Development LogUpdated July 18

Straight from
the build.

Games, applications, and immersive work — written up as it happens. Newest first, always.

Development logGamesJuly 18, 2026 · King Summon

Creature roster hits twelve, each with a full ability kit

The roster is finally deep enough to test real counterplay — twelve creatures across three combat roles, and the first synergies that genuinely surprised us in playtest.

Also newAll recent →

AnnouncementJul 15

Playtest sign-ups open for King Summon

First external build goes out to Discord members this month.

Product logJul 11

Portal ships every internal build automatically

Push a build, testers get it in minutes, feedback attaches to the version played.

Development logJul 07

Summon queue reworked after playtest feedback

Turn order is readable now — players stopped losing to the interface.

AnnouncementJul 03

Two artists join the team

Character and environment work moves in-house from this sprint on.

Jump toXR · 5 Recent · 9 Studio · 7

21In these sections

4Projects in build

XR

Notes from the headset.

Everything we learn building for immersive hardware — runtimes, input, and the ideas that didn’t survive contact with a headset.

Our XR stack

Featured update

TechnologyJuly 09, 2026·4 min read

One build, two headsets: the OpenXR sandbox runs

A single codebase now targets two runtimes with no per-platform forks. Hand tracking and room-scale locomotion are next on the bench.

More from the XR track

Research & engineering notes

View all

01

Capture artwork

ResearchJun 26, 2026·5 min read

Hand tracking without a controller fallback

Pinch, grab, and point carried the whole test build. What broke was confidence, not accuracy — players need to see the hand register before they trust it.

02

Capture artwork

EngineeringApr 08, 2026·7 min read

The 90fps budget on standalone hardware

Where the frame time actually goes on a mobile chipset, and the four passes we cut to hold a steady 90 without dropping visual density.

03

Capture artwork

ResearchMar 22, 2026·5 min read

Passthrough as a design material, not a gimmick

Mixed reality stops being a novelty the moment the room matters to the mechanic. Three prototypes where it did, and one where it did not.

04

Capture artwork

TechnologyMar 05, 2026·6 min read

Why we chose OpenXR over per-platform SDKs

One integration, several runtimes, and a smaller surface to maintain — plus the tradeoffs we accepted to get there.

Recent

What we’ve shipped lately.

AllDevelopment logGame updateTechnologyAnnouncement

Key art

Game updateJun 30, 2026

Number Forest gets a full lighting pass

The Fibonacci canopy finally reads at a glance — and puzzle sightlines survived the rework.

Key art

Development logJun 20, 2026

Combat prototype milestone

The core summoning and battle loop is playable internally. Feedback is already reshaping pacing.

Key art

Development logJun 16, 2026

Battle pacing: turn timers and the surrender rule

Two small rules cut the average match from nineteen minutes to twelve without losing depth.

Jun 10, 2026

First realm complete, start to finish

Number Forest is our first fully explorable mathematical environment.

Game update

May 28, 2026

Ability tooltips rewritten from scratch

Players were losing fights to unclear wording, not to bad decisions. Every tooltip now states its numbers plainly.

Game update

May 21, 2026

Nngtw Portal enters internal beta

Build distribution and playtest feedback in one place, dogfooded daily.

Product log

May 15, 2026

OpenXR integration research

Evaluating cross-platform XR across Unity and Unreal.

Technology

Apr 30, 2026

What room-scale means for a strategy game

Notes from three failed prototypes, and the one spatial control scheme that stuck.

Research

Apr 24, 2026

Puzzle authoring tool cuts build time in half

Designers now assemble a realm end to end without an engineer in the loop.

Engineering

Load older updates

Studio

The studio itself.

About Nngtw

AnnouncementJune 12, 2026

The studio moves to a four-project pipeline

Two games, one application, and one XR research track — each with a named owner and a fixed monthly review. Here’s how we decided what to run in parallel, and what we deliberately said no to.