We recommend using this Ad Maker to try out ideas quickly and to build visual references for your creative agency or the marketing team you work with. The scoring of creative effectiveness is there to help you select the best generations and test every idea before spending a minute on design. Most generations are not ready to be final ads. However, you will notice that some of the image could be published as ads. In those cases, you can add your logo and your product in the editor, as well as make final alterations. If you want to generate images directly with your product inside of them, contact us so we develop a custom model for you.

The heatmap represents the elements that will attract most attention from viewers in the first 2 seconds of exposure. It works with a computer vision model trained by Memorable researchers. The model has learned with data from millions of human reactions to ads, collected through online cognitive experiments.

We use a set of deep-learning models with two goals. The first one is to generate and alter images based on a text prompt. The second group of AI models perform the most important task: predicting the marketing impact of the generation, so that users can assess the potential of each idea.

Yes, but the quality of the generation is lower than for other elements. We recommend using the Ad Maker for assets that are not centered on human figures or don’t display humans at all.

Our paid plans include high-definition exports with 1024x1024 sizes. You just need to click on the Upscaled Export button. Our free plan downloads images with a 512x512 resolution and a Memorable logo.

The Ad Maker generates images to optimize for marketing impact on Facebook and Instagram. However, it can be used for plenty of other static ads. Try it out and tell us about your use cases!

Our paid plans give you full rights to the publication of the generated images. This is not included in the free plan.

Since the generation follows a randomization process, getting to the exact same image is extremely unlikely. It doesn’t only require users to type the next prompt, but it also requires the model to randomly select the same root number. This has a likelihood of one in billions. To take this even further: as the generative model is retrained, it becomes virtually impossible to replicate old generations.

Paid plans include full commercial rights and enable users to download the images in high quality. The free plan gives you access to an image with the Memorable logo and does not provide publishing rights.

In our experience, it can take between 5 to 10 credits to generate and make multiple alterations and tests until you have a ready-to-publish ad. If you already master prompt generation, it’ll take you a bit less 😎

This metric helps you build a stronger brand. Ad recall measures the likelihood that an image is remembered after being seen once. The score shown is a percentile over all ads in the Memorable database. We collect millions of human responses in a cognitive experiment of aided recall. The experiment checks if viewers recognize having seen an element before or not. As those responses become massive, we start getting enough signals for our machine-learning models to identify patterns associated with higher or lower recall. This results in models that can predict with high accuracy the probability that a viewer will recognize a visual element after seeing it once. The model deployed in the ad maker is a simplified version of the models in our creative platform. Contact us to access the full version.

This metric helps you build a differenciated ad. Distinctiveness measures how different your image is with respect to ads in the selected industry. The score shown is a percentile over all ads in the Memorable database. For each industry, we take thousands of ads from our database to measure what an average asset from that industry looks like. With the help of mathematical distance metrics, we are able to measure how dissimilar the generated image is from the average asset for that category. The model deployed in the ad maker is a simplified version of the models in our creative effectiveness platform. Contact us to test your ads with the full version.

This metric maximizes the attention of social media users on your ad. It estimates the proportion of attention going to your ad **vs** the rest of the content displayed in the social platform. The score shown is a percentile over all ads in the Memorable database. After simulating the generated image in a social media context, we are able to use our attention models to predict how much attention the asset is getting when competing with other visual elements in the channel, like buttons, avatars, other images, etc. Our predictive models for attention are trained using massive data that samples human gaze for different images and videos.

This metric helps you drive likes and shares on social media. The score shown is a percentile over all ads in the Memorable database. This is a simplified version of our models to predict likes and shares on social media. It recognizes different creative elements driving engagement online, and uses them to predict the impact of your ad. The full version of this model is channel and brand-specific. Contact us to pretest your ads using it.