🔥 Stage 0 算法仿真已通过
🔥 Stage 0 Algorithm Simulation Passed

让普通相机拥有
空间智能

Give Your Camera
Spatial Intelligence

多传感器融合 + 端侧 AI 推理 + Canvas 游戏级渲染输出——
一款便携模块，让 Vlog、VR 拍摄、影视创作全面升维。

Multi-sensor fusion + edge AI inference + Canvas game-grade rendering —
A portable module that elevates Vlog, VR, and film production to the next level.

申请早期体验 Apply for Early Access 查看 Demo View Demo
Algorithm Demo
Algorithm Demo
内行看门道，外行看热闹

以下展示传感器融合 + AI 推理的初步仿真结果，真实呈现 Stage 0 的技术能力。

Showing initial sensor fusion + AI inference simulation results — genuine Stage 0 technical capability.

Sensor Fusion Demo
Stage 0 仿真结果 Stage 0 Simulation

深度相机 + mmWave 雷达融合输出

Depth Camera + mmWave Radar Fusion Output

OAK-D-Lite 双目深度与 24GHz 毫米波雷达协同感知——室内环境中，即使有部分遮挡，设备仍能稳定检测人体位置与运动轨迹。

OAK-D-Lite stereo depth + 24GHz mmWave radar co-sensing — even with partial occlusion indoors, the device stably detects human position and motion trajectory.

9轴 IMU 轨迹 9-Axis IMU Trajectory

运动轨迹实时记录

Real-time Motion Trajectory Recording

ICM20948 提供高精度姿态数据，与视觉/雷达融合，输出稳定的6-DoF运动轨迹。适合第一人称 Vlog 和 VR 拍摄的场景重建。

ICM20948 provides high-precision attitude data, fused with vision and radar, outputting stable 6-DoF motion trajectory. Ideal for first-person Vlog and VR scene reconstruction.

端侧 NPU 推理 Edge NPU Inference

RK3588 实时场景理解

RK3588 Real-time Scene Understanding

6 TOPS NPU 在端侧完成 AI 推理——场景分类、物体检测、深度估计全流程实时，延迟低于 50ms。

6 TOPS NPU runs AI inference edge-side — scene classification, object detection, and depth estimation in real-time with under 50ms latency.

Hardware + Software
Hardware + Software
为什么是 VistaSense

不是单一传感器，而是多源融合产生的「空间理解」，这是单相机、单雷达永远无法实现的能力。

Not a single sensor — it's multi-source fusion that produces spatial understanding, something neither a single camera nor radar can achieve alone.

📷

深度相机

Depth Camera

Luxonis OAK-D-Lite
Intel Myriad X VPU
4 TOPS 算力

Luxonis OAK-D-Lite
Intel Myriad X VPU
4 TOPS

📡

mmWave 雷达

mmWave Radar

Seeed XIAO 24GHz
4-6m 人体检测
部分遮挡穿透

Seeed XIAO 24GHz
4-6m human detection
Partial occlusion penetration

🧭

9轴 IMU

9-Axis IMU

TDK ICM-20948
高精度姿态估计
运动轨迹重建

TDK ICM-20948
High-precision attitude estimation
Motion trajectory reconstruction

⚡

RK3588S 主控

RK3588S Main Control

6 TOPS NPU
端侧 AI 推理
无云依赖

6 TOPS NPU
Edge AI inference
Cloud-independent

🎨

Canvas 渲染

Canvas Rendering

推理结果直接驱动
游戏级实时构图
RK3588 GPU 加速

Inference drives rendering
Game-grade real-time composition
RK3588 GPU acceleration

🔧

开源生态

Open Source Ecosystem

硬件 + 固件 + SDK
MIT 许可
Skill 扩展体系

Hardware + firmware + SDK
MIT License
Skill extension ecosystem

Use Cases
Use Cases
三个真实场景

内行看门道，外行看热闹——三个具体的创作场景，看 VistaSense 如何改变拍摄方式。

Three specific creative scenarios — see how VistaSense transforms the way you shoot.

01

自拍时终于知道取景好不好了

Know Your Shot Framing in Real-Time

你举着相机，专注表演，但永远不知道画面里到底拍到了什么。VistaSense 实时分析你的机位——视角太窄、构图偏移、主被摄距离不对——它像是站在镜头后面的导演，在你拍摄时就告诉你"往左半步"、"退后一点"。你不需要在后期里才发现素材不可用。

You hold the camera, focus on performing, but never know what's actually in frame. VistaSense analyzes your position in real-time — whether the angle is too narrow, composition is off, or subject distance is wrong — like a director behind the lens, telling you "step left" or "back up a bit" while you shoot. No more discovering unusable footage in post.

第一人称 Vlog First-Person Vlog
02

6DoF 追踪，零后期

6DoF Tracking, Zero Post-Production

VR 直播最怕的是"观众看到的是抖动的碎片"。VistaSense 融合深度相机 + 9轴 IMU，实时输出 6DoF 运动轨迹和场景理解——你的每一个转身、每一个走位，画面都在正确的地方。

The biggest fear in VR live streaming is an unstable, fragmented view. VistaSense fuses a depth camera + 9-axis IMU to output real-time 6DoF motion trajectory and scene understanding — every turn, every movement, the frame stays exactly where it should be.

VR 直播辅助 VR Live Streaming
03

现场就知道镜头语言对不对

Know If the Shot Works — On Set

传统预监只能看画面有没有穿帮。VistaSense 在预监画面上叠加实时语义层——哪里是视觉焦点，哪里构图过密，哪里缺少纵深感——导演和摄影指导在拍摄现场就能判断：这个镜头能不能用。

Traditional on-set monitoring only shows what's in frame. VistaSense overlays a real-time semantic layer — highlighting visual focus points, detecting compositional density, identifying missing depth — so directors and DPs can judge right there: does this shot work?

影视预监 Film Pre-Visualization
System Architecture
System Architecture
三层闭环：感知 → 推理 → 渲染

借鉴 Aerial VLN 分层架构，设备从被动记录升级为自主选择最佳观测位置的具身智能节点。

Inspired by the Aerial VLN layered architecture — the device upgrades from passive recording to an embodied AI node that autonomously selects the optimal observation position.

▲ 高层规划 — LLM/VLM 语义理解、机位决策
▲ High-Level Planning — LLM/VLM semantic understanding, camera position decisions
◈ 接口层 — 目标位姿/航点指令传递
◈ Interface Layer — Target pose/waypoint command delivery

传感器融合

Sensor Fusion

Depth + Radar + IMU

→

AI 推理

AI Inference

场景理解 + 构图

Scene + Composition

→

Canvas 渲染

Canvas Rendering

最优取景输出

Optimal framing output

→

实时画面

Live Output

VR/Vlog 直出

VR/Vlog direct output

Pricing
Pricing
透明定价，开源生态

消费级

Consumer

$499 – $699
Early Bird 预售
Early Bird Pre-order
  • 多传感器融合模块 ×1
  • Multi-sensor fusion module ×1
  • 基础固件 + SDK
  • Basic firmware + SDK
  • 社区支持
  • Community support
  • 固件持续更新
  • Firmware updates
  • 开源硬件文档
  • Open-source hardware docs

专业级

Professional

$999 – $1,499
首批限量
First batch, limited
  • 高规格传感器套装
  • High-spec sensor kit
  • 高级 AI Skill 套装
  • Advanced AI Skill bundle
  • 优先固件更新
  • Priority firmware updates
  • 一对一技术支持
  • 1-on-1 tech support
  • 软件生态优先接入
  • Priority access to software ecosystem
Early Access
Early Access
成为第一批用户

告诉我们你是谁、你关注什么——我们正在寻找 VR 拍摄、Vlog 创作、影视制作领域的早期共创者。

Tell us who you are and what you care about — we're looking for early co-creators in VR filming, Vlog creation, and film production.