USING AI TO RECREATE A FAKE BEER AD
REAL PRINT ADS BY BENEDIKTINER (& THEIR AGENCY)
FAKE AD with MIDJOURNEY AI + PHOTOSHOP BY ME
A CASE STUDY – »BENEDIKTINER WEISSBRÄU« AD
MADE WITH MIDJOURNEY AI + PHOTOSHOP
Concept & Creative Direction Marc Perino
AI Prompting & Retouching Marc Perino
Tools ChatGPT | Midjourney AI | Topaz Photo AI | Adobe Photoshop AI
Motif »Benediktiner Weissbräu«
-
When I was driving my father to a doctor the other day I passed a huge billboard with a beer advertisement for »Benediktiner Weissbräu« – a moderately big German brand which belongs partly to the larger »Bitburger Braugruppe«. Since I had no time to look at the billboard closely enough while driving I later came back and took a photograph of the billboard to study it later on.
When I got home I slowly formed in my head the idea – maybe even for a series? – to (re)produce existing print ads in Midjourney AI (Artificial Intelligence) and Photoshop to see what is possible today. I consider this a case study for myself and others in the industry who might be interested in this topic since AI is slowly creeping into every aspect of our professional and personal lives.
-
Normally when you even bother to look at a print ad for a product you probably don't think much about the creation of it at all. How and where or by whom it was produced? Or if you are even looking at a realistic image in the first place?!
I guess most people assume that advertising is »all a lie« anyway. We all know the images of a giant thickly topped juicy burger – or insert food of your choice here – next to a cold refreshing bubbling colorfully saturated drink. And then you walk into the next restaurant and what you will get on your food tray is a sad, small and underwhelming variant of what you have been promised in the advertisement.
But I don't want to bash any clients or agencies for »bending the rules« a little bit as long as the product that you receive works, tastes and/or looks at least similar in reality as promised. I worked for a long time in advertising – and still do – to know how the »sausage is made« – sometimes literally. Essentially we all know that advertisers are always go to great lengths to let their client's product look in the best way possible by special lighting, highly selective choices and processing of ingredients, materials and/or skilled retouching of the final images in the end.
But that brings me to the point I am trying to make. Retouching images to enhance images or create new photographic worlds has been in advertising for decades even before the invention of Photoshop. And that is completely OK with me since I do retouching and compositing myself and have improved products and humans in Photoshop for years to make them look a little more perfect than in reality. I won't go into the ethics and morals of retouching here because that might be a completely new blog post.
But in reality there has always been a »silent agreement« between customers and advertisers that what you see in an ad is probably not a 100% accurate depiction of a product or situation it is used in. The exact percentage varies from ad to ad. But the enhancement of reality is fueled by the idea that brands want to build an emotional connection to their customers rather than a realistic one. These emotions generated in an ad are supposed to leave a positive imprint in the minds of the potential customers before buying the product. And I am OK with that too. Apple for example is a master in this field.
-
In today's print ads it is quite common to build whole worlds in Photoshop that don't exist in order to place the product in the most visually appealing environment. And these ads are quite often stitched together from many many images – sometimes hundreds.
In the past retouchers have mostly relied on stock photography because you were able to to use images that were physically and logistically prohibitive to acquire otherwise. This process has been established for years, even decades and is used up to this day.
In the two original »Benediktiner Weissbräu« print ads next to this article you will see that in both images that the mountains in the back are completely different, the lens flare of the sun is placed somewhere else, the clouds in the sky have different formations and of course the placement of the product + hand is different.
So this ad is clearly a composited image because it has to serve many needs. This is totally fine by me because it has a positive vibe, the beer looks tasty and the image fits the message perfectly. Like I wrote in the section above it is more about the emotion the image conveys rather than an accurate depiction of reality. Without knowing the details of how this ad was produced – and I guess by an agency in conjunction with a photographer for the product shot – I can deduct from the image that it is probably made up of several different elements (which are called »assets«) because in reality it would next to impossible to capture such a moment where every element fits together perfectly at the right time and place. At least in a predictable and logistically and monetarily sound manner. Additionally it would be impossible to use the image in different formats – think of widescreen/landscape, portrait or square formats on different social media platforms. The beer + hand would be »cemented« into the background image and could not be moved. When you composite an image from individual assets they can freely be moved and customized for different uses cases.
Also images of THE PRODUCT – in this case the glass of beer – always receive a special treatment. That is mostly that they are photographed in a controlled studio environment where every detail in lighting, composition, material and appearance is tuned carefully until the client is happy. This is not impossible but very costly to produce on location. The end result might look a little bit artificial but as we established before this illustrative character of an ad is contributing to the emotion that the image should transport. So far so good...
-
But, hey Photoshop – there is a new sheriff in town called »Midjourney« (I know it is not the only AI at the moment but the most popular one). Since »generative AI« – that is AI which literally generates something, in this case »images« – has come upon us humans in the past few months with the speed of light and is probably here to stay, my prediction for advertising photography or advertisements in general is that AI will be quickly replace parts of the photography process or at least will be incorporated heavily into it.
I think that every image asset that is »generic« will be produced by AI in the future. That might not be the case of the product itself at the moment – although e.g. in the automotive industry it is quite common that the cars you see in an ad are completely CGI (computer generated imagery) – but everything else you see: the generic forest, the generic sky, the generic hand, the generic whole human model will probably generated by AI. Maybe in the next few months you can even upload images or CAD (computer aided design) files of products into the AI and place it perfectly lit into the already fake environment. In any case the time of stock photography might come to an end as we know it.
-
Like I wrote in the intro I was thinking about the possibility to completely generate »real ads« in Midjourney AI and wanted to explore this idea in a real world scenario. In the fake ad I created with Midjourney (MJ) I tried to come as close as the system allowed me. Since at the moment MJ is a little bit unpredictable if you try to create a motif with many different elements completely within the system I decided to approach it in a more »classical way« by looking or creating the assets individually. For such an assignment I would normally look for stock photos or background plates or would be provided with those by the agency.
So first I created the landscape (the »background plate«) with MJ and tried to establish a light direction. It took me a while to generate the »correct« landscape and of course the monastery in the background has nothing to do with the monastery in the original ad. This is why I emphasized the word »generic« several times. In this particular case you still see the limitations of AI because the system does not know that the monastery is a particular one which deeply connects to the beer brand. Probably because the first Benediktiner monks brewed their beer in that monastery. Today I think it is produced somewhere else but as we established before it is all about the emotion of the product and not an actual depiction of reality.
Anyway…after I extended the sky of the background with Photoshop’s generative fill (Adobe’s AI) I created a generic male hand with a generic beer glass. Since MJ is an American company and its AI model was probably trained with more US based images it was not possible for me to generate a German wheat beer shaped glass. I guess they are not that common in the USA but I don't know for sure. The hand and the caustic reflections of the liquid which were generated together with the generic beer glass on the fingers were quite acceptable to me.
So for the hero shot of the glass I cheated and resorted to the homepage of the company that produced the beer and found a similar product shot of the beer glass. I »put« the glass into the hand generated by MJ and was good to go.
Next I composited everything together and tried to match the emotion and atmosphere of the original ad. Of course they are not identical but in a real world scenario there would be no »original« to compare to because the artist would create the image from scratch in the first place. The agency probably just would have provided him with a drawn storyboard first. BTW creating storyboards with AI could be another blog post in itself.
-
Like I mentioned in my other article (link) there are still limitations with MJ. The images are still quite low in resolution and I had to use another AI (Topaz Photo AI) to upscale them. Additionally sometimes it is necessary to retouch artefacts or image defects which are produced by the AI system. In this case there was no need for it.
But in the end the idea of this experiment was only to see how much you can push technology and how viable some of these workflows are for creatives in the future or right now in the present. Because it is clear for me that this technology is evolving in giant steps and some creative industries will be affected heavily by it. And I'd rather stay at the forefront and incorporate it in my current workflow than to stay behind.
