Three cameras in, one live panorama out: how MAST Stitching Studio does it

Three cameras around a pitch, and you want one picture of the whole pitch, live. MAST Stitching Studio joins your cameras into one seamless panorama on your NVIDIA GPU, and sends it out as a live stream or a file.

ANPR on a phone at your gate: read every car, build your White List, send each plate to your system

MareArts ANPR Mobile reads number plates with the phone camera, on the phone itself, even offline. Here is how a car park or a gate uses it: read every car, learn which plates belong, and send each plate to your own system the moment it is read.

Myth: you need the backup app to open your backup. Fact: with BACK it is plain folders

Myth: you need the backup app to open your backup.

Fact: with BACK, your backup is plain folders. BACK copies your folders to another disk on a schedule, and any file manager on any machine can open the copies. No archive format, no database.

Where was this photo taken? Read it on your own computer with MIV

Where was this taken? Your camera and phone write it into the picture. MIV, the MareArts Image Viewer, reads it on your own computer and shows the nearest town, without sending the position anywhere.

Hide a moving car in a video: by hand in three clicks, or just ask your AI app

A car drives through your clip and you need to hide it. In VIDE, our video editor for Windows, macOS and Linux, a mosaic can follow the car as it moves, frame by frame. You can set it up in a few clicks, or ask an AI app on your computer to do it for you.

Moim beta testers wanted: your AI coding apps as one team, in one office

If you use Claude Code, Codex, Cursor, Gemini CLI or VS Code, each of them normally works alone, in its own window, and cannot reach the others. Moim is a desktop app that puts them in one office on your own computer, as one team. We are looking for early beta testers.

moim-pose: walk-cycle sprites from one picture

Open source · ComfyUI · Wan 2.2 · FLUX.2 [klein]

moim-pose (formerly marearts-spritestride) turns one side-view picture of a character into a walk cycle: 12 frames that loop, the same person in every frame, checked automatically, with the height and head size locked and a stride that keeps the feet from sliding. It runs on an 8 GB GPU through ComfyUI: a video model draws the whole cycle at once, an image model makes each frame sharp, and every model it uses is Apache-2.0 or MIT.

Rin and Otto, each drawn from one sentence, walking and carrying a box in 12-frame loops
Rin and Otto, each drawn from one sentence: a walk and a walk with a box, 12 frames each, the same person in every frame. The far leg is drawn darker. These are checked drafts: some frames still need a person's look and a redraw, see "It is not 100 %".

MareArts MSYS: the whole computer in one console, for you and your AI agent

Windows · Linux · macOS · MCP

MareArts MSYS is a free system monitor that shows the whole computer in one console: CPU per core, memory, GPUs, temperatures, fans, disks, network, containers and the busiest processes. Your AI agent can read the same numbers through MSYS's own tools, so you can simply ask it what is slowing the machine down.

MareArts MSYS side bar and console in VS Code, dark theme

NaN is not one bit pattern: sorting floats by integer keys

Floats · sorting · NaN

float("nan") and 0/0 both print as nan, but they are not always the same bits. A sort that never compares the floats, and instead ranks an integer key made from those bits, can give those NaNs a rank against each other. That breaks a stable rule: equal values keep the earlier column. Fold every NaN to one bit pattern before the key is built.

marearts-tinydeform: deformable attention in one CUDA call

Open source · CUDA · PyTorch

marearts-tinydeform is our first open-source CUDA kernel. It runs the value projection and multi-scale deformable attention of a DETR-style decoder (RT-DETR, D-FINE, DINO, Deformable-DETR) as one call, gives the same result as the usual PyTorch code, and is up to 4.7× faster than the original Deformable-DETR CUDA op when there are few queries. Apache-2.0, free for any use.

unittest.mock, from stand-ins to spies

Python · Testing

A mock is a stand-in you put where a real function or object used to be, for as long as a test runs. It answers the way you tell it to, it remembers every call it got, and when the test ends the original comes back. This guide covers both halves — controlling and recording — and then the part most tutorials skip: wraps=, a spy that lets the real code run.

A | B means either: reading union type hints in Python

Python · Type hints

def sensor_id(sensor: Sensor | int | None) -> int has three bars in one signature and looks like bit arithmetic. It is not arithmetic and it is not new syntax to memorise: each bar is the word or, and the whole annotation is one sentence — "a Sensor, or an int, or nothing".

MareArts ANPR App: On-Device Plates, MMC Vehicle Info, Team Sync, Web and Desktop Viewers

MareArts ANPR is a license-plate recognition app for iOS and Android, with a web viewer and a desktop viewer on the same account. Plates are read on the phone. Cloud MMC (make, model, colour, type, side, nation), team sync, rules, maps, and stats sit on marearts.com.

API and ABI, and the line between them

C · Python · Linking

Two acronyms one letter apart, describing promises made at two different moments. An API is what you agree to when you write the call. An ABI is what your compiled bytes agree to when they meet somebody else's compiled bytes — and unlike the API, breaking it usually produces no error at all.

The X-macro, from #define to runtime reflection

C · Preprocessor

C forgets the name of every field the moment it finishes compiling. The X-macro is how a C programmer gets those names back — by writing the list once and replaying it as an enum, a string table, a descriptor table, and a set of registration calls that run before main.

The list that remembers: mutable default arguments in Python

Python · Language

def append_to(item, target=[]) looks like it starts from an empty list every time. It does not. One list is created when the def statement runs, and every call for the rest of the program's life shares it.