BeamNG.Earth
A 1:1 scale streamed Earth for BeamNG.drive: real elevation and satellite imagery, floating origin, LOD streaming from a companion server
Explore Kingdom Come: Deliverance's world in Kingdom Come: Deliverance II. Experimental converter setup, feature status, tutorials and validation boundaries.
Translation pending
GamesPython · LuaGitHub @SamG-CoderOnlineOwner unconfirmed
Have you tried it? One tap is enough. No sign-in needed.
Experimental compiled KCD1 terrain and material converter for the Kingdom Come: Deliverance II development runtime.
Taken from the repository README
Loading the full README.
Read the full README ↗Explore the world of Kingdom Come: Deliverance in Kingdom Come: Deliverance II.
An experimental world and gameplay converter built around the assets in your installed games.
Website · Download setup · Installation guide · Feature checklist · Report an issue
Development screenshot supplied on 22 October 2026. The exact build was not identified; this visual snapshot does not establish release-wide gameplay acceptance. See the checklist below for scope.
The full frame and original game HUD are retained in these web-sized copies. See screenshot provenance.
Follow development on YouTube and X / @SamGCoder.
KingdomComeGlueMapper reads compiled KCD1 and KCD2 archives, converts the original landscape and assets, and builds a separate region for KCD2. The combined setup packages the world, coach travel and supported gameplay conversions into one mod.
The goal is to preserve KCD1's authored world while using compatible KCD2 runtime systems. Buildings and terrain retain their source placement; interactions receive native contracts where implemented. Wildlife uses KCD2 models, animations and AI at converted KCD1 habitats.
The main Rataje world is imported geographically. Complete KCD1 gameplay, campaign progression and every interactive object are still being converted. Separate source maps and every DLC or alternate quest state are not covered by a whole-game completion claim.
The checklist describes source implementation, with dated development-build observations below. It is not a release manifest or a whole-game acceptance result. See the release, source and runtime status for the 22 October 2026 source review, newer gameplay work and its remaining checks. Historical counts retain their original observation dates.
The normal travel route is Launch mod → Continue a Trosky save → coachman → Travel to the Rattay region. Setup also connects the bounded Skalitz opening and selected gameplay conversions to its build recipe. This does not complete the KCD1 campaign. KCD1 saves are not converted.
SHA256SUMS.txt and self-test.json. See the 0.3.28 release notes for changes since the last published release, contributor credits and validation, and the KCD1:MP guide for the separate experimental multiplayer export.Updating or resuming? Choose Activity → Restore / upgrade build… (called Resume failed setup… in older builds) and select the original job folder containing job.json. Supported completed recipes migrate through the shared upgrader without a full map rebuild. Verified current packages and compatible finished lighting tiles are reused. Stale LOD validation can be repaired as a single section using verified saved build data, without rerunning completed gameplay or lighting stages. The clothing repairs are part of normal Setup conversion and saved appearance upgrades. Keep the original build workspace available. If map conversion never produced a complete package, that phase restarts. Logs and partial outputs are retained.
Current source also upgrades introduction and training handoffs, pub cleaning and resting, source-selected dice work, and supported voiced market calls. These use authored actor assignments, work windows and contacts. The work ownership audit records assignments belonging to other campaign profiles and unresolved coverage. The guides distinguish package validation from the remaining in-game checks and source behaviors awaiting conversion.
Setup v0.3.28 includes parallel tile baking and fresh NPC appearance performance improvements. See the performance evidence and limits for hardware, measured stages and release scope. These measurements are not a duration guarantee for the downloaded installer, another bake mode or your machine.
See the Windows setup guide for requirements, instructions, recovery and rollback, and the launcher details for each mode and current limitations.
Use Windows and Python 3.13 for the setup development environment:
git clone https://github.com/SamG-Coder/KingdomComeGlueMapper.git
cd KingdomComeGlueMapper
py -3.13 -m venv .venv
.venv/Scripts/python.exe -m pip install -r requirements-setup.txt
.venv/Scripts/python.exe tools/setup_app.py
Python 3.14 also works with the Visual Studio C++ build tools installed:
the pinned xatlas==0.0.11 has no cp314 wheel and must build from source.
Python 3.13 avoids that compilation through its Windows wheel. See
contributing for setup, local tests and issue reports.
To build a distributable setup executable:
.venv/Scripts/python.exe tools/build_setup_exe.py
The executable and self-test report are written to outputs/release. Build its GPU payload first as described in the Windows setup guide. KCD1 Map builds, installs, and resumes require the official KCD2 Modding Tools. Older build_*_probe.py commands are focused experiments with specific baseline requirements, not substitutes for the combined setup workflow.
How to read this checklist: a checked item means the described conversion exists in the current code. It does not mean every instance has passed in-game acceptance. Unchecked items are remaining work, including validation of implemented features. Runtime evidence and limitations are stated separately.
Waterfall shader materials.Evidence: terrain/buildings/vegetation and selected ground repairs have in-game confirmation. Water-layer repair preserves retained records, and the flow-speed correction targets Waterfall materials specifically. It is not a universal adjustment to every water shader. Lighting conversion exists, but full visual parity remains open.
Interior exposure adaptation is corrected across the imported map's time-of-day profiles, with user confirmation in a sampled house. Broader lighting validation remains open; see interior lighting.
Environment coverage: the installed conversion includes 1,307 particle emitters, 21 fog volumes, 1,355 audio entities and 931 linked audio boundaries across the main map's selected initial world. Missing KCD1 sounds and reverbs use explicit native KCD2 equivalents; these are semantic substitutions, not identical recordings. The corrected build reaches gameplay, resolves all 1,190 nonempty imported play-trigger references in the live engine, and received user confirmation on 13 October 2026. This is map-wide conversion coverage, not inspection of every individual effect. Quest/later-state layers and separate maps remain outside this pass. See the environment conversion audit for exclusions, validation and the reserved player-ID fix.
Sources: terrain converter, static assets, brush repair, water selection, water timing, lighting, voxel packaging.
Music coverage: 119 original KCD1 scenes and 80 area bindings are installed for the main map's supported initial world. Runtime probes verify Skalitz, its inn, countryside, Rattay, Talmberg and leaving the imported-world boundary. Original town, pub and countryside scene playback is visible in the native debug overlay. Full listening acceptance, quest music and native-level return travel remain open. See the music conversion audit.
Container coverage: the installed development build registers 2,769 converted stashes and preserves nine shared personal chests. All 174 generated loot tables include registered item names and GUIDs (2,208 entries); the user confirmed starting-home fruit sacks work after this fix. Containers are temporarily unowned. Merchant/NPC-generated stock, campaign transfers, wearable fit and persistence remain separate acceptance work. See the container conversion audit.
Evidence: seating, herb picking, washing, loose-item pickup and the selected rental/bed fixes have sampled user gameplay confirmation. The inspected development level contains 274 washing triggers and 303 cooking groups; the pickup conversion report records 1,773 placements. These counts are coverage snapshots, not individual acceptance tests.
World beds: the development build adds 1,385 sleep contracts and preserves the two existing rental/miller beds. A sampled raised bed completed a one-hour native sleep cycle; a ground bed opened the native sleep menu with its lower comfort rating. Other families, placement clearance and save/load remain acceptance checks. See the bed conversion evidence.
Ladders: the installed development build contains 113 native ladder entities with their static duplicates removed. The user confirmed climbing works at the sampled test location on 13 October 2026. Other locations and NPC climbing remain separate acceptance checks.
Sources: chairs, benches, cooking, washing, herbs, pickups, player chests, doors, ladders, rental.
Evidence: runtime sampling found native boar, hare, red-deer doe/stag and roe-buck actors. Two boars changed position between samples. The 752 figure is configured population, not a simultaneous active-actor count. Hare areas use native regional populations rather than an exact reproduction of KCD1's player-centred spawning.
Sources: wildlife converter, wildlife scope and validation, navigation, character conversion.
Evidence: coach access, map display and selected retail services have recorded gameplay confirmation. Registering native combat, shop, repair or sharpening graphs does not establish every actor or world object is wired to them. The full original campaign remains unfinished.
Sources: combined setup backend, region builder, gameplay graph registration, campaign tools.
Current boundary: the development test loads all 109 source objective states, awards both Speech choices, advances the original opening objectives and returns to gameplay with correct actor placement. Speech 7 survived a quicksave/reload without replaying the opening conversation. Opening journal text and region are now confirmed in game. Other stat branches, quest markers and subsequent interactions remain unverified; this is not a completed character-creation or campaign conversion. The ordinary setup still uses coach travel. See the quest conversion checklist for implementation milestones and per-quest acceptance requirements.
The following snapshot compares source mission/layer entities selected by the converter's initial-layer filter with the installed development level on 13 October 2026. Source identities are deduplicated. These are conversion targets, not instructions to force every quest-state object into the live world.
| Area | Source audit | Current conversion gap |
|---|---|---|
| Storage | 2,778 Stash entities | 2,769 converted stashes registered at runtime; nine shared personal chests preserved. Sampled storage/fruit-loot tests passed; ownership, special-service contents and persistence remain open. |
| Fish traps | 63 stashes using three fish-trap model types | Included in native storage conversion; water/fish baskets are additional variants. Specialised behaviour and replenishment remain unverified. |
| Nests | 103 | No converted nest registrations in the inspected level. |
| Digging | 79 holes: 63 grave pits and 16 buried sacks | No converted digging registrations. |
| Ambient fish | 41 fish-group entities | Separate from trap/container interaction; not converted. |
| Ladders | 113 | Converted to native climbing entities; sampled player test confirmed. Another 26 are deferred with inactive layers. |
| Grindstones | 11 | Global sharpening service registration does not activate these instances. |
| Alchemy benches | 7 | Complete native activity conversion remains. |
| Shooting targets / dice cups | 8 / 13 | Activity and participant conversion remains. |
| Beds | 1,387 | 1,385 new native sleep contracts plus two preserved existing beds; 438 inactive-layer beds remain deferred. Family alignment and persistence validation remain. |
Some KCD2 systems use different entity classes, so class counts alone are not proof of missing functionality. These gaps were cross-checked against the converter implementations. Counts describe the inspected source/build, not every game version or a published release guarantee.
Installed KCD1 archives + installed KCD2 reference assets
|
Terrain / geometry / material conversion
|
Native gameplay contracts and dependencies
|
World + travel package verification
|
Mods/kingdomcomegluemapper
The setup's conversion plan builds the world, packages its dependencies, calls the region builder for gameplay/travel, and combines the result. Source transforms and identities are retained or explicitly remapped; unsupported dependencies are reported rather than assumed compatible.
| Entry point | Responsibility |
|---|---|
| setup_app.py / desktop_setup.py | Setup interface, preflight, build orchestration and installation. |
| setup_campaign.py | World packaging, registration and campaign-oriented developer commands. |
| region_travel.py | Destination lifecycle, travel, gameplay converters, lighting and wildlife. |
| travel_bridge_package.py | Combine world/travel packages and apply supported compatibility repairs. |
| tests / CI workflow | Regression checks, Python compilation and selected CLI entry points. |
Run the test suite from the source environment:
.venv/Scripts/python.exe -m pip install -r requirements-test.txt
.venv/Scripts/python.exe -m unittest discover -s tests -v
The Validate tooling workflow checks Python 3.11 and 3.14 on Windows. It runs on manual dispatch (Actions → Validate tooling → Run workflow), not on push, so it does not gate a pull request; run the suite locally before opening one. It does not have the games installed and cannot verify appearance or gameplay. Local binary checks, frozen-executable checks, development-client tests and retail gameplay are separate validation steps. The Python 3.14 environment needs a C++ toolchain for xatlas, as described under Run the current source.
The Quest & NPC Tests emulator runs converted quest
and controller logic offline, with deterministic timers, explicit engine
outcomes, save/replay, and searchable HTML/JSON/JUnit reports. Start with
.venv/Scripts/python.exe tools/quest_lab_test.py demo --output outputs/quest-lab-demo.
It is separate from consumer Setup; unsupported operations remain visible and
native gameplay still requires engine verification.
The Clothing & Body Test workbench provides a standalone 3D viewer and repeatable audits for male/female clothing, layer combinations, and all KCD1 soul records. It is a developer tool, separate from the consumer setup application. Launch it with .venv/Scripts/python.exe tools/clothing_lab_app.py.
Historical documents contain dated experiment results and commands. Use this checklist for current scope, the current code for implementation, and release notes for a particular downloaded build.
For a useful issue report, include the converter commit/release, game version, location or coordinates, a screenshot and reproduction steps. State whether the failure concerns appearance, collision, interaction or persistence. Attach relevant log excerpts or report files without distributing game archives.
Known visual follow-up: shallow mud puddles lose their wet surface over imported ground (#18). The physical-surface correction and the missing wet appearance are separate issues.
The original tooling is released under the MIT license. Game assets, trademarks and screenshots belong to their respective rights holders and are not covered by that code license. Third-party tool notices are retained under third_party.
No game archives, extracted textures, meshes or generated world packages are distributed by this repository. Conversion requires your own installed games. KingdomComeGlueMapper is an independent experimental project, not an official Warhorse Studios or publisher release.