Key concepts and terms
Formats
Formats are the types of image and video file types that Image and Video Manager can store, display and manage, such as JPEG, PNG, MP4, and AVIF. For the full list of supported formats, see Supported image and video formats.
Transformations
Transformations are changes or adjustments applied to the images or videos to alter their appearance, format, size, based on specific needs and optimization requirements. These changes are applied automatically based on the predefined policies. These transformations are key to generating optimized derivative images or videos that fit the specific requirements of different devices, network conditions, or use cases. See Add image transformations and conditions to learn how you can change the appearance of your original images.
Pristine vs. derivative assets
Pristine image or video
A pristine resource is the highest-quality version of a media file as it exists on your origin server, free from any additional compression, alterations, or quality degradation by intermediary systems. This version represents the file exactly as it is served from your origin, unmodified by Image and Video Manager, though it may differ from the file's original capture or creation if you modified it before uploading.
Derivative image or video
A derivative resource is a version of the original (pristine) media that has been modified or processed to meet specific needs, such as optimization for performance, size, or format. The derivatives are created by applying transformations to the pristine file.
Real-time vs. Offline derivatives
Real-time derivatives
A real-time derivative is an immediate, fast processing operation that is applied to an image or video after receiving the first request. This derivative is designed for speed, as it prioritizes speed over image fidelity, and typically generates a low-quality, short-lived derivative. Due to its speed-oriented nature, the real-time process does not apply perceptual quality enhancements and only supports basic image formats (that is, JPEG, PNG and GIF) by default, although these can be overridden by specific policy settings. The result of a real-time derivative is a derivative of a single format optimized for preview or quick access to the asset, while more robust transformations are performed asynchronously.
Offline derivatives
An offline derivative is a high-quality, comprehensive processing operation that is initiated after a successful real-time transformation. It is performed asynchronously and is not intended for immediate delivery, so it can focus primarily on fidelity, perceptual quality and support for a wider range of formats (that is, AVIF and WebP) or resolutions. Offline derivative creates derivatives that are intended for long-term access and are therefore suitable for final or production-ready versions of the asset. These derivatives often require more processing time, but provide higher quality results suitable for consumption or storage by the end user.
Real-time and Offline comparison
Aspect | Real-time | Offline |
---|---|---|
Speed | Prioritizes speed, fast but temporary response | Slower but more thorough processing |
Image quality | Lower fidelity, trades quality for speed | Focuses on maintaining high perceptual quality |
Transformations | Limited and simplified | More advanced, customized based on policies |
Supported formats | Default to a single generic format, can be overridden | Supports multiple formats based on device or browser |
Usage | Immediate response for the first asset request | Long-term optimized storage and delivery |
Perceptual quality | Not supported | Supported, ensuring minimal quality loss |
Lifespan | Short-lived | Persistent, used for subsequent requests |
In summary, real-time processing offers a quick, temporary solution by delivering a lower-quality image to fulfill an immediate request, while offline processing takes more time to generate higher-quality, optimized images that are stored and served for future requests.
Perceptual Quality
Perceptual quality refers to how visually appealing or acceptable an image or video appears to the human eye, and is not strictly tied to technical metrics such as file size or resolution. It’s about maintaining a balance between visual clarity and file optimization so that media looks "good enough" without taking up unnecessary space or bandwidth.
In Image and Video Manager, perceptual quality is managed using algorithms that adaptively adjust quality, resolution and format to preserve the appearance of your media while optimizing it for faster load times and lower bandwidth:
- Low Perceptual Quality. Allows for more aggressive compression, tolerating some visible artifacts (slight blurring or loss of fine detail) in exchange for smaller file sizes. This setting is useful when file size reduction takes precedence over perfect visual reproduction, for example, for thumbnails or preview images.
- Medium perceptual quality. Ensures a balance between visual fidelity and file size reduction. This setting aims to preserve detail as much as possible while achieving a moderate level of optimization. This is a good default setting for most use cases where clarity is important but extreme precision is not required.
- High perceptual quality. Prioritizes visual quality over file size savings. The algorithm minimizes compression artifacts and rejects optimizations that could lead to visible degradation. This setting is ideal when the clarity of the image is crucial, for example, for high-quality display images. In the case of very compressed or low-quality input, Image and Video Manager can avoid additional compression to prevent further loss of quality.
How Perceptual Quality Works
The perceptual quality algorithm in Image and Video Manager evaluates recognizable elements in an image (such as edges and details) to detect compression artifacts caused by optimization. If an image is already heavily compressed or has no recognizable features (such as color gradients or low-texture areas), the algorithm may determine that further compression would degrade the quality beyond your chosen perception setting. In such cases, Image and Video Manager may retain the original image or reject excessively degraded derivatives.
If you have problems with the perceptual quality settings, we often recommend a static quality setting of 85, which corresponds to a medium to high perceptual quality.
Policy management
Policy
Policy is a customizable container for the rules and configurations that determines how Image and Video Manager processes, optimizes and delivers images and videos. It defines the specific transformations and actions, such as resolution adjustments, format conversions, and performance optimizations based on device types, or user preferences. With a policy you can tailor how your media is handled in a flexible and scalable manner, ensuring that the Image and Video Manager aligns with your business and your users' objectives.
Policy set
Policy set is a collection of policies, and hence, a set of predefined rules and configurations that govern how media is processed, optimized, and delivered. This set defines how various transformations are applied to images or videos in different situations, ensuring they meet performance, quality, and user experience objectives.
Property
To learn more, see Property Manager documentation.
Rule
To learn more, see Property Manager documentation.
Updated 28 days ago