center holes remain readable. The thinner AirBlast late rim and rounded PickupGlow contours are disclosed interpretations rather than identical alpha. The actual patterned Canvas::shade pickup comparison preserves every transparent hole and never paints the generated mask RGB; native logical anchors remain authoritative. Its test is local indexed composition, not a full Host timing/world-neutrality proof.

MediumAir hold: 101d8–101e4 thirteen frames (e833d7e3792b82ba3951bd980055ae42f4dfe1b959ac7f0b890ce33725525e7f) preserve the early hot-patch ordering, but final 101e2–101e4 rings shrink inward by 1–2 logical pixels and lose 27–38% of original opaque pixels. The last frame shifts its alpha centroid 1.37 pixels right and 0.88 up. The thin HD path visibly sits inside the original at the top/left/bottom; its original outer path, width and gaps require a generated correction before whole-family approval. This is a visible shape finding, not a universal IoU threshold. Evidence: .reference/hd-full-review/effects-context-cohort/, including 101e4-logical-difference.png.

Large effect cohort and concrete holds, 2026-10-11

All 96 candidate frames across Lightning14, GroundBlast42, SmallAir14 and Flare26 pass fresh actual Rust 1–4× loading (384 outputs), with no new reserved palette class. Every one of the 24 source/1×/4× pages and sixteen native 2×/3× pages was independently inspected. Lightning 1018e–1019b (5eb9ba29895acc6ecaca25b3012bc20c9bee1b8818930d97da31d4e22c7d8e6f) is approved: branch growth/retraction, lower anchor and terminal compact spark remain readable; newly thin continuous branches are slightly cooler blue than the original lavender.

Three complete families remain held for visible source differences:

• GroundBlast42 10164–1018d (9b6060fad4b39b5d9b43e18087eca2ae028be2c7f34886ecd01abec7b1f7ee60): late smoke and embers are substantially overlit. Occupied-pixel RGB at 10172 changes about 69/27/15→103/48/25, 10174 49/25/20→78/43/33, and 10176 35/30/29→68/58/56. Source fragment ordering remains recognizable, but dim late smoke must retain its original fade rather than a new glowing coal treatment.
• SmallAir14 101fc–10209 (badd1b4badbfa982c418b9100ca61350d8cd5d31f43429533eaeaf321a875a60): final 10209 drops the original left-side ring fragments at 1×. Its left bound moves from x=1 to x=7; 23 source opaque pixels become eight. Restore the native fragment positions and enough physical width to survive the supported presentation scales. Earlier phases are coherent.
• Flare26 10252–1026b (b5b7b03d5b4c64f1c2213e54bca11e7a175fb1779488f395fa45f794e554911c): 10267–10269 fill original transparent center holes with opaque dark/red pixels. At 10268 logical (16,16) changes from transparent to black index 0; 10269 gains 46 opaque pixels inside its central 7×7. Exact loaded 4× pixels composed over a checkerboard show scenery being hidden. Restore original center-hole growth and late gaps in generated artwork; a broad mask overlap score misses this problem.

Evidence: .reference/hd-full-review/effects-large-cohort/independent-review.json binds every page and candidate hash. 10268-transparent-hole-proof.png and 10209-logical-difference.png isolate the transparency and disappearing-fragment findings. The checkerboard is a read-only visualization of exact Rust-loaded masks, not a full Host simulation capture. No candidate pixels were edited; all three holds were sent to the artist and coordinator.

Original portraits and title logos approved, 2026-10-11

All nine original portraits/ID pictures 10111–10119 and both title logos 10013/10014 pass fresh independent source-bound Rust loading at scales 1–4, with all 44 indexed outputs current and error-free. Every source/1×/4× row and native 2×/3× picture was inspected. Faces, hair, clothing, earrings/eyepatch and slanted ID-card backgrounds retain the original character cues. The English title/subtitle and four Chinese characters preserve their original lettering and spacing, with newly authored metal/stone detail. Alpha contours are interpreted, not identical; original logical canvases and anchors remain authoritative.

Candidate hashes: portraits bf318311594ef16469922c19047b36dbb98c8b8c38fe3a03f86028729a82781e (frame binding c1124bf4fbbd8a79132ffc148bc9f92b18e59c5b49e2c4ddec278d95c5f9776d), English logo 0c9c021b59bafce53ca9a2f3482b1d632bd97378c3082886ee2422c227de1908, Chinese logo 723476d5b40c450dc939f9dc31d7d77b087e4c1ee9160481dde7de67a5c32fbe. Ten static sources gain no reserved palette class. The Chinese source already has 478 fire-class pixels at index 245; its 1× replacement preserves 478 pixels within the allowed 245–249 class rather than introducing a new animation class.

The actual SDL SD/HD four-ID Main-menu comparison was independently viewed and its changed-pixel bounds recomputed: surrounding UI is byte-identical, names/IDs remain readable, and portraits stay inside their menu frame. The five full portraits and both logos remain retained catalog artwork, with no claimed active menu route. Evidence: .reference/hd-full-review/interface-cohort/ and the owner's interface-other-10111-10119/runtime-review/live-menu/menu-proof.json.

The final Stormwall 1280×800 and aspect-correct 1920×1080 actual SDL captures were also inspected: indexed terrain, foreground, HUD and letterbox placement are clean. This contextual check does not resolve the separate original-industrial terrain grain hold or establish original-art completeness.

Runtime routes and remaining gates

The former original beam/effect routing gaps have dedicated native composition paths. Beam shafts, muzzle/power/impact companions and secondary smoke/sparks/glows preserve original frame IDs and original anchors. Original shadows deliberately retain the original logical masks and collision dimensions.

Code review of the newly extended HUD, timed currency pickup and light-mask routes confirms they request HD frames at the presentation scale while retaining source logical placement. Light masks call indexed Canvas::shade instead of painting RGBA colors. Generic HUD assets keep nearest-neighbor source fallback; bitmap font text remains sharply scaled. Focused runtime fixtures now pass at 1× and 4×: generic HUD and equipped-weapon icons use supplied HD pixels at the original logical anchor; a sparse red RGBA shield replacement uses its own alpha holes and transforms background index 37→44 through the light table rather than painting its red foreground. World/pilot state remains unchanged. The broader 13 original-rendering checks also pass. A dedicated full Host::scene currency fixture now passes at 1×/4× and interpolation α=0/0.5: synthetic native N$_PIC stays green in SD while supplied red/blue HD halves appear at the interpolated source anchor, with every pickup pixel and its surrounding black border checked. At 4× the half-step anchor is (246,410). World digest, RNG, pickup countdown/previous-art-position/glow-frame, pilot and clock state remain unchanged; no HD diagnostic appears. This proves route behavior separately from artistic acceptance; no currency remaster is approved by that synthetic test.

Canonical terrain candidate review and actual terrain composition checks are ongoing. Fresh isolated Rust review decoded all 32 tiles in the first two groups at scales 1–4 without diagnostics; all eight paged source/1×/4× contact sheets were inspected. The first group (10360–1036f) keeps the broad tree/grass/dirt placements. The second group's original canopy drift is now resolved in its corrected candidate; compositor grain ridges remain a separate presentation gate. Every candidate tile requires independent source geometry review plus neighboring placements, mixed HD/original joins, destroyed states and scroll phases. Opaque alpha-mask overlap alone is trivial and is not an artistic quality gate. Rendering test success adds no artwork coverage.

Opening-credit pictures now have a separate source-palette rendering route: the active card exposes its unfaded POG/CYG palette; supplied HD artwork is quantized using its complete static 256-color palette before the existing cinematic fade. The integration agent reports 19 HD-loader, seven portable-cinematic and two full-Host opening-card fixtures passed, including 1×/4× index 250, start/half/full/end fades, absent/rejected artwork fallback and unchanged world/pilot/RNG plus the 569-tick synthetic cue stream. Two original-data cinematic checks preserve opening 966 ticks and AGX 925 ticks. This is route evidence reported by the implementation owner, separately from independent artistic review; actual opening-card replacements remain unapproved. See docs/hd-opening-cards.md. AGX films and bitmap fonts retain their explicit classic policy.

Open findings

• Full art generation and registration are unfinished; preserve the per-ID checklist through every cohort.
• Generic HD palette quantization is corrected and independently passes all current supplied frames at scales 1–4. Preserve this source-class gate for every future family.
• Firing-player seven-frame sequence is now approved after specific straight-frame contour correction and fresh complete all-scale review. Preserve those source-bound contour checks.
• Generic HUD, lighting-mask and timed-currency routes now pass supplied-art runtime fixtures at 1×/4×, preserving indexed shading and original/interpolated anchors. Review their actual generated artwork separately when candidates arrive.
• Resolved source hold for terrain 2010d–2011c: rejected initial picture 20111 invented a tree canopy. Its original palette contains olive grass and brown dirt only; the remaster adds a bright green canopy at the upper-left covering about 15% of logical pixels. The source hash and opaque alpha-mask overlap pass, demonstrating why those checks do not establish source geometry. The separate v2 correction now passes all source and contextual checks; preserve the original rejected evidence. Coordinator, artist and terrain engineer were notified. Proof: .reference/hd-full-review/terrain-2010d-2011c/runtime-contact-sheet-002.png and 20111-original-enlarged.png.
• First canonical terrain group 10360–1036f is independently approved. Corrected 20111 no longer invents a canopy and the entire second-family contextual review is complete. Both industrial cohorts remain held for repeated join grain ridges.

This record will be extended as new complete families arrive. Original-derived artwork and review renders remain in ignored data/ and .reference/; authored mod artwork and sounds remain Git-tracked under RustRaptor/mods/.

2026-10-11 resumed terrain acceptance

The final mirrored-detail compositor is independently approved after all 66 full-corpus gallery pages were viewed, including all 1,706 cases: 1,379 cardinal pairs, 234 intact/target contexts and 93 destroyed-state partner swaps. The seven source/pristine/previous/fixed panels were also inspected. Every gallery digest was recomputed and matches the frozen index. The reviewed terrain_hd.rs SHA-256 is b30b4a65a9e646e9fe106041ad718eebd2740c6c90354267a7b17b770e50ebb0; the source hashes of hd.rs, assets.rs and render.rs are recorded alongside it in .reference/hd-full-review/terrain-ridge-final-independent.json.

Repeated vertical/horizontal grain ridges and pale seam strips are gone. Roof/wall/pipe geometry, broad grass/dirt boundaries and corner placement remain source-coherent. The previous presentation hold is resolved for these exact candidates: terrain-2010d-2011c (e65fe2737df0f02746cd5024decd6d00e6aeffa0574566e24b7ab803c592e3fb), terrain-300ed-300fc (9524b51b512e6cc02cd5f4f343c6391b022b912b1caf8d19e1cc7201fdd1628a) and terrain-300fd-3010c (0206f86be19d8ccc3eceeb3893191ac29fb628bfd8df03a919cf07b5e4759bde). Earlier source-art and complete scale 1–4 reviews remain applicable to their unchanged raw/crop bindings. Registration belongs to the coordinator.

Foreign damaged-state neighbors remain original fallback. Their coarser material patches are visible and add no HD destroyed-art coverage. The first canonical terrain family still has no declared destroyed partner. This approval covers these 48 remastered source pictures and the final composer presentation, not the uncompleted 5,753-picture catalog. The renderer owner's 13 portable checks, corner/grain regression and native 32-phase cache/damage checks are complementary runtime evidence; the visual review does not manufacture a new gameplay/performance result.

Restart checkpoint: .reference/hd-full-review/restart-checkpoint-2026-10-11.json. The next independent queue is Flare26 v2, Muzzle4 v4, SmallAir14 v2, Coconut4, then the nine reserved enemy families (35 frames): 10311–10314, 10334, 1033d, 1033e–1033f, 10340–10343, 20053, 20056, 2006f–2007b, 20080–20087. Existing historical effect holds remain attached to their rejected candidate hashes until each new correction is reviewed. Death Ray/Energy equipment is reserved to another reviewer.

Current render.rs differs from the frozen corpus due to menu font color/text clipping and a separate cosmetic lightning function drawn after terrain. Root confirmed Canvas scaling, blit, indexed shading and terrain composition are unchanged. The exact current/frozen hashes and this scope are retained in the independent evidence; no terrain acceptance is inferred from unrelated menu tests.

2026-10-11 prior cohort journal

The source/enemy/equipment cohort at .reference/hd-full-review/enemy-equipment-cohort/ was independently inspected completely at original and Rust scales 1–4, with fresh 168 scale outputs for 42 frames. All 42 were approved: Ground Missile pickup v6 (1), Turret (4), Pulse equipment (2), Energy Grab equipment (4), door 10142–10150 (15), bonus 10151 (1), treads 10300–10303 (4), hovercraft 10304–10305 (2), statics 102fa, 102fd, 10327, 10331 (4), 102fe–102ff (2), corrected TARG10 v3 10161 (1), and fighters 1032f, 10309 (2). Exact candidate digests and frozen candidate copies remain in that cohort's candidates.json and runtime evidence. Two peripheral single-pixel Energy Grab holes and tiny peripheral fighter holes differ from source, while the primary transparent loops/slots are preserved. Fighter 10309 has brighter copper material (occupied mean RGB 64/47/42→82/58/52), disclosed and accepted independently of the separate bright-soil issue. TARG10 v3 restores subdued brown soil and supersedes its earlier orange-ground hold.

The eight-frame addon .reference/hd-full-review/soil-and-tiny-projectiles/ also passed every source and actual scale 1–4 view. Corrected ground 1013b (23be1ef3cca01b1c6b82b0984cbd07c8817b8c4a3a83edfe5c140828e0f5c9e2) and 10160 (90938e509abe1f82c9c7aba48e322697a36947808dc885a538a6b1e17ead8966) restore quiet rust-brown soil and supersede their old candidates' brightness holds. Forward Guns/MiniGun four 2×2 phases 101c3–101c6 (1193e3810f7be2063d3e29dfda07b060bd2c232d0df044303577099261140bf2) preserve each gray/white native mask and phase. Mine two phases 101d1–101d2 (5da8b059289e3147b433adf1128d86aaf3417e4f1678213ac17e3e6dce8ee204) preserve pink/burgundy centers and gold corners, with somewhat duller warm patches accepted. No live scene occurrence is claimed by these gallery checks. Coordinator confirmed all 40 addon source/scale PPMs match the independent outputs byte for byte.

Six further complete phases were approved in .reference/hd-full-review/laser-and-pulse/: Forward Laser equipment 102c8–102cb (4d849e980722dd1ddc948d7a7faab98f268c8c97dc112689f3215286f863111b) retains four rod heights/light indicators and restored opaque backing; Pulse Cannon projectiles 10223–10224 (9c362e95417cc6a7025bd70012890ca9126fa6236c19aa3721b320d9aa5c70ef) retain exact logical alpha, the muted lavender arches, open centers and native offsets. This does not approve the separately held Forward Laser projectile family.

The independently frozen 215-frame manifest (69c1f9d46c5e5137b444bcde002f6d0216312b613cacc91c3aaa20f60face5e2) passes a fresh isolated actual Rust validation of all 860 scale outputs with zero diagnostics and no stale source-loader flag; its source and candidate bindings are in .reference/hd-full-review/snapshot-215/. Each active family was independently reviewed in the earlier cohorts. The complete comparison is now finished: all 1,075 original/scale PPMs match prior independently reviewed cohort outputs byte for byte, recorded in .reference/hd-full-review/snapshot-215/prior-reviewed-byte-comparison.json (c5f7bd987e086f5a5edd2a1d3fbf223b651347c5a27107f8841ecdbd025f850c). This closes the historical215-frame comparison without implying a new whole-pack visual review or relabeling it as the later active manifest. Coordinator reported active coverage 221 after the six later phases, before the new terrain acceptance.

Original player runtime captures at .reference/secondary-effects/native-player-banking-flames-{sd,hd}.png and native-player-family-{0,1,2}-{sd,hd}.png were independently inspected for complete HD bank families, SD/HD flame phase pairs and firing context. Original anchors, banking alignment, cyan flame phases, localized amber flash and clipping are clean; root's capture test also preserves world digest. The capture was made against the reported 165-frame manifest 3ba076d98f644f39ca6769633f91f83c0b20a9af44036d2afe7905839ddd3310, so that historical route proof must not silently be relabeled as a later whole-pack capture.

The earlier portrait SDL comparison is historical at its previous menu location. The menu is now logical 400×225 with portrait (30,45); current individual .reference/ui-all/ captures and .reference/interface-hd/pilot-menu-400x225/ evidence are the relevant route proof. Pre-existing audit-sheet images are stale. The interface owner independently recomputed the four outside-photo comparisons and reports byte-identical surroundings; current-layout independent viewing by this reviewer remains pending.

2026-10-11 effect corrections and Coconut

Independent review of the complete 48-frame combined candidate produced 192 current asset-loader outputs at scales 1–4, with no diagnostics or added reserved indices. All 12 original/1×/4× pages and four native 2×/3× boards were viewed. The broader game-source freshness flag remains true due to unrelated game/menu work, so this is standalone asset/rendered-raster proof, not a new full-Host capture. Full candidate/raw/page bindings and decisions are in .reference/hd-full-review/effects-corrections-and-coconut/independent-review.json.

Approved: Muzzle four phases v4 101c7–101ca (bf3d51a434d70a68c42412bf6ad36b88e3e1b0b366f11c1113fbeebac538dd26), SmallAir fourteen phases v2 101fc–10209 (939b87d98431626c0ab9c72685ccd0de48f5259b1fc722a3fedf9e1c96bc2fe8), and Coconut four phases 1024d–10250 (153ab82bdab15987be1e92a5779eaa8de4f7386ebf5cec35a60bb399de2c1916). Muzzle's dim stages now keep source brightness and distinct native bounds. SmallAir's final 23-pixel source mask, bounds and separated left fragments are restored; the other 13 frames' complete source/scale outputs are byte-identical to the earlier reviewed cohort. Coconut retains native offsets, stepped contours, opaque near-black spots and dull tan/burgundy patches, with restrained husk detail.

Flare26 v2 (ac360b618f5538a06b8989bb79e7282e4720b5811c484a01c1c34c40a0a822d2) remains held for a new, separately specified material/coherence issue, rather than its resolved transparency defect. All scales of 10267–10269 now show original background at center (16,16); source central 7×7 background counts 12/30/46 correspond to HD 1× counts 14/32/46. The other 23 frames' five source/scale PPMs are byte-identical to the preceding reviewed version. However, the three generated corrections use dark-bordered rectangular pixel-cell slabs and brick-like lava grain, visible at 2–4× and in unchanged raw generated-v2.png. This abruptly differs from adjacent organically detailed fire/ember phases. The artist has been asked to regenerate only these three with original holes/branches and adjacent remaster style, explicitly excluding beveled/outlined pixel-cell grids and metal/brick material. Earlier rejected hashes and original-hole evidence remain preserved.

Independent current-layout portrait review is now complete: all four SD/HD photo crops and one complete Main-menu view were inspected, and all eight capture hashes plus every full-pair difference were independently recomputed. Every changed pixel stays in the screen portrait bounds (96,184)–(224,312); surrounding UI bytes match. The 400×225 canvas at portrait (30,45) keeps portraits, pilot text, menu labels and footer readable without clipping. Evidence .reference/hd-full-review/portrait-current-layout-independent.json binds exact fdf2ff0 capture binary da750499800d33884ddc8ea2beccd3b0339058012525b60805341335d9d28bcb and unchanged portrait frame binding; it does not claim newer navigation changes were captured.

2026-10-11 reserved enemy cohort

The nine reserved enemy families, 35 original pictures, have completed independent original/1×/4× review on all nine pages and native 2×/3× review on all five boards. The isolated actual Rust loader produced 140 outputs with no diagnostics or stale source flags. All candidate and raw artwork hashes remained unchanged; original and HD reserved-palette index counts are zero. Exact candidate/source/raw/output/page bindings and per-family decisions are in .reference/hd-full-review/enemy03-reserved-35/independent-review.json.

Approved 33 pictures in eight complete families: 10311–10314, 10334, 1033d, 10340–10343, 20053, 20056, 2006f–2007b, and 20080–20087. The rotor sequences retain orientation and native offsets, the machinery keeps recesses and materials, and both complete shutter-to-turret sequences preserve stage progression. The opaque olive/gray/green ground contexts preserve original material regions; this gallery does not prove their terrain joins or live enemy reachability.

The two-frame 1033e–1033f family remains held at candidate 65966d76a5311858e8c29649f7c0658421a52372689ef6f3d93ddc619659cfc2. Frame 1033f has a detached gray vertical strip at actual HD 4× column 0, rows 33–59, also visible at 3×; the original left column is entirely transparent. .reference/hd-full-review/enemy03-reserved-35/1033f-alpha-detail.png records the artifact. Its apparently favorable native mask score does not excuse the border artifact. The artist should inspect the unchanged atlas cell boundary and exact source canvas before proposing a correction; no source body pixels should be trimmed to conceal it.

The Git root is now RustRaptor; source-art and ignored review evidence paths remain in the parent workspace. This reviewer has staged and committed nothing. The coordinator reports active pack 298 after admitted terrain and independently cross-reviewed boss cohorts, before admission of these 33 enemies and the already approved 22 corrected effects. Coverage remains incomplete against all 5,753 required original pictures.

2026-10-11 laser power and endpoint corrections

Seven further complete phases are independently approved: Enemy Laser Power v3 10245–10248 (2d5ceba65e0745348119442db9eb2f19bc60a0c07678054e79554e1583deb21e), Bomb v2 101d3 (8225689631f1b65ed4704a6fb0bce65b0130213d41ed8850affd990fdcc232c8) and Enemy Missile v2 101d4–101d5 (69b53e415efc1ac52c927cd13ab72aa7481dc61572a20a3f6d1442b2482e72a4). The fresh actual Rust loader produced 28 outputs without diagnostics or either stale-source flag. Every native mask is exact, source/native bounding boxes match, original anchors remain, and no reserved palette indices were introduced. All source/1×/4× rows and enlarged exact 2×/3× outputs were viewed.

Laser Power keeps pale broad early cores and darker narrow late phases with fine heat grain. Bomb and Enemy Missile keep red/gray/blue material regions, fin positions and exhaust phase offsets; somewhat darker gray highlights are accepted with readable fine casing detail. Evidence and all raw/candidate/source/output/page hashes are in .reference/hd-full-review/laser-power-and-endpoint-projectiles/independent-review.json. This is raster approval, not new live scene or simulation proof.

2026-10-11 gold flash and original standalone opening cards

Gold pickup flash 102db–102de is independently approved at candidate 7c4b3f9bb38ad5ac12fc55da8866a92588fc68bbc78273b11ad54f943ef2b290. Every phase has an exact native mask/bounds/origin and preserves the large → medium → small/dim → medium sequence. The repeated 102dc/102de originals, raw artwork and every actual scaled output are byte-identical. Complete source/all-scale viewing confirms phase distinction and restrained glow with zero newly animated palette indices.

All three complete original 320×200 opaque cards are independently approved: APOGEE 1002b (3475eab74c4a6123f19ce480c4c5288aace2981b7f852eba7929f0d76aa3390f), CYGNUS 1002d (9451cf6d88341ce3a69616c822cc9da2f481b7495dd4b0e0105effbf644efda0) and TAIWARN 1002f (8c609ff0008d2a6a8c9330eb6e14c4ad458abc89cdf7ad09fe9295ba93a10d93). APOGEE retains its title, complete tagline and planet/swoosh/star composition, with disclosed stronger nebula/star detail. CYGNUS retains exact two-line lettering, slabs, perspective and shadows, with disclosed brighter blue faces and metal grain. TAIWARN retains the original Traditional Chinese title/body/company/closing glyphs and five body line breaks, geometric logo and full-width red rule. The metadata crop removes precisely one pixel per edge of each untouched 1586×992 RGB output; no lettering, logo or planet is clipped.

Each card uses its authentic unfaded POG/CYG/TAI source palette with all 256 static indices allowed. All 64,000 original pixels of each independently decoded source raster match its authenticated prepared source PNG. Every original/native1× pair, complete native2×/3× board and full4× card was viewed; all 15 source/scaled card PPMs match the fresh independent renderer byte for byte. The combined seven-frame pass has 28 outputs, zero diagnostics and no stale source flags. Exact raw/source/palette/candidate/output/page bindings are in .reference/hd-full-review/gold-and-opening-cards/independent-review.json.

This is card artwork/raster acceptance, not a new SDL fade/cue proof. The prior engine tests cover APOGEE/CYGNUS still-picture presentation and fades; TAIWARN remains retained catalog art without an active opening cue. Original AGX films remain classic, and this does not change the separate sharp bitmap-font policy.

2026-10-11 refreshed complete catalog coverage audit

The independent strict source decoder again checks all 6,449 original INDEX entries and finds exactly 5,753 standalone pictures, including 4,768 outside sprites/. Exact source byte lengths, span terminators, source/palette/RGBA hashes, logical canvases and header options match the frozen completion ledger. Every picture still has remaster_required: true. All 674 picture-job target memberships preserve those source hashes/canvases and cover all 5,753 IDs; there are no missing, foreign or excluded original picture IDs.

The captured active manifest contains 340 source-bound pictures without duplicate IDs, source mismatch or anchor/canvas drift. Its SHA-256 is 648241ad22ff0863dc3824270d7275cd0184204369349ad0a359a4a08dc32097; this snapshot is a metadata/coverage audit, not a new whole-pack artistic or live-rendering approval. Candidate stages and registration remain distinct from completion. Evidence .reference/hd-full-review/catalog-coverage-2026-10-11.json (c612c947b0f36be2314698ac4e19f75acdaf8101afa8e025bf4082f35993e9ac) binds the complete original inventory and all 674 job files.

2026-10-11 HUD picture family source-stroke hold

The complete 11-picture HUD family 10005–1000f remains held at candidate afdf80893e42db3bd2c2139106d9b5526d2f15e9e7219f08c965f6827cd242c4. All three source/1×/4× pages and all three enlarged source/all-scale boards were independently viewed after a fresh 44-output actual Rust pass, with no diagnostics, stale flags or newly introduced reserved colors. These are N*_PIC catalog pictures, not the four original bitmap fonts.

Digit 7 1000c loses the original bright lower-diagonal pixel (3,6) (RGB223/138/24) at native HD1 and gains dark exterior pixel (1,5) (RGB89/28/4). This narrows/moves a real source stroke. Its favorable 0.926 native mask overlap does not establish stroke fidelity. The other 10 pictures preserve readable strokes, core holes and row placement, with separately disclosed peripheral dark alpha changes. A targeted genuine correction must restore the native digit7 contour before the complete family can be accepted; the remaining 10 must remain unchanged or be reviewed again. Exact source/raw/page/output/pixel-difference evidence is .reference/hd-full-review/hud-pictures-11/independent-review.json. No family approval or active live-route claim is implied.

2026-10-11 — Terrain 103b0/103c0 complete current-composer review

Approved exactly the 32 source-bound tiles in terrain-103b0-103bf (candidate 5e92494eb43d393b6e471890d0f62b70d1df1d0e3937e9987473b7b9aedf8758) and terrain-103c0-103cf (3b60736654d48b933215d20ed4b5f5c3ffa58b44705b2dc73e6a0ec65f0551e6). Both unchanged raw atlases, all eight source/1×/4× pages and eight native 2×/3× pages were visually inspected. Bronze machinery, black recesses, silver pipes, mesh/grate bands, brown soil scars and cyan water boundaries remain source-aligned and readable; authored rivets, casing grain, pebbles and ripples add consistent detail.

The reviewer compiled a private diagnostic helper against the current library and unchanged grain-fixed terrain_hd.rs (b30b4a65a9e646e9fe106041ad718eebd2740c6c90354267a7b17b770e50ebb0). Every one of 120 declared cardinal pairs and 33 actual 3×3 samples was rendered at all four scales, producing 612 cases. Every replacement applied with no HD diagnostics; whole opacity, per-pixel original palette class and reverse overlay order checks passed. All 39 current full context pages were then visually inspected. Pipe/roof/canal edges and corners remain coherent; the repeated join ridges from the superseded compositor do not recur.

Exact dated acceptance and all 1,384 original/scaled/context PPM hashes are in .reference/hd-full-review/terrain01-next32/independent-review.json, SHA256 ca8730305ce6cf3990ef40c22d30bfff74c28152d16524fc64e7426b92ea4d8a. Raw hashes are 3950fc29f19b409b8150249c29b01d685e36c30fcf712eb8cee086db651057a2 (b0) and 8c522dec6bf065bee49af637fc2a9da4fd20895b16603e923c1328f673e46471 (c0). Current source component hashes equal those used for rendering. Historical owner probes against older libraries are not the acceptance evidence.

Every target has an empty state_partners list: this adds zero remastered destroyed partners. Original source scars/fragment pictures are preserved as themselves. Unremastered external neighbors remain lower-detail original fallback and still require their own genuine remasters. This isolated tile/context approval is not a new native Host scrolling, damage, fullscreen or world-neutrality capture. Registration remains coordinator-owned.

2026-10-11 — HUD picture family N7 v4 correction accepted

Approved the complete eleven N0–N9/N$_PIC picture family interface-other-10005-1000f, exact v4 candidate a2d85dbca5f76f94eafbe70325f8a2907599fc0ff6aae89e171925d62518034f. The missing lower N7 stroke in the previous afdf8089… candidate is resolved: its native occupied layout now exactly matches the source, the lower step and shortened dark final foot survive all four scales, and the invented (1,5) alpha is gone. All eleven originals and actual Rust 1–4× outputs were visually inspected on three complete five-scale boards after a fresh error-free isolated loader run. The row placement, digit strokes and core holes remain readable. These are HUD pictures, not the four preserved classic bitmap fonts.

The brightness difference is deliberately disclosed, not treated as exact RGB preservation: original gold (3,6) 223/138/24 becomes copper 158/56/12 at 1×; adjacent (2,6) remains gold 231/158/24. Original shadow (1,6) 85/12/4 quantizes to olive 85/81/12 at 1×. The complete native lower diagonal stays opaque and readable, and its higher-scale authored surfaces remain gold-orange with shading consistent with the other numeral remasters. This localized shading variation is acceptable; another regeneration is not required by this review.

The other ten raw/crop/source bindings are unchanged. All fifty of their original and scaled PPMs byte-match the earlier independently reviewed snapshot, including the previously disclosed minor peripheral shadow-alpha changes; core holes remain intact. The untouched new N7 raw PNG was also inspected. Exact source points, source/raw/candidate/runtime page hashes and all 55 PPM hashes are in .reference/hd-full-review/hud-pictures-11-v4/independent-review.json, SHA256 0265bf01fa89e99ac83ef4c413adb82101058386e8ce9db11e8bce7778784298. This resolves the earlier whole-family artistic hold only for this exact v4 binding. Existing native HUD and currency supplied-art route tests remain separate; no new full Host capture is claimed here.

2026-10-11 — Riverbank and crossing terrain 103e0–103ef accepted

Approved the exact sixteen-picture candidate 8d16ab47efc069ab373b80647ff9f680a8338202e562c1dfd6bac49d9bcb6cd7 after viewing the untouched raw atlas, all four source/1×/4× sheets, all four native 2×/3× boards and every one of 64 current context pages. Source coast curves, rock placement, pale banks, moss/grass regions and both steel-sided/wooden crossing orientations remain aligned. Original plank spacing, posts, rails, beams, water gaps, shadows and debris positions survive all four scales; added ripples and ground grain introduce no new canopy, furrows, material crossing or recurring join ridges.

The private immutable current-composer corpus covers 283 declared cardinal pairs, 44 actual target 3×3 samples and 27 source-declared damaged-center swaps at scales 1–4: 1,416 cases. Every replacement applied with no diagnostics; whole opacity, original palette classes and reverse overlay order passed. The nine damaged partner pictures (1050c, 1050d, 10510, 10511, 1050e, 1050f, 104ea, 104eb, 104ec) retain original fallback: zero damaged partners are remastered by this family. External original neighbors also remain lower-detail fallback and still require genuine remasters; that visible texture-density difference is disclosed.

Evidence .reference/hd-full-review/terrain01-103e0-103ef/independent-review.json, SHA256 2427f5c4b0d264b71654804c3bfbeee41ea69d424e9a04c7355bdf927031cb7a, binds all 2912 source/scaled/context PPMs, raw atlas 1de996634c0edcca2e843c70fa11d6722de8760924582b484dd63e0c3285e745, exact source/canvas/crop records and every viewed context page. Current relevant source hashes match the frozen proof, including grain-fixed composer b30b4a65a9e646e9fe106041ad718eebd2740c6c90354267a7b17b770e50ebb0; corpus proof hash is 525a80107a2ef93b671506d09b59aade7f4dc1a5e715991221eba1c49535e964. Opaque mask overlap is not artistic proof. No new Host scrolling, fullscreen, damage or simulation-neutrality capture is claimed. Registration remains coordinator-owned.

2026-10-11 — Eight original UI pictures accepted

Approved five complete frozen families: SYSTEM DAMAGED 10011 v2 (117d479efc9b40629197076f5e38e6a20c017bd1164df2bbb585bef489e5d61e), counters 102df–102e0 (afb2eecafc8c73b4dc00d830b6cd8390e7988533f6b0ffd11e7133a4191ed087), checkbox OFF/ON 1012d–1012e (2eb314ddc9c38a152eabb47210867caa39a32bdd6bcdcecdaedde014e54138bf), indicator ON/OFF 1012a–1012b (09ee9f9239be52063cabae5c5b8fd23c1a205f621b5412c6526ac3e2b1ab6850) and question 1012c (3cda159ce12c3f6dc52ccb1e8eef471ca271aabf756456390da33c36c9099765). Every untouched raw image, original/1×/4× row and complete native 1–4× board was viewed after a fresh actual Rust 32-output pass with no errors, stale flags or new dynamic palette labels. Exact source/canvas/crop/raw and forty source/scaled PPM hashes are bound by .reference/hd-full-review/ui-pictures-eight/independent-review.json, SHA256 260a6256f99bbbaab606ea01f778defe67e0732e95f3d12c05b0908b37dbf06f.

SYSTEM DAMAGED v2 restores exact original logical lettering cells and warning spacing; brighter gold/orange bevels preserve readable apertures and dark undersides. Both gold counters preserve their distinct symbols, central recess/lettering and outline shape. At native1 some black outlined cells become olive/brown, while higher scales retain dark readable edges; the disclosed shading change is acceptable. The two indicators preserve a clear bright-red versus dim-burgundy state with a continuous textured lens; the old tiled ON draft remains rejected. The question hook, dark shadow, open core, stepped contour, separate dot and row spacing remain source-faithful, with brighter blue edge detail.

Checkbox acceptance explicitly includes material judgment: OFF preserves its 28-cell gray perimeter and dark empty center, while ON retains the original segmented crossed-stroke positions and grayscale pattern. Brushed metal grain and bevels within source-gray cells remain appropriate and readable at all scales. Native brightness is not exact: five source52-gray cells (0,0), (1,1), (2,2), (4,4), (5,5) rise above64; (6,2) changes from77 to48. These are disclosed shading variations rather than lost/moved source strokes. Whole opacity and mask overlap alone did not establish this approval.

This is isolated artwork/raster review, not a new live menu/HUD capture. The four original bitmap fonts remain under their separate sharp-font policy. The suspected straight-alpha resampling diagnostic remains separately journaled; no loader or alpha policy changed, and this approval binds current actual output. Global admission remains coordinator-owned.

2026-10-11 — Shoreline, rock and fence terrain 103f0–103ff accepted

Approved exact complete sixteen-picture candidate 3e7b6f7e4f6355013cb815eb590862a72441bee983fc3145e8a18ab17b52f11f after viewing all sixteen untouched raw thumbnails, four source/1×/4× sheets, four native 2×/3× boards and every one of 83 current five-column context pages. Original shoreline bays/corners, grass banks, fence rail positions/bends/material widths and four-/six-rock group counts and relative offsets remain aligned. Added brown ground/rock grain and teal ripples preserve source materials without invented canopy, regular furrows, new props, recurring seam ridges or material crossing. Fine checkerboard fence mesh smooths at native 1×; finer dark lattice remains visible at 2×–4×. Authored water-wave frequency/angle varies between tiles, and original fallback neighbors remain visibly lower-detail. These accepted detail differences are disclosed.

The private immutable current-composer proof covers 442 declared cardinal pairs and 48 actual target 3×3 layouts at scales 1–4: 1,960 cases. All replacements Applied with no diagnostics; whole opacity, original per-pixel palette classes and reverse overlay order passed. All sixteen source state-partner lists are empty: zero declared and zero remastered damaged partners. External original machinery/wreck/shoreline neighbors remain separately required remasters; their presence is not destroyed-art coverage. Exact native comparison of three original fallback neighbor regions found 0 changed pixels, resolving misleading thumbnail stripe appearances without modifying artwork.

Evidence .reference/hd-full-review/terrain01-103f0-103ff/independent-review.json, SHA256 35b0871570d1e4f57561b3d80e117371a616feee4a51a7d6090c9fc1eeb88da8, binds all 4000 source/scaled/context PPM hashes, sixteen raw PNG hashes, source/canvas/crop bindings and all 83 reviewed context-page hashes. Corpus proof SHA256 8c641364439cd4232e19f0e2991befd9219638731696f3143be7bdfdd55db619 pins immutable library e8b521c4d9a6f2fdf087668c264487014925df129cdd4fa79959eb15286cf224 and unchanged grain-fixed composer b30b4a65a9e646e9fe106041ad718eebd2740c6c90354267a7b17b770e50ebb0; every current relevant source hash still matches. Current job hash d1a770311acd4a3959b49acfcc9cac93eea90c64157099fd6056edf2edcb9107 differs from the corpus input only by target status/count metadata; all source/context/partner declarations are unchanged. Opaque-mask overlap alone did not establish approval. No new Host scrolling, fullscreen, destruction or simulation-neutrality capture is claimed. Global admission remains coordinator-owned.

2026-10-11 — Four complete original tank animation families accepted

Approved the exact fifteen-picture cohort: 30074–30077 (9eb74fb9566b3ef05a4dc3234298cf9ee5d307d9836bd2c119bab9ef4f44a97d), 30080–30084 (1ace4e117a41b4a2790441b94331eb05bc7d66eb711100a0db325d068726758e), 30085–30088 (fb0785c9c88080cbee4602e784892ad29a71cdedbb2ded6f579e901e7c3db4dc) and 3008b–3008c (05e1ca21fd8f7d7b02459175b03d963235f2b0eb86612276fffb52e4305ebfe1). All four untouched raw atlases, every original/native1×/4× row, every native2×/3× row and enlarged whole-family source/4× animation board were independently viewed. Fresh immutable actual Rust decoding produced 60 scale outputs with no HD diagnostics, relevant-loader staleness or new dynamic palette indices. All 75 original/scaled PPMs match the owner’s frozen final evidence byte for byte. Built-in original outputs and repository raw copies have identical authenticated hashes.

The gray/teal tracked machinery keeps its central ladder, dark aperture and four authentic tread phases. The copper four-slat machinery preserves all five original highlight/recess stages, gray side bars and lower split legs; its pale copper bevels are slightly brighter, and source dark bands remain opaque recessed material. The bronze/gray family preserves the central framing, side pods and original fork-length progression (phase0 occupied through row29, later phases through row28). The horizontal red family keeps its original aperture, narrow slot, four gray rake bars, pale marks and distinct two-phase tread patterns. Shared physical/native scale and empty-margin alignment retain every raw alpha≥16 and alpha≥32 pixel without RGB/alpha edits.

Small-scale differences are disclosed: copper loses source black boundary (3,27) at1× in all five frames and at2× only 30082/30083; bronze loses 3/4/2/4 source contour cells at1× and (21,28) at2× only 30087/30088; red loses black (1,14) at1×. Every original source cell survives3× and4×. Gray/teal loses no whole source cell at any scale, with one added peripheral native cell in each later phase. Partial higher-scale boundary coverage remains recorded separately; native bounds/major limbs are not a crop-only approval.

Original stationary body colors are not byte-identical across authored frames. Independent actual indexed comparison exactly reproduces the owner’s row-based shade-change counts: gray/teal264/333/333 (4/5/4 cells at≥32-channel difference), copper89/124/126/128 (0/0/4/3), bronze171/167/167 (0/1/1), red51 (8). Most changes are small; original stationary logical block means at4× differ by≥32 in only4/6/3 of543 gray/teal cells,0/0/4/3 of341 copper cells,2/1/0 of485 bronze cells, and0 of275 red cells. Full-phase visual review found coherent material/architecture rather than displaced parts. These limited shade variations are accepted and explicitly recorded, not silently treated as original-body invariance.

Exact source/raw/candidate/crop/built-in provenance, all75 PPM hashes, viewed-page hashes, indexed source-class/missing-cell evidence and phase measurements are bound by .reference/hd-full-review/tank-families-15/independent-review.json, SHA256 f6c62881aa1b2c8f035cd588b6d4e11b8b543af4a03a6a56616613afe6be7839. Immutable library is e8b521c4d9a6f2fdf087668c264487014925df129cdd4fa79959eb15286cf224; broader unrelated game-source freshness is true. This is isolated picture/raster acceptance, not a new native Host enemy occurrence, timing, terrain, simulation-neutrality or fullscreen claim. Global registration remains coordinator-owned; root800 handoff is unchanged.