NEXVDO SDK — CAPTURE

Frames in GPU memory, in your first session.

Ingest capture cards, multimedia files, and network cameras. Move frames through GPUDirect DMA, then render, compose, or hand them to the next module.

Request an evaluation See Capture capabilities

Illustrative sketch - not a published API

source = capture_card | multimedia_file | network_camera
frame = dma_capture(source, destination = gpu_memory)

adjust(frame, brightness, hue, contrast, saturation)
transform(frame, scale, crop, rotate)
output(frame, render_or_encode)

Capture cards · files · network cameras

GPUDirect DMA to GPU memory

Windows · Linux · embedded

CAPTURE CAPABILITIES

Move the frame. Keep control.

GPU PATH

Capture into GPU memory

NVIDIA GPUDirect enables high-performance DMA capture direct to GPU memory. It removes a later memory-transfer step from the pipeline.

RENDER + COMPOSE

Shape the frame before output

ThumbDraw supports multi-screen rendering. Adjust brightness, hue, contrast, and saturation. Scale, crop, and rotate with CPU or GPU processing.

WORKFLOW DEPTH

Carry overlays and imaging data

Use ChromaKey, OSD, layouts, and XML-driven animation. DICOM, PACS, WORKLIST, and HL7 support clinical imaging workflows.

PUBLISHED SURFACE

Three input paths. Four referenced frameworks.

YUAN publishes the input classes, framework surface, and operating-system surface. The per-input compatibility mapping has not been published.

Capture input paths × framework × operating system. Unpublished mappings remain explicit.
Input path	Framework	Operating system
Capture cards	Input-to-framework mapping — pending publication	Per-input OS support — pending publication
Multimedia files	Input-to-framework mapping — pending publication	Per-input OS support — pending publication
Network cameras	Input-to-framework mapping — pending publication	Per-input OS support — pending publication
Published module surface	DirectShow · Media Foundation · V4L2 · GStreamer	Windows · Linux · embedded
Resolution and format matrix	Resolution matrix — pending publication	Format matrix — pending publication

SOURCE TO OUTPUT

Keep the frame on the fast path.

The capture path starts with a card, file, or network camera. GPUDirect DMA places the image in GPU memory. The frame is then ready for rendering or encoding.

YUAN's Capture module page states: “only 4 APIs, allowing you to capture video and audio within just five minutes.” The API reference itself is pending publication.

Pair with YUAN hardware →Medical imaging workflows →

01Source

02DMA

03GPU memory

04Render / encode

Illustrative Capture pipeline from source to render or encode

EVALUATION ARTIFACTS