The Everything Project

Foreward

There’s a lot of uncertainty in creative industries at the moment. AI was shoved into all of our lives without anyone really asking for it, and has been making leaps and bounds in its ability to create stuff in recent months. Understandably a lot of people are scared or at least uneasy about the post AI future in creative industries.

As someone who approached the AI onslaught with what I can only describe as cautious optimism, I thought this project done recently for Leicester, Leicestershire & Rutland Road Safety Partnership (LLRRSP), was a great example of AI being used alongside the creative process, instead of replacing it. In fact there were a lot of very traditional, almost “old school” design techniques used during this project, alongside the comparatively cutting edge AI tech.

Intro

LLRRSP sought an impactful anti-drink driving campaign that would resonate with a wide range of age groups. After careful exploration, we found inspiration in the iconic 80s Roadside Volkswagen Advertising as well as the advertising style of Apple in the 80s and 90s. We chose this approach because it not only connects with older audiences who grew up during that era, but also taps into the current 90s renaissance in art and design. This choice allows us to be perceived as both nostalgic and cutting-edge, appealing to both the older generation and the younger, forward-thinking audience.

Theres also a certain something about the striking imagery in a white abyss, with these almost tongue in cheek, very short copy, designed to be read in a very short space of time, all these factors made these adverts seem like the perfect pool to draw inspiration from.

Challenges

The imagery is a huge component of these, so finding something that fit the theme and gave the right look and feel was a very important part of capturing the very specific feeling of these ads. After an initial search a free use image was found:

This image felt perfect, save for one pretty crucial factor, it was really low resolution. A few months ago this would have been a dead end, we’d have had to either re think the project or try to source another image, which is hard to do with small budget projects like these. However thanks to some fairly basic AI tools, making this usable was a quick, hassle free process.

Ai Process

The first step was upscaling the image to remove the graininess. This process was done using an open-source AI tool called replicate, full of different models for very specific use cases, one of which is image upscaling, there are multiple upscaling models on replicate, however, my go-to is called “nightmareai/real-esrgan”

The next step required the removal of the background, again, AI tools help out here, not that it wouldn’t have been doable without, but AI turns what was previously an hours-long task, into something that can be completed in a matter of seconds. For this we used photoshop’s inbuilt “select subject” tool

Older Techniques

Because of the printed nature of the original versions of these ads, it always looks like something is missing having an artificial, digital replication of that. We employed a couple of techniques for the final creative to give the ads this authentic old-school look.

Statics

To authentically give the creative a printed look, we utilised the tried and tested method of rescanning. Printing designs out using a laser printer, then re scanning at ultra high DPI.

Animated

For animated creative, we couldn’t rescan. The number of frames needed for a 15 second animation would have wasted over 300 sheets of paper just to get this one animation, so we had to re think how were approached it.

Step 1 6 pieces blank pieces of paper were lightly creased/crinkled up and then scanned in to use as a background, this was looped to extend to the full length of the animation

Step 2 A series of effects were added to the animated composition to emulate the look of ink bleeding on paper

Step 3 The frame rate was reduced from 30 to around 20, old stop frame animations that this sort of creative is emulating generally run at slower frame rates, so this helps emulate that look more closely.

Step 4 A small random amount of movement and rotation was added to each frame, although very subtle, this helps sell the idea that a person has manually scanned in each of these frames by hand, giving it the slight look of imperfection