Drive the dormant overrideTexture path from host hi-res texture lookups - #5
Open
vjravenlight wants to merge 2 commits into
Open
Drive the dormant overrideTexture path from host hi-res texture lookups#5vjravenlight wants to merge 2 commits into
vjravenlight wants to merge 2 commits into
Conversation
The DR_PSYX_TEX / overrideTexture mechanism (TF_32_BIT_RGBA split + direct-sample shader) was never emitted by any host. Feed it per textured prim from HiresOverride_LookupByTpageClut (weak stub here; strong table lives host-side), so registered hi-res replacements actually draw: - ApplyHiresOverride() before AddSplit in POLY_FT3/FT4/GT3/GT4 and SPRT/SPRT_8/SPRT_16. textureId is already part of the split key, so batches open/close exactly at matching prims. - overrideTextureWidth/Height carry the ORIGINAL TIM native pixel size, so tpage-relative UVs map 0..1 across any upscale factor. - Reset the override in ClearSplits so it cannot leak across frames. - gte_shader_32_rgba: discard alpha < 0.5 — PSX colour-0 transparency for overrides (opaque prims ignore blending, holes must discard); authored soft alpha above the cutoff still blends on semi-trans prims. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Prim UVs restart at each tpage, so a TIM spanning several tpages (320px backgrounds, the boot logos) drew every chunk sampling the override from x=0 - the image duplicated per chunk. The lookup now also returns the tpage origin position inside the replaced TIM (native texels); it rides the split in drawenv.tw.x/y (tw.w/h already carry the size), is part of the split-keep key so two chunks of the same override cannot share one uniform, and lands in the 32-bit shader as u_texOffset added to the UVs before texelSize normalization. Removes the known v1 limitation from the original hook commit. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Author
|
Follow-up pushed: the known v1 limitation (multi-tpage surfaces duplicating the override per chunk) is now fixed - the lookup returns the tpage origin's offset inside the replaced TIM and the 32-bit shader adds it via a new u_texOffset uniform (split-keyed so chunks can't share one uniform value). Builds clean, boots and runs; in-game visual pass on a wide background pending. |
SlickAmogus
pushed a commit
that referenced
this pull request
Jul 10, 2026
…lookups Port of PR #5 onto current master. Textured prims (POLY_FT3/FT4, GT3/GT4, SPRT/8/16) now consult a host-provided HiresOverride_LookupByTpageClut (weak stub here, so non-SH hosts link unchanged) and route through the existing TF_32_BIT_RGBA override path when the host registered a replacement for the prim's tpage/clut. - overrideTextureOffsetX/Y + u_texOffset uniform: shifts tpage-relative UVs when the prim's tpage sits partway into the replaced TIM (multi- tpage surfaces draw as several prims whose UVs restart per tpage). - Splits additionally key on the offset (carried in drawenv.tw.x/y, which were dead for splits; tw.w/h were already repurposed) so two chunks of one override texture can't batch into one offset uniform. - gte_shader_32_rgba: alpha<0.5 discard for PSX colour-0/PNG-hole transparency on opaque prims; soft alpha >=0.5 still blends on semi-transparent prims. - Beyond the PR: DR_PSYX_TEX packet state is tracked separately and restored on lookup miss / ClearSplits, so that path keeps its exact persist-until-changed semantics instead of being zeroed per prim. Byte-identical when the host registers no overrides: lookup returns 0, all override state stays 0, split boundaries and bound shaders unchanged. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FuUfmVsV1UrMpoBqRmUhMv
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Companion to SlickAmogus/silent-hill-decomp's hi-res texture override system (
pc_port/src/hires_override.c): the game has been registering hi-res replacement textures since that file landed, but nothing ever drew them —HiresOverride_LookupByTpageCluthad no callers.PsyCross already contains the perfect delivery mechanism: the
DR_PSYX_TEX/overrideTexturepath (forces aTF_32_BIT_RGBAsplit and binds a plain RGBA texture with the direct-sample shader) — but no host ever emits that packet, so it was dormant. This PR feeds it from the host's override table:ApplyHiresOverride(tpage, clut)beforeAddSplitinPOLY_FT3/FT4/GT3/GT4andSPRT/SPRT_8/SPRT_16. SincetextureIdis already part of the split key, batches open and close exactly at matching prims — no batching rework.overrideTextureWidth/Heightare fed the original TIM's native pixel size, so the prim's tpage-relative UVs map 0..1 over the replacement regardless of its upscale factor.ClearSplits()resets the override so it can't leak across frames.gte_shader_32_rgbagainsif (fragColor.a < 0.5) discard;— PSX colour-0 transparency for overrides. Opaque prims ignore GL blending, so transparent texels must discard; authored soft alpha ≥ 0.5 still blends on semi-transparent prims (SRC_ALPHA carries the STP-style 50% blend).Tested end-to-end with the SH1 port: a PNG override registered for the boot-logo TIM renders in-game (see the game-side PR in silent-hill-decomp, which adds PNG loading + discovery).
Known v1 limitation: a surface wider than one tpage (e.g. a 320px background drawn as two tpage-sized prims) samples the whole override from each prim (UV origin per tpage, not per TIM). Single-tpage textures — items, characters, HUD sprites — map correctly. A follow-up can thread the tpage→TIM origin offset through a uniform.
🤖 Generated with Claude Code