Optimizing image quality can be a time consuming process as many factors contribute to rendering acceptable results. The outcome is partly subjective because individuals perceive image quality differently. Structured experimentation is key.
Adobe Dynamic Media Classic includes more than 100 image-serving commands for tuning and optimizing images and rendering results. The following guidelines can help you streamline the process and achieve good results quickly using some essential commands and best practices.
See also Smart imaging.
Try out and discover the benefits of Dynamic Media image modifiers and Smart Imaging, using Dynamic Media Snapshot.
Snapshot is a visual demonstration tool, designed to illustrate the power of Dynamic Media for optimized and dynamic image delivery. Experiment with test images or Dynamic Media URLs, to visually observe the output of various Dynamic Media image modifiers, and Smart Imaging optimizations for the following:
To learn how easy it is to use Snapshot, play the Snapshot training video (3 minutes and 17 seconds).
As a best practice for image format, start with the most common setting
Dynamically reducing image size is one of the most common tasks that Dynamic Media Image Serving performs. It involves specifying the size and, optionally, which downsampling mode is used to downscale the image.
&hei=<value>. These parameters automatically set the image width in accordance to the aspect ratio.
&resMode=<value>controls the algorithm used for downsampling. Start with
&resMode=sharp2. This value provides the best image quality. While using the downsampling value
=bilinis faster, it often results in the aliasing of artifacts.
As a best practice for image sizing, use
Image sharpening is the most complex aspect of controlling images on your website, and where many mistakes are made. Take the time to learn more about how sharpening and unsharp masking works in Adobe Dynamic Media Classic by referring to the following helpful resources:
Best practices white paper Sharpen images in Adobe Dynamic Media Classic and on Image Server.
With Adobe Dynamic Media Classic, you can sharpen images on ingestion, on delivery, or both. Usually, however, you sharpen images using only one method or the other, but not both. Sharpening images on delivery, on a URL, typically gives you the best results.
There are two image-sharpening methods that you can use:
Simple sharpening (
&op_sharpen) - Similar to the sharpen filter used in Photoshop, simple sharpening applies basic sharpening to the final view of the image following dynamic resizing. However, this method is not user-configurable. The best practice is to not use
&op_sharpen unless required.
Unsharp masking (
&op_USM) - Unsharp masking is an industry standard filter for sharpening. The best practice is to sharpen images with unsharp masking following the guidelines below. Unsharp masking lets you control the following three parameters:
amount (0-5, strength of the effect.)
radius (0-250, width of the “sharpening lines” drawn around the sharpened object, as measured in pixels.)
Keep in mind that the parameters
amount work against each other. Reducing
radius can be compensated by increasing
Radius allows finer control as a lower value sharpens only the edge pixels, whereas a higher value sharpens a wider band of pixels.
threshold (0-255, sensitivity of effect.)
This parameter determines how different the sharpened pixels must be from the surrounding area before they are considered edge pixels and the filter sharpens them. Threshold helps to avoid over-sharpening areas with similar colors, such as skin tones. For example, a threshold value of 12 ignores slight variations in skin tone brightness to avoid adding “noise”, while still adding edge contrast to high contrast areas, such as where eyelashes meet skin.
For more information about how you set these three parameters, including best practices to use with the filter, see Sharpen images in Adobe Dynamic Media Classic and on Image Server.
Adobe Dynamic Media Classic also lets you control a fourth parameter: monochrome (
0,1). This parameter determines if unsharp masking is applied to each color component separately using the value
0 or to the image brightness/intensity using the value
As a best practice, start with the unsharp mask radius parameter. Radius settings that you can start with are the following:
Gradually increase the amount from 1.75 to 4. If sharpening is still not the way you want, increase the radius by a decimal point and run the amount again from 1.75 to 4. Repeat as necessary.
Leave the monochrome parameter setting at 0.
This parameter controls JPG encoding quality. A higher value means a higher-quality image but a large file size; alternatively, a lower value means a lower quality image but a smaller file size. The range for this parameter is 0-100.
To optimize for quality, do not set the parameter value to 100. The difference between a setting of 90 or 95 and 100 is almost imperceptible, yet 100 unnecessarily increases the size of the image file. Therefore, to optimize for quality but avoid image files becoming too large, set the
qlt= value to 90 or 95.
To optimize for a small image file size but keep image quality at an acceptable level, set the
qlt= value to 80. Values below 70 to 75 results in significant image quality degradation.
As a best practice, to stay in the middle, set the
qlt= value to 85 to stay in the middle.
Using the chroma flag in
qlt=parameter has a second setting that lets you turn on RGB chromaticity downsampling using the normal value
,0(default), or turn it off using the value
,1). This setting usually results in better image quality, especially for synthetic images with lots of sharp edges and contrast.
As a best practice for JPG compression use
jpegSize is useful if you want to guarantee that an image does not exceed a certain size for delivery to devices that have limited memory.
jpegSize=<size_in_kilobytes>). It defines the maximum allowed size for image delivery.
&jpegSize=interacts with the JPG compression parameter
&qlt=. If the JPG response with the specified JPG compression parameter (
&qlt=) does not exceed the
jpegSizevalue, the image is returned with
&qlt=as defined. Otherwise,
&qlt=is gradually decreased until the image fits in the maximum allowed size, or until the system determines it cannot fit and returns an error.
As a best practice, set
&jpegSize= and add the parameter
&qlt= if you are delivering JPG images to devices with limited memory.
As a best practice, to achieve a high image quality and small file size, start with the following combination of parameters:
This combination of settings products excellent results under most circumstances.
If the image requires further optimization, gradually fine-tune sharpening (unsharp masking) parameters by starting with a radius set to 0.2 or 0.3. Then, gradually increase the amount from 1.75 to a maximum of 4 (equivalent to 400% in Photoshop). Check to see that the desired result is achieved.
If sharpening results are still not satisfactory, increase the radius in decimal increments. For every decimal increment, restart the amount at 1.75 and gradually increase it to 4. Repeat this process until you achieve the desired result. While the values above are an approach that creative studios have validated, remember that you can start with other values and follow other strategies. Whether the results are satisfactory to you or not is a subjective matter, therefore structured experimentation is key.
As you experiment, the following general suggestions are helpful to optimize your workflow:
&product_high$. The custom preset name in a URL path calls these presets. Such functionality helps you manage commands and quality settings for different usage patterns of images on your website and shortens the overall length of URLs.