Dashboard Guide

Information box –

1) Upload image –
Size and Format
Formats supported: png, jpg
Image file size: not larger than 5MB.
Use white background and NOT transparent.
Transparent pixels with no colour will be treated as black.

Recommended image dimensions: 1024x768, 1366x768, 1280x720, 1920x1080

Use your highest quality images. We recommend on using PNG files.
You can use JPG format with 100% quality using no compression.
Product and Package image
Prefer landscape or square images.
Add at least 10% of blank margins.

Website screenshot:
For an accurate screenshot, you should create the screenshot manually, using a browser add-on, and then upload the image using the menu Upload > Image File To create a web snapshot, you can use a browser add-on such as:
* Awesome Screenshot
* Fireshot
* Paparazzi for Mac OSX
Capture the visible part of the webpage.
Create a website screenshot which contains only the webpage itself. Do not include the surrounding browser UI elements.

2) Settings –
View Type
This parameter currently has no impact on the heatmap. By selecting a value you are helping Feng-GUI system to learn more about the content inside input images.

View Distance
Set the viewer distance from this image. Default is a computer screen.
Available options are: Desktop Screen, Mobile, Print, Indoor Signage, Package Design and Outdoor Billboard.

Add Area of Interest –
Either add Area of Interest or choose auto AOI.
With auto AOI, our algorithm will create the auto areas of interest in final output

3) Visual Features –
Visual Features are major contributors to the calculation of AOI score. Drawing attention requires the object to have a significant contrast from surrounding objects and background.

Intensity Features - The percentage level of intensity.
Intensity is another way of saying luminance contrast, brightness, black/white contrast. So, it’s more like the other versions of color contrast. Design change can be done by changing intensity i.e by adjusting “brightness” of areas/objects (dark/light).

Edges Features - The percentage of edges, texture and orientation.
Design change can be done by Edges are created by shapes, which can be created by objects and text. And edge occurs when a group of very similar tightly packed pixels are adjacent to a different group of similar, tightly packed pixels. They can be changed in many ways, such as making the shape larger, changing spatial relationships, changing colour schemes, removing blur or shadows, changing orientation (change angle of the object), adding or reducing edges, adding or removing face etc.

Red Green Contrast - The percentage of Red-Green difference.
Design change can be done by changing colour contrast Red/Green

Blue Yellow Contrast - The percentage of Blue-Yellow difference.
Design change can be done by changing colour contrast Blue/Yellow

Our vision system picks up both high- and low-levels of colour, so when reports indicate the presence of R/G or B/Y contrast, that means there is either a lot of colour, or a lack of colour. You can witness this by examining images within any software tool that lets you measure the R/G/B levels of specific areas. So, you can increase the impact of these elements by changing R/G/B levels to increase or decrease colour saturation, hue, etc.

A high contrast between an element and its surrounding will increase the element's visibility. Measure the colour contrast of an AOI element and alter it to reduce or induce its visibility. Identify the dominant colour of the element (foreground) and the dominant color of its surrounding (background).
Measure the colour contrast ratio of the two colors. Change the element or its background luminance to increase or decrease the colour contrast ratio.
Colour Contrast ratios can range from 1 to 21 (commonly written 1:1 to 21:1).
Minimum contrast for reading text is 4.5:1
The contrast of black foreground over a white background is 21:1

A note for you –
Keep in mind that attention-getting power is always relative everything within the analysed image/photo/design is evaluated by the model. Therefore, it’s the “mix” of the above visual elements and other such as size, intersection, text, skin and face detection in a specific image that dictate the attention getting potential. These elements are the “building blocks” of visual attention, but they work together to create the overall effect.

You can also increase the attention-getting power of your visual priorities by modifying non- priority areas/objects; especially those that results show have strong attention-getting power. We call these areas distractors. For example, let’s say you want the product variant name on a package to be more prominent, but the ounces/grams text is attracting significant attention. You can do 3 things:

Strengthen the product variant name visual elements
Weaken the ounces/grams text visual elements
or you can do both.

This is perhaps the most common question we receive, and because each image/scene is unique and the elements contribute in a “relative” manner, it’s difficult to identify “best practices” in the conscious viewing sense. There is no formula, but thankfully designers are naturally good at this. Once they understand the model’s output, they are best equipped to recommend what to do next.

The Visual Elements result, is intended to inform the designer (and team) to help them understand where the Visual Elements exist within a design/image, and to provide more detailed data that may help guide what to try next.