Should you try Arnis?
Use Arnis for a small trial of a place you want to build in Minecraft. Arnis generates Minecraft worlds from OpenStreetMap and elevation data. Its documented workflow starts with selecting a rectangle on a map, then choosing generation settings. That makes it worth evaluating for a geographic starting map, while the quality of your particular location still needs inspection.
Our evidence supports a narrower conclusion than a finished-world review: the source installed, compiled and completed its test command in our sandbox. We did not generate a region, launch the GUI, open a world in Minecraft or measure generation speed. This correction uses the existing run; it does not report a new execution.
What our October sandbox run established
In our run on October 3, 2026, at commit 8e3fac4, installation took 109 seconds, the build took 421 seconds and the test command took 201 seconds. The harness records command success, exit status, elapsed time and timeout status separately from parsed test counts. Each command returned zero and did not time out.
The retained compiler output says the build used the dev profile with debug information. Treat the build duration as source-setup evidence, rather than a release-build or world-generation benchmark.
The attached stored parser summaries report 2736 passed, 0 failed and 2736 total. However, the retained log repeats the same Cargo summary before and after a replay marker. Those stored aggregates may count results more than once; they do not establish a unique or exhaustive test total. The raw footer also includes ignored tests. Ignored tests are not failures, and the aggregate skipped-test field is unknown.
The successful command is useful evidence that this checkout completed its automated checks in this environment. It does not establish what those checks cover or whether a particular exported map works. There is no recorded security audit, and scan coverage is unknown.
The MrKeyoor testing methodology explains how command outcomes and parsed counts reach the review. That page supplies provenance context; its current revision does not establish which procedure version governed this older run.
One World changes the expansion decision
At the tested commit, the README documents One World mode for extending a persistent Java world. It says successive areas share a Web Mercator frame, while the first run fixes scale, terrain mode and build height; rotation is off. The earlier page's blanket recommendation to avoid Arnis for persistent expansion therefore does not fit the supplied commit-specific documentation.
That documentation establishes an available workflow, not a result from our sandbox. We did not test adjacent-area joins, repeated generation or preservation of player-built structures. Before using it on a valued world, test adjoining areas and repeated runs on a disposable copy.
The commit-specific README lists Java Edition 1.17+, Bedrock Edition and Luanti. The supplied current README lists Java Edition 1.21.4+ instead. Check the requirements for the exact build you download rather than treating either statement as a compatibility result from our run.
Choose generation inputs before enlarging the area
The tested commit documents three modes. The default geo-terrain mode combines mapped objects with elevation. Geo-only places OpenStreetMap objects on flat ground. Terrain-only generates elevation terrain and skips the OpenStreetMap query and Overture fetch.
Use that separation to make a trial easier to inspect: choose terrain-only when terrain is the question, or include objects when roads and buildings matter. Compare the result with the place you intended to recreate before enlarging the selection. Our build and test timings cannot predict the cost or accuracy of that larger job.
Which alternative fits the task?
TerraPlusPlus is an alternative for a modded Earth-generation workflow that requires Forge, CubicChunks and CubicWorldGen. Its README describes public terrain datasets, OpenStreetMap roads and teleportation by latitude and longitude. Choose it when that mod setup fits your intended world. Its existence is no longer a reason to claim Arnis lacks a persistent-world option, and we have no comparative performance or accuracy measurements.
WorldEdit is a Java Edition in-game editor with selections, schematics, copy and paste, brushes and scripting. It is a possible companion after geographic generation when you need to change blocks. Choose it directly when editing an existing world is the task; it does not replace Arnis's documented geographic-input workflow.
How the ratings are assigned
For assessed dimensions, the rubric is: 1 means a recorded blocker, 2 means substantial documented friction, 3 means a documented route with limited validation, 4 means a documented route plus successful relevant sandbox checks, and 5 requires broader validation across the intended workflow. These are editorial ratings, not benchmark scores.
Setup ease receives 4 because a download route is documented and source setup succeeded in our sandbox, although the desktop application was not launched. Documentation receives 3 because the supplied README explains the basic workflow and generation modes and points to a wiki; the full wiki was not evaluated. Community support quality and operational maturity remain unassessed. Repository attention and a successful test command do not resolve those questions.