iiinova — where your digital self keeps evolving

v1sea @v1sea.com · 22/09/2025

Coming down with the sickness again, maybe there are codebase visualizations that are useful. Interactive with bidirectional editor integration and runtime analysis. Apologies for the slop prototype.

230 3

v1sea @v1sea.com · 22/09/2025

Let me get this straight, it is a set of PNGs of an AI eagle head with different brown colored feathers than the base AI eagle body. Got it.

AI generated Eagle, head on with feathers that do not match the pattern or color of the body of the eagle

Browser debug visual for downloaded Eagle head single png on checkerboard background. Head is cut off from body. 1671x1300 px

030 3

v1sea @v1sea.com · 17/09/2025

Here is a VGGT example where all but one image is from daytime. The single nighttime image fails to integrate and is estimated to be in the free space to the left, outside the view of the daytime images.

010 1

v1sea @v1sea.com · 17/09/2025

The VGGT vision transformer model couldn't really handle mixed time of day images, but this new Apache-2 licensed model is a big step up. Variable lighting and weather really mess with existing VPS solutions. A popular one needs two scans at least 5 hours apart. map-anything.github.io

130 3

v1sea @v1sea.com · 05/09/2025

Butlerian Jihad at home. 2010s programming just to feel something.

Screenshot of host file reading.
"##
# Host Database
#
# localhost is used to configure the loopback interface
# when the system is booting. Do not change this entry.
##
127.0.0.1 localhost
255.255.255.255 broadcasthost
::1 localhost

127.0.0.1 claude.ai
127.0.0.1 api.anthropic.com
127.0.0.1 anthropic.com
"

010 1

v1sea @v1sea.com · 30/08/2025

Got the mesh registration to run in ~100ms, down from 400ms, by fusing the kernels and moving the correspondence work to the gpu. The extra time is from deserializing the test data. bsky.app/profile/v1se...

020 2

v1sea @v1sea.com · 29/08/2025

These are 4 of the Mapillary images

Image 1/7 total, corner adjacent to the large flower pot. tree partially obscures top right. Parking garage on left.

image 2/7 looking at intersection with flower pot to the right across street with parking garage on the left across street.

image 3/7 next to flower pot that is out of frame. Looking forward is stone building with parking garage on the right across street

image 4/7 with flower pot across street in front of us with stone building to the right. Few cars driving around.

000 0

v1sea @v1sea.com · 29/08/2025

This is the headset image.

Sidewalk looking at stone building on right with large flower pot on left.

100 1

v1sea @v1sea.com · 29/08/2025

The VGGT model works as a VPS with only a few images. One attention head estimates the 6dof camera pose of each image. Here is a rough example with 7 warped Mapillary images and 1 XR headset camera image in red. While the error is about 2m, you couldn't do this at all with SfM.

120 2

v1sea @v1sea.com · 14/08/2025

Here is 1/10 images used before unwarping. vgg-t.github.io

Screenshot of Mapillary UI of one of the ten images used in the reconstruction. It is of the Seattle Convention Center with the left side taken up by the canopy.

020 2

v1sea @v1sea.com · 14/08/2025

I got some very promising 3d sidewalk point cloud data from unwarping Mapillary data and feeding it to VGGT(Visual Geometry Grounded Transformer). The output seems much better than traditional Structure from Motion + Multi-View Stereo dense reconstruction.

1141 14

v1sea @v1sea.com · 10/08/2025

I think I've hit the limit for GPS accuracy in the city for AR positioning. One path forward could be doing mesh registration on a 3d model of the city. Maybe by running Structure from Motion on the street level Mapillary data. Maybe combined with USGS airborne Lidar.

Screenshot of Mapillary steetview image showing white building on left, red brick 20 floor building on right. Small map with routes in the bottom left and 4 way arrow in the bottom center.

USGS Lidar of the same location, clearly shows how little data there is at the street level as the buidling occludes them.

010 1

v1sea @v1sea.com · 09/08/2025

Ok, the GPS in red is so wrong here :(

100 1

v1sea @v1sea.com · 09/08/2025

Demo: while you were living in sin I was studying the GNSS.

110 1

v1sea @v1sea.com · 07/08/2025

I was really surprised by how well the tracking held up in the dark with visionOS 26.

010 1

v1sea @v1sea.com · 07/08/2025

Alleyway GPS sensor fusion test. Head position in white. GPS in red. Fused state in green. It is tuned for open fields and forests, but maybe with a bit more math this could hit sub meter accuracy. At this scale the odometry is very accurate.

120 2

v1sea @v1sea.com · 04/08/2025

I was surprised to find NOAA publishes the 3d weather forecast data for the continental US + Alaska. This appears to match observations of nearby mountains. Very little smoke today. I get why hiking apps don't provide search on this data now :), but there might be a way.

Map showing in blue the low clouds of the HRRR model. Mount SI appears covered by clouds.

Webcam showing 4:44PM PT Mount Si webcam, the peak is in the clouds.

110 1

v1sea @v1sea.com · 31/07/2025

Would anyone actually try a free outdoor navigation app for the vision pro? The tech works in the mountains, but it is too dangerous to recommend outside flat parks. My initial tests were putting a 10m tall pushpin at a gps coordinate and keeping it there for miles.

120 2

v1sea @v1sea.com · 30/07/2025

Demo showing Claude building a race car with voxels. The UI is all buttoned up now, hopefully app review approves this feature.

200 2

v1sea @v1sea.com · 27/07/2025

Today was such a beautiful summer day for a hike according to my database.

120 2

v1sea @v1sea.com · 24/07/2025

Working on a little website to find hikes. One feature is filtering by weather forecast. Very thankful for the open data from NOAA and the USGS. All the blue dots are the real-time positions of nearly 1000 buses and ferries.

Map of Washington and Oregon state Western sides with overlay of NOAA HRRR visibility data and nearly every Puget Sounds bus and ferry. Data show clear interior and low visibility on Western side of Olympic mountains.

010 1

v1sea @v1sea.com · 21/07/2025

Claude and I friendmaxxing at the beach.

130 3

v1sea @v1sea.com · 14/07/2025

Claude's little plan.

# BUILDING PLAN

## 1. Architectural Style Analysis
This request suggests a **Medieval European Castle** style, specifically drawing from Norman and Gothic fortress architecture. The emphasis on making it "BADASS" indicates we should focus on imposing defensive features like massive walls, towers, battlements, and a commanding presence that projects power and strength.

## 2. Historical Context
Medieval castles (11th-15th centuries) were primarily defensive structures designed to:
- Control territory and trade routes
- Provide refuge during sieges
- Project royal/noble authority
- Serve as administrative centers

Key features include:
- Massive stone walls with crenellations (battlements)
- Multiple defensive towers
- Central keep (main fortress tower)
- Gatehouse with portcullis
- Concentric defensive layers
- Arrow slits and murder holes

## 3. Material Palette
Authentic materials for medieval castles:
- **Primary**: Stone (grey stone, greystone) for walls and towers
- **Secondary**: Brick (brick grey) for inner structures
- **Accents**: Dark materials (charcoal, slate) for defensive features
- **Roofing**: Slate or dark stone for conical tower roofs
- **Foundations**: Gravel stone for base stability

## 4. Structural Elements
- **Proportions**: Vertical emphasis with towers rising 2-3x wall height
- **Walls**: Thick (3-4 blocks) with crenellated tops
- **Towers**: Cylindrical or square, with conical roofs
- **Central Keep**: Tallest structure at the heart of the castle
- **Gatehouse**: Fortified entrance with twin towers
- **Courtyard**: Open central space for activities

## 5. Implementation Strategy
1. Create foundation and base platform
2. Build outer curtain walls with corner towers
3. Construct imposing gatehouse entrance
4. Build central keep tower
5. Add defensive details (battlements, arrow slits)
6. Create interior courtyard features

---

# JAVASCRIPT IMPLEMENTATION

```javascript
// BADASS 150x150x150 MEDIEVAL CASTLE CONSTRUCTION

210 2

v1sea @v1sea.com · 14/07/2025

This is claude-sonnet-4-20250514 building a sand castle via the CRDT ops after 16 seconds of thinking.

120 2

v1sea @v1sea.com · 14/07/2025

Little faster. Now it does 4296.3 CRDT ops/sec (6x faster) and chunk meshing is 192μs (78x faster)

110 1

v1sea @v1sea.com · 13/07/2025

Little better with some tuning. 2256.8 ops/sec

000 0

v1sea @v1sea.com · 12/07/2025

WIP: performance test for the CRDT. Getting around 637.9 ops/sec with intermittent state compaction.

210 2

v1sea @v1sea.com · 08/07/2025

I think I can reuse the front end of the mesh registration system to detect when some mesh chunks have enough geometric detail to be found in the future. If this works well then I can automate the fine grain (<10cm) geospatial localization. Sunk cost fallacy is my moat.

020 2

v1sea @v1sea.com · 07/07/2025

Moved ISS to the GPU to join FPFH, now the full ISS + FPFH + TEASER pipeline runs in under 400ms for typical environments on Vision Pro. Red mesh is the prior saved one we are searching for in the blue mesh.

161 6

v1sea @v1sea.com · 02/07/2025

Due to economic headwinds beyond my comprehension, I'm pivoting to LLMs. I'm adding an LLM building assistant to my voxel editor. To support this the data layer is now a CRDT. Notice how only the llm edits are rolled back.

140 4

v1sea @v1sea.com · 26/06/2025

got SharePlay integrated for volumetric windows, working on testing immersive spaces before shipping. I was able to abstract over the hub-and-spoke tcp/udp transport and the SharePlay peer-to-peer transport. Due to better user experience SharePlay gets priority.

020 2

v1sea @v1sea.com · 24/06/2025

IVAS at home. Another ads-b tracked takeoff displayed in the vision pro. This feature seems worth polishing up and shipping.

030 3

v1sea @v1sea.com · 24/06/2025

Testing: real-time ads-b overlay via pi->(ble)->vision pro. Soon my app will offer Full-Spectrum intelligence of the urban theater.

010 1

v1sea @v1sea.com · 19/06/2025

I added 70 solid colors to the palette. My next priority is shipping SharePlay and making the outdoor modes more intuitive.

010 1

v1sea @v1sea.com · 10/06/2025

Debug vis from walking around alleyways with visionOS. GPS in red, odometry(head position) in white, fusion in orange/green. The main issue is deriving the north heading without a compass. Urban areas have significantly worse gps noise making state convergence hard to maintain.

010 1

v1sea @v1sea.com · 10/06/2025

WIP: bus-ray-vision. Building on my gps localization system for visionOS, I render the real-time position of all Puget Sound transit. Seems more like a novelty than worth shipping tbh.

061 6

v1sea @v1sea.com · 04/06/2025

I'm really impressed by the results. This makes it significantly easier to learn some new domain of algorithms. Still gotta read and implement the papers, but this is so much better imo.

Screenshot showing Claude Research for vehicle position estimation. It lists 438 sources. It suggests Extended and Unscented Kalman Filters along with HMM-based map-matching algorithms.

010 1

v1sea @v1sea.com · 23/05/2025

tired: deleted and replace blocks wired: push em

070 7

v1sea @v1sea.com · 17/05/2025

I've published the first version of the ClassiCube port to visionOS. The in game wrist menu is not complete, but you can place and delete blocks. It adds a Metal render backend and then bridges to the AR Compositor. github.com/v1sea/MacroC... I might try to get Bevy to support this next.

1221 22

v1sea @v1sea.com · 17/05/2025

If so it doesn't appear in the gpu trace. I get the same rasterization rate map when looking different directions. This is all v2.5

Timeline view visionOS metal renderer. Shows two command buffers with one post process pass each with two events.

Zoom in on single command buffer in timeline view.

explicit timeline view showing execution time of vertex, fragment, and compute kernel.

110 1

v1sea @v1sea.com · 15/05/2025

I'm considering adding my custom map renderer with the voxel world overlay(needs a bit more polish). I'm also actively hunting for a backend or AR software role in the Seattle area. DMs are open. Thank you.

1183 18

v1sea @v1sea.com · 07/05/2025

Sometimes the origin move will be vertical only. Other times the new origin is at the users feet. Video is of sensor replay from 3k elevation gain. Notice the two separate pillars of reference frames.

000 0

v1sea @v1sea.com · 07/05/2025

Wrist compass is now exposed in public build, so if you walk 100 feet or so it will appear. It does long distance drift compensation, handles OS origin recenters, recovers from tracking loss. It is mostly an AUKF fusing odometry and GNSS from a phone anywhere on you..

users forearm with AR compass pointing nearly North with real compass on the ground. Ground is grass with few white flowers.

120 2

v1sea @v1sea.com · 25/04/2025

Demo: 2025 starter home. I've shipped a ~4x faster version of the relocalization pipeline by moving the most costly component to the GPU for a typical time reduction from 3s to 100ms.

020 2

v1sea @v1sea.com · 13/04/2025

Here is the full animation cycle for both hands. I think this is good enough to ship.

000 0

v1sea @v1sea.com · 13/04/2025

new menu, there when you want it.

141 4