01
Understand
Find the real problem — not just its initial wording.
VIC·TORI
Human × AI thinking tested against reality.
If you need competitive advantages.
VIC·TORI investigates technical possibilities that may give your product capabilities beyond the expected.
VIKTOR
TORI
AT WORK
HOW WE WORK
Beyond the initial task.
Toward new capabilities.
We examine what is possible, challenge our own hypotheses and follow the evidence—even when the most valuable opportunity lies beyond the original question.
02
Reconstruct
Reveal the structure, constraints, signals and relationships that matter.
03
Hypothesize
Build competing explanations and possible solutions instead of committing to the first plausible one.
04
Build
Turn the strongest structure into the simplest working mechanism: algorithm, model, code or another concrete solution.
05
Test against reality
Measure what actually works. Keep what survives. Change what does not.
HOW WE WORK IN PRACTICE · REAL INDEPENDENT R&D CASE
From a visual defect to a new product model
A real investigation, not a fictional success story: analysis of a public interior visualiser, controlled experiments, mistakes and critical opposition. The work led to two focused engineering proposals—and a broader product-level concept.
Original evidence · Tests & revisions · Independent critique · Three proposals
We do not protect the first idea.We keep what survives contact with reality. Explore a new capability →
SELECTED WORK
Different domains.
The same engineering habit.
Find reliable information, recover hidden structure, and turn it into a working mechanism. The images below are real outputs and working prototypes; some projects have detailed pages, others are shown here only as compact evidence of the range.
SIGNALS · COMPUTER VISION · GEOMETRY · INTERACTION · REAL-TIME SYSTEMS · CREATIVE TOOLS
SIGNAL ANALYSIS · MEDICAL ENGINEERING
REAL ANALYSIS OUTPUTHolter ECG
Long, noisy recordings turned into a first-level expert pipeline: heartbeat detection, delineation, intervals, beat classes, rhythm analysis and diagnostic support.
COMPUTER VISION · PREPROCESSING
ORIGINAL → ENHANCEDUniversal Image Enhancement
A training-free visual-information layer for visible, IR, thermal and other imagery. The aim is not decoration: reveal weak structure before measurement or recognition.
MACHINE VISION · INDUSTRIAL INSPECTION
REAL INDUSTRIAL PIPELINEMachine Vision
Production-oriented vision pipelines: PET preforms, plastic-product defects, geometric recognition and other tasks where the final criterion is reliable inspection, not a pretty image.
GEOMETRY · PATTERN RECOGNITION
CUSTOM ALGORITHMSGeometry & Pattern Algorithms
Custom solutions developed when standard detectors were not enough: perspective rectangles and camera-relative horizon geometry, line and sphere detectors, N-dimensional curve approximation, and frame-by-frame 2D→3D reconstruction.
CONTEXT · COLLABORATIVE SYSTEMS
Contextual Interaction Systems
COMmix and WEBRTS: comments and reviews attached to the exact object, fragment or shared screen; comments can themselves become review objects, preserving the structure and history of discussion.
REAL-TIME AI · SYSTEM ARCHITECTURE
↓
VIRTUAL OPERATORVIRTUAL OPERATOR
↓
SYNTHETIC DIRECTOR → SCREENS
Artificial Broadcasting
A fully automatic broadcast from one wide-angle camera. Situation detectors feed virtual operators; a synthetic director selects and combines streams according to events, show logic and real-time constraints.
CREATIVE SYSTEMS · HUMAN INTERACTION
Gekibana / Creative Systems
A complete server + application for creating visual works at the level of intent rather than manual technique — without first mastering brushes, paints or the low-level mechanics of the medium.
FROM THE TECHNICAL ARCHIVE
Four more lines of inquiry.
Historical implementations and exploratory work, each with original evidence and a detailed account of what was actually built.
VISUAL REPRESENTATION
Visual Structure
From color fields and gradient lines to drawable objects.
Read case study →
PATTERN GEOMETRY
StarTagger
A relational signature for constellations of points.
Read case study →
DOCUMENT ANALYSIS
Technical Drawings
Discovering meaningful geometry in engineering sheets.
Read case study →
WEBRTC COLLABORATION
Ouité
Video communication as a shared workspace.
Read case study →OPEN R&D / PARTNERSHIPS
New Ideas.
Real Engineering Partners.
Selected directions grounded in our previous work, open for focused technical exploration and paid R&D collaboration. Each direction starts with a technical opportunity and a measurable question.
WORK WITH VIC·TORI
Independent engineering work, scoped validation, and a decision based on evidence — not a promise of results before a test.
Explore New Ideas & Partnerships →REALITY LAB
Reality is allowed
to say no.
Some ideas can look convincing for a long time when the answer is delayed or ambiguous. In the Reality Lab we choose environments where decisions can be frozen first and an independent outcome arrives later.
The purpose is not to protect a successful story. A miss remains visible; a failed hypothesis becomes evidence about the method.
Open Football / Reality Lab →CLOSED EXPERIMENT
ENG0
The public prediction experiment did not establish a robust predictive advantage. The frozen record remains part of the archive instead of being rewritten after the fact.
CURRENT RESEARCH
ENG1
Research continues on a different structure: replay, blind hypotheses, internal conflicts and new unseen games. Hits and misses are recorded as new evidence.
View October frozen testFREEZE→PREDICT→OPEN REALITY→LEARN
WORKING DIALOGUES
The reasoning is also part of the work.
Selected dialogues show hypotheses being proposed, attacked, revised and sometimes rejected. They are not a transcript archive; they are a window into how Human × AI work changes while the problem is still alive.
Read selected dialogues →Work with VIC·TORI
Your hardest technical problem may be your next competitive advantage.
Bring us an unresolved problem, a difficult constraint or an idea worth investigating. We challenge assumptions, examine the evidence and work towards practical engineering solutions.
Viktor
Independent engineering R&D, signal and image processing, computer vision and unconventional technical directions — from a specific limitation to a capability competitors may not have.
vic@vic-tori.ai
LinkedIn
Tori
A second line of reasoning: challenge assumptions, compare alternatives, and keep the argument connected to evidence.
tori@vic-tori.ai