by By: Michaela Horvath
In order to gain inspiration and possible content for my summer internship, I created an artificial intelligence campaign for Low Tide Wine co. I have worked with their larger company for a few years now, and they recently created a more casual label for younger audiences. Still the same hand-crafted wines, but with a new label for more aesthetic-driven clients! To help the business gain an idea of how to market to a new audience, I created the Boozy Beach Days campaign. A fun-influencer based social media campaign that will contact their target audiences.
In this campaign there will be three influencers who get delivered three bottles of Low Tide wines of their choosing. These three bottles will need to be featured on their Instagram pages in either a reel, or carousel post. These posts can be of the wine on a table of food, the wine in glasses, and or them holding the bottle in some way preferably at the beach or with beach-like scenery. It can also be a food-focused video of things that pair well with each wine. This will help gain tagged photos, and break into new audiences that are regularly difficult to contact for small businesses. These images may look something like this:
The next part of the campaign will be a bundle sale. Because the initial drop of Low Tide there are only three different wines, Grenache Rosé, Alberiño, and Chenin Blanc; with the purchase of all three, you get a voucher for a free bottle of your choice. Because these wines are yet to be tasted, the audience will get the opportunity to choose their favorite and get an additional bottle on the house.
The mock-up images created by AI are very effective in imagining what the real-life photos would look like, however, they would not necessarily be usable for the real campaign. It is easy to see mistakes because of teeth and facial features that aren’t in the correct places, and don’t look quite right. It is very fun to test and imagine how these campaigns would look using AI. It only took a few different tweaks in verbage for the images to turn out better. At first, the concept of the women sitting at a table didn’t come across well. The second time, the women holding the glasses were more animal-like than human. But, a third time’s a charm. The third time, with hyper specific language the images came out much better.
Michaela, I enjoyed reading about your campaign and think you did a great job planning out all the fine details. The AI images are very interesting because they fit the aesthetic you’re going for but I always find it difficult to get realistic human features. I think you came up with a great name for the campaign and would love to see this come to life.
Hi Michaela, I enjoyed reading your passage. I thought you had lots of great ideas for how to implement this campaign. Your AI images also turned out very well! Much better than mine when I tried.
Hi Michaela,
It’s cool to see someone else choosing AI for the second blog post. The first photo in particular looks really convincing. You can kind of tell that the food is fake (because no food has that much contrast) but you really have to be looking for it to make that observation. I could also see it as being altered with photoshop. Thanks for the submission!
Hi Michaela,
I like how your incorporated skills gained through class assignments into your professional work. AI generated content is a great way to create mock-ups for clients without having to invest money into professional shoots. While I see the problems with the hands and teeth, I think the content you created looks very realistic and similar to content we might see today advertising this product. Great post!
Hello!
I loved your vision for this campaign and creative ideas. While AI tends to appear somewhat spooky to me most of the time, it really is a great way to storyboard ideas and make a tangible list to go off of when it comes to creating content. Nice work!
Michaela,
I enjoyed this campaign proposal a lot. You did a great job of generating images that match the vibe of the brand and campaign. I spent some time generating images and also found that hyper-specific commands worked much better than vague ones. The better we become at crafting the right command, the better the A.I. generator will perform.