OK, that was clickbaiting. The breakthrough didn't just happen today. It's a process. But in March 2024, another important hurdle in using AI to create marketing materials has been overcome - the problem with the lack of consistency in AI-generated images. You can read here why this is so important and what next steps can be expected. And at the end, there's a personal confession.
The problem so far: too little efficiency!
Let's think about the previous process when we were looking for images for a campaign with a series of required assets, for example with a person. Either we do a shoot. The result is dozens of images with a model that are consistent in themselves and we have a selection of motifs to generate further situations and new assets with afterwards. Or we can use an image database, such as Getty Images. There, too, we can find a complete series for a model from one or more shoots. And this is the only way we can generate consistent stories or find exactly the two images we need right now.
Previously, this was not possible with AI, or only with great difficulty. For example, if you wanted a picture with the same person but in a different context or a slightly different position, you had to realise that although this was possible with skill, a lot of patience and a bit of luck, it was ultimately inefficient!
And efficiency is also the key reason why AI has not yet been used across the board for this type of agency task. It's amazing what you can already generate with AI, but in many real-life use cases, the classic way is still the faster way!
The new status with the "Character Reference" feature from Midjourney
Midjourney then introduced the "cref " feature on the night of 12 March 2024. Although this is not yet perfect, it significantly reduces the problem described above. With Midjourney, we will be able to reach our goal much faster and therefore more efficiently in future than by laboriously searching through image databases.
And that's just one of the improvements that really catches the eye! Every day there are new tools or simply smaller or larger leaps towards better results, easier operation or extended areas of application.
Let's think of speech-to-text applications: Until a few months ago, these were simply too artificial and too poor even for organic social media posts. Today, as with ElevenLabs, for example, the quality is already sufficient in many cases.
And now, also in March, came the extension: you can now create small sounds by simply entering text - and it's really quicker than searching for them in a sound database. This is just one of many examples. And that's why I'm coming to the end:
Yes, it's happening NOW!
If this continues at this pace, we as a digital agency will see in the coming weeks and months how the use of generative AI for more and more tasks will make the leap from the experimental stage to daily use: first for static images and then for moving images. Initially for simpler always-on assets without a media budget (already part of everyday life, but sometimes not really efficient) and finally also for regular assets of greater complexity that are pushed with a media budget.
Between enthusiasm and FOMO: a personal confession
I've been working in a marketing agency for over 20 years now (even in the same one!), even though I always need something new. I sometimes think that in many other industries I could never have stayed this long. I can get bored easily. But here in digital marketing, the world has been and continues to be reinvented every few years.
And so I also threw myself into the topic of AI about three years ago. That's a good thing, because I'm always amazed and enthusiastic. At the same time, however, I have to admit that I currently have FOMO several times a week. I haven't experienced that in the last 20 years. Yesterday I tried a tool and failed with the task - one day later a new feature, an improvement or a completely new tool comes around the corner. And then it works! Or even worse: there was a better tool yesterday - I just didn't know it yet!
My suspicion (and my hope): It's not just me - but many people who deal with this topic. Because even if I did nothing else but try out new AI tools and features in different application situations all day long, it would still be impossible to keep track of everything at the moment. It's just too much.
We will probably have to live with this situation for the next few months (years?). At least we won't be bored ...
