AI image generators and DAM: how to use them professionally in marketing

The revolution in image generation through AI

Artificial intelligence is developing rapidly and producing innovations that are massively changing our day-to-day work - especially in marketing. AI image generators enable you to create visual content quickly, efficiently and in high quality. Whether for social media, e-commerce or traditional advertising: tools such as DALL-E, Adobe Firefly, Photoshop, Google's Gemini (Imagen), MidJourney and Stable Diffusion and, more recently, Janus offer a wealth of possibilities.

But beware: the actual relevance and application of these tools in practice is often different from what is communicated in social media. The management of AI-generated material in Digital Asset Management (DAM) systems in particular requires a well thought-out strategy.

When dealing with AI-generated images, you should not only focus on the creative possibilities, but also on the scaling and organization of the assets created. But before we go into detail, let's start with one of the most important tools in the corporate context: Photoshop.

Photoshop: Increased efficiency through AI retouching

Even before Adobe integrated AI functions into Photoshop, the tool was the first choice for retouching and image editing. In the past, it meant removing distracting elements, making color adjustments and refining details with a lot of patience and precision. This could take time - and was not exactly scalable.

Today, generative AI takes over many of these tasks. All you have to do is select the area to be edited and enter a short description - and Photoshop will provide three suggestions for optimization. If none of them fit, you can generate further variants. This saves time and ensures consistent, high-quality results.

It gets particularly exciting when it comes to adapting images for different formats. Do you need a motif for a hero image, but also for a skyscraper banner? No problem: Photoshop can intelligently add missing image areas - with astonishing precision.

But beware: as the number of generated assets increases, so does the organizational effort. This is where DAM comes into play: If you don't want to get lost in a chaos of variants and versions at some point, you should define a clear naming and sorting strategy right from the start.

DALL-E & Imagen: AI image generators for social media

When it comes to generative AI, ChatGPT is the best-known tool - and because DALL-E is directly integrated into ChatGPT, it is one of the most widely used image generators. Most of the AI-generated images on LinkedIn come from DALL-E, because it is very easy to use: enter the prompt, get the image.

The strength of DALL-E (and Google's Imagen) lies in its low threshold. Even without a great deal of technical know-how, you can generate attractive images with just a few words. And if the result is not quite right, certain elements in the image can be edited afterwards.

But there are limitations:

• Text in the image? Difficult! Letters often look strange because the system does not really generate characters, but only imitates similarities.
• Little control over details. The options for intervention are limited - ideal for quick social media posts, but less so for sophisticated brand designs.
• Hardly any individualization. The generated images often have a certain style that does not always suit a brand.

For small and medium-sized companies, DALL-E is a great option for creating content for social media at low cost. However, if you work in a company with a larger marketing budget and value brand-specific designs, you should take a closer look at Adobe Firefly.

By the way: Canva has also integrated its own generative AI, which can be combined directly with the tool's social media templates. If you use Canva anyway, it's worth taking a look at this function.

Adobe Firefly: Customized AI images for professionals

While DALL-E & Co. are aimed more at the general public, Adobe Firefly is aimed at professional users. The tool is included in Adobe Creative Cloud and offers impressive control over the generation process (without Adobe CC, you get 25 credits/month, which is too little for an in-depth test).

What's special about Firefly:

• Using your own images as a base: Instead of just generating completely new motifs, you can specifically adapt existing material.
• Create CI-compliant content: Companies can integrate their own style elements and thus generate images that match the brand identity.
• Better results than entry-level tools: Firefly offers finer details and more realistic renderings than many other AI generators.

In short, if you need high-quality, customized marketing visuals, Firefly is one of the best options. If you also know a bit about coding, you can also control and automate the tool via Discord - originally a gamer messenger.

Midjourney: The favorite of creative people

Midjourney is the AI tool that is generating the most enthusiasm among designers and digital artists - and for good reason.

• Diverse styles: While DALL-E often looks “artificial”, Midjourney creates images with an impressive aesthetic.
• High level of control: You can upload existing images, use them as a reference and develop them further in a targeted manner.
• Efficient workflows: Thanks to various configuration options and community-driven development, Midjourney is incredibly powerful.

The tool is subject to a fee - with prices starting at 10 USD per month, it's basically low-threshold. But if you're serious, you should invest in a Pro or Mega plan, as this is the only way to keep the generated images private. However, there is a hurdle if you want to automate: With Midjourney, this works via Discord, which can be unfamiliar at first for inexperienced users.

For companies with high visualization requirements, Midjourney is also a real alternative to classic stock photos - especially if a large number of images in the same style are required.

Another advantage: automation via Discord makes it possible to transfer generated images directly into a DAM. This allows you to use Midjourney not only creatively, but also in a scalable way.

Stable diffusion: For real tinkerers

If you want even more control over the generation process, you will end up with Stable Diffusion at some point. The big advantage? It's open source. That means:

• Full control over the results (at least with enough technical know-how).
• Train your own AI models to get exactly the style you need.
• Free to use if you have the necessary hardware.

However: Stable Diffusion is not an entry-level tool. To use it efficiently, you need technical understanding and a fair amount of computing power. But if you familiarize yourself with the topic, you can generate extremely individual and brand-specific content - and with a flexibility that proprietary tools often do not offer.

The open source approach also has great advantages. On platforms such as Hugging Face, a community for AI development, you can test different models, analyze how they work and compare their performance.

And that's exactly where we took a closer look at Janus - the image generator from the Chinese AI start-up DeepSeek, which has recently made headlines. Our focus was on testing how well the model recognizes objects in images. Without going into too much detail: It works. However, we found that there are errors that are probably due to translation problems.

In terms of image generation, we were not yet completely convinced by the results - we recommend continuing to monitor developments here. The market for AI image generators is currently extremely dynamic and it is quite possible that there will be significant improvements in the coming months.

However, it is interesting to note that DeepSeek has already achieved impressive benchmarks in the field of text generation. So perhaps it is worth keeping an eye on future developments - especially for companies that want to invest in generative AI in the long term.

Efficient digital asset management (DAM) for AI-generated images

The use of AI image generators brings many advantages - but also challenges, especially when it comes to organizing and managing the generated assets. Without a well-thought-out DAM strategy, chaos can quickly ensue. To ensure that your image generation works smoothly with the DAM, you should consider the following aspects:

Structured, standardized file naming & metadata

AI image generators often produce numerous variants of a motif. To maintain an overview, you should develop a clear naming structure, e.g. image type, project name and image number. Also add relevant metadata, such as the AI tool used.

Use automated workflows

Some tools such as Adobe Firefly, MidJourney or Stable Diffusion offer the option of transferring generated images directly to digital asset management via interfaces. With the TESSA DAM, you can support automation, set up workflows, automatically keyword the images or sort them into the right category.

Separate AI assets from originals

As AI-generated images often only serve as a starting point and are further processed, it is advisable to store them in a separate area of digital asset management. Sometimes it is the other way around - the AI uses originals as reference data or a starting point. We recommend marking these accordingly. This makes it easier for you to distinguish between source material, originals, edited versions and final assets.

Observe rights & licenses

There are different usage rights depending on the AI tool. Some generated images are free to use, others are subject to restrictions. Document the origin and license conditions directly in the DAM. The TESSA DAM provides a corresponding infrastructure here.

Plan scaling strategically

As AI drastically accelerates image production, the volume of assets also grows. Make sure that your DAM has enough storage space, powerful search functions and well thought-out categorization to remain efficient in the long term. This is easy with the TESSA DAM. You can purchase additional storage space at any time.

With these measures, you ensure that AI-generated images can be used not only creatively, but also in a structured and efficient way.