Probe Report Parameters
Complete reference for Probe Report output parameters
Overview
The Probe Report provides canonicalized, ML-enhanced metadata by analyzing and reconciling outputs from both FFprobe and MediaInfo. This reference documents all available parameters in the Probe Report response.
Probe Report uses machine learning heuristics trained on over 1 billion media assets to correct inconsistencies and provide authoritative metadata.
Parameter Reference
Path | Data type | Optional | Description | Example |
---|---|---|---|---|
codec->id | uint | No | Numeric codec ID | 1983148141 |
codec->four_cc | string (4 char) | Yes | Codec FourCC (four-character code used to uniquely identify media codec/formats) | “mp4v” - mpeg4 video |
codec->name | string | Yes | Human readable video codec name. Multiple four_cc can mean the same codec name. For example “h264” and “avc1”, “mpg4” and “mp4v”, etc. | ”mpeg4” |
size->width | int | Yes | Video frame width | 1920 |
size->height | int | Yes | Video frame height | 1080 |
size->display_aspect_ratio | float | Yes | Display aspect ratio. Rounded to 3 decimal points. | 1.778 (for 16:9) |
size->sample_aspect_ratio | float | Yes | Sample aspect ratio. Rounded to 3 decimal points. | 1.333 (for 4:3) |
frame_rate->numerator | uint | No | Frame rate numerator | 30000 |
frame_rate->denominator | uint | No | Frame rate denominator. Cannot be 0 | 1001 |
pixel_format->color_model | string | Yes | Frame color model | ”RGBA” |
pixel_format->components_order | string | Yes | Component order for the color model. | ”GBRA” |
pixel_format->chroma_subsampling | string | Yes | Chroma subsampling | ”4:2:2” |
pixel_format->bit_depth | uint | Yes | Bit depth | 8 |
pixel_format->has_alpha | bool | Yes | Is alpha channel exists | false |
start_time | float | Yes | Start time aka delay. Rounded to 3 decimal points. | 0.033 |
duration | float | Yes | Video duration. Rounded to 3 decimal points. | 12.079 |
frames_count | uint | Yes | Video frames count | 362 |
bitrate->average | uint | No | Average bitrate. Typically - Steam size / duration | 301878 |
bitrate->nominal | uint | Yes | Nominal bitrate. Aka target bitrate. | 300000 |
bitrate->minimum | uint | Yes | Minimal stream bitrate | 298646 |
bitrate->maximum | uint | Yes | Maximum (peak) stream bitrate | 305943 |
color->primaries | string | Yes | Chromaticity coordinates of the source primaries | ”bt709” |
color->transfer | string | Yes | Opto-electronic transfer characteristic of the source picture | ”bt2020-12” |
color->matrix | string | Yes | Matrix coefficients used in deriving luma and chroma signals from the green, blue, and red primaries | ”ycgco” |
color->range | string | Yes | Color range (for YUV color space) | “tv” |
Parameter Categories
Codec Information
Essential codec identification and format details:
Video Dimensions
Frame size and aspect ratio information:
Temporal Information
Frame rate and timing details:
Pixel Format
Color and pixel structure information:
Bitrate Analysis
Comprehensive bitrate statistics:
Color Space
Advanced color space and HDR information:
Example Response
Data Types
Numeric Precision
- uint: Unsigned integer (always ≥ 0)
- int: Signed integer
- float: Floating point, rounded to 3 decimal places where specified
- bool: Boolean true/false value
String Values
- four_cc: Exactly 4 characters (may contain nulls)
- codec names: Standardized codec identifiers
- color values: Standardized color space names
Common Use Cases
Format Validation
Verify codec compatibility and format specifications
Quality Analysis
Analyze encoding quality and bitrate efficiency
Compatibility Check
Check device/platform compatibility
Transcoding Decisions
Determine if transcoding is needed
Machine Learning Enhancements
The Probe Report applies ML-trained heuristics to:
- Correct inconsistencies between FFprobe and MediaInfo outputs
- Validate suspicious values using learned patterns
- Fill missing data using statistical models
- Normalize formats across different tool outputs
- Detect encoding artifacts and quality issues
These enhancements are based on analysis of over 1 billion media assets and continuously improved through ongoing training.