FIXING APPLE’S FLOP, GPT-4O REVIEW, IPHONE VIDEOGRAPHY TIPS
CaJu’s Fresh Picks 01
The following is a repost of our newsletter! To get articles like these delivered to your inbox as soon as they come out, consider signing up HERE or at the bottom of this page.
Housekeeping: Welcome
First and foremost, welcome to CaJu’s Fresh Picks! This semi-regular newsletter will try to cover the latest goings on in the Marketing, Production, and Analytics industry spaces. A pretty wide net, but then again that is our modus operandi here at CaJu Creative.
I think at a later date I’ll delve more into the personal experiences of launching one’s first business or launching one’s first business with one’s spouse in the first year of one’s marriage, but for this inaugural edition of Fresh Picks, at the risk of immediately souring the novelty of it, I’m going to be jumping right into a special Industry Update Edition! That’s to say the way this newsletter will usually function is, when we have something worth sharing, I’ll post it here first and foremost.
That might be tips and tricks, that might be our commentary on the latest industry news, or it might be sneak peeks or first looks at content we’ve made in-house. For an as-of-yet undetermined amount of time, that content will only be available through the newsletter, then after some time, it will be posted permanently on our blog.
Once a month, I’ll do a Fresh Picks Industry Update Edition that serves as a proverbial foot in the proverbial pool that is the trifecta of industry we find ourselves in. So with the pomp and circumstance out of the way, I want to humbly thank you for being some of our first subscribers and cut the ribbon on our first newsletter!
*snip*
Marketing: Fixing The Worst Apple Ad I’ve Ever Seen
One of the classes from college that has served me best in my career has been the Advertising 101 elective I took in my final year. The first day of that class was essentially an exposition of old Apple advertisements. From their “1984” ad to the first iPhone keynote in 2007, Apple’s advertising ideology has always been groundbreaking and mind-bending in subtle but effective ways. Or at least that’s how I used to feel.
In recent years, I feel that they’ve lost a bit of their sheen. I believe this is in no small part due to the fact that more and more every year they become the only game in town for a growing segment of the population who have been trained to associate their polish with a vision of the future. However, starting with their “Shot on iPhone” campaign from years back, I started to suspect that they were starting to sell the steak rather than the sizzle, so to speak.
That suspicion was confirmed with the recent release of the Vision Pro, their fantastically expensive and isolating VR headset, which has all but since crashed and burned. A near-completely banal device whose main purpose is to further isolate us from one another? At this cultural moment? Really? There must be something I’m missing that Apple sees. They couldn’t be this far gone right?
Anyway, they’re that far gone.
Fast forward to the recent “Crushed” ad. I, like most people online, had a visceral reaction to the ad. A giant metal hydraulic press crushing symbols of human artistic expression like pianos, paint, guitars, arcade cabinets, etc. It was immediately obvious what Apple wanted to say. “Look at all this stuff! We’ve been able to push all of this stuff into the thinnest device yet.” But that’s not how it comes across. I mean just watch it.
Crush! by Apple: https://www.youtube.com/watch?v=ntjkwIXWtrc
What this conveys to me is ignorance boarding on arrogance and waste. Ignorance in that I could see what they were going for, but I simply couldn’t believe that they wouldn’t have anticipated backlash. Arrogance in the belittlement of not only the manufacturing labor and care that went into all of those objects but also the artistic labor they’re associated with. Waste in the destruction of all of these objects that mean so much to those whom this commercial is supposedly aimed at.
The fact that Apple had to issue an apology is like finding out the pilot of your airplane thought he could get away with dozing off for a bit. It’s nice to hear you’re going to try to do better, but I’m more concerned about why you thought this was a good idea in the first place.
Thus, we get to the true root of the issue: artists must not be the audience of this advertisement. Anyone who owns a drumset or an upright grand piano won’t be throwing them out for the latest iPad anytime soon. However, someone resentful of their inability to play an instrument might be first in line on launch day. An artist who already paints on canvas might get an iPad for practice, but someone who learns to paint on an iPad might never see the need for a canvas.
If you’ve been online and witnessed some of the conversations around generative AI “art” and those who proudly call themselves “AI Artists” or “Prompt Engineers,” then you already know Apple’s audience for this iPad. Unfortunately, it seems Apple is either ignorant to the shift in how people are perceiving them, or they’re perfectly fine no longer standing for the empowerment of creatives, likely because there’s more money to be made in being the boot that stamps them out.
Good advertising must see the products of itself as cultural touchstones within the greater context of the cultural moment. Just like Apple’s “1984” ad addressed anxieties about the moment head-on. Right now, our zeitgeist is concerned with the meaning and role of creativity in human life as the prevalence of Generative AI threatens to doom humans to toil or unemployment while AI turns all our creative endeavors into derivative slop.
That is the feeling that should’ve been addressed in the ad. Instead, it played into these anxieties. Apple became “Big Brother” up on the TV screen when it needed to be the hammer.
So what would I have done differently? With Apple’s budget, this ad could’ve been literally anything, but the first conversation I would’ve liked to have with the stakeholders is “Who do you want to buy this thing and why do you think they should?” Because if you’re targeting the artists who have traditionally made up your most dedicated fanbase, maybe it’s a bad idea to crush the symbols of their livelihoods with 2 tons of cold steel.
If I were really trying to sell a computational device to artists, I wouldn’t be headlining its thinness. My 2018 iPad Pro is plenty thin and I have no ambitions to replace it anytime soon. If I wanted to advertise iPads to artists, I would want to show how they make you better at using the creative objects you love. You + iPad = More of what you love, not less. You could write around the “less is more” cliche, but what might be better is showing the artistic process of trying and failing over and over again. Show off the fact that you can practice on the iPad without having to buy a new canvas every time. Partner sleeping? No problem! Write your music on the iPad, then play it in the morning!
The iPad isn’t the hydraulic press crushing the things that you love. iPad is the wedge stopping the crushing weight of self-doubt through ease of practice.
You can have that one for free, Apple marketing team.
With the iPad, you’re “You-Plus.” Unfortunately, the Crush ad seems to convey more “You-Minus” type vibes.
Samsung had an incredibly fast and clever response to Apple’s ad called UnCrush. Check it out below.
Uncrush by Samsung: https://www.youtube.com/watch?v=cMrVz0d_MF0
Unfortunately, it also misses the mark by tainting its ad’s closing moments with pro-AI sentiment; a subject also accused of dehumanizing art. I think that’s indicative of a bigger conversation about the limits of these new “AI” systems. More on that below…
Analytics: “Hands-On” With GPT-4o
Here’s my review in its entirety: GPT-4o is better than GPT-3.5 and faster than GPT-4. While it may not perform some tasks as well as GPT-4, this is likely due to OpenAI’s efforts to optimize and scale the model, possibly for their rumored partnership with Apple later this year.
If you use AI for tasks like sending emails, GPT-4o can help you do it faster. For data analysis, it offers similar performance but with increased speed.
That’s it. That’s the review.
Now, if you, like me, are skeptical of many of the promises of Gen AI, then read on, fellow heretics.
The main limitation of the current generation of AI, including GPT-4o, is novelty. By novelty, I mean creating that which is truly new as the human mind does; the intaking of stimuli to create meaning-enriched outputs greater than the sum of their parts.
Generative AI excels at repeating and iterating on existing patterns but struggles with creating truly original content. To understand these limitations, it’s essential to grasp the underlying technology. Large Language Models (LLMs) like GPTs are essentially solving complex mathematical problems derived based on given inputs. The promise of Artificial General Intelligence (AGI) becomes ridiculous once you realize that these models are just advanced statistical tools and that no actual thinking or creativity is being done.
One of my favorite examples describes GPT’s text generation as akin to “creating blurry JPEGs.” Just as AI art generators ingest millions of images and tokenize them—breaking down pixel patterns and assigning numeric values to certain words, word sequences, partial words, individual characters, or mixed character sequences—GPTs do the same with text. You input a prompt, and the model generates and solves a statistical equation to produce the most likely sequence of tokenized character sequences.
Generative AI cannot give you what it has not seen before. If it’s giving you something that seems novel, it’s because you’ve never seen the source data before. If you ask ChatGPT to make up a word, it may do so, but only based on existing labor already done for free and scraped from the internet. Generative AI is entirely dependent on the outputs of the very same artists that it threatens to put out of work. As more and more data that these models scrape from online becomes AI-generated itself, pre-generative AI internet text will become as valuable as pre-nuclear age steel.
Untainted human creativity is what fuels these technologies. Hence OpenAI’s recent announcement of a partnership with Reddit, a source for organic human-written training data (provided it protects against the flood of AI-generated text).
That said, I think it’s important for creatives today to understand that these technologies aren’t going away anytime soon and that when used with understanding an of their limitations, they can increase the creative professional’s productivity. My rule of thumb is “never rely on ChatGPT for a final draft,” but it is pretty good if you’re looking to throw some stuff at the wall and see what sticks. Just keep in mind that if someone else types in more or less the same prompt, they’re going to get more or less the same answer because the models are, more or less, pulling from the same dataset.
To give the devil his due, these tools are amazing for repetitive or tedious tasks. Sorry in advance for some data-rhetoric, but in Data Science, a common method for finding patterns in large series of text files is called “regex” or regular expressions. They are anything but regular. For instance, if I want to catch a particular pattern of characters that include at least one uppercase letter, lowercase letter, digit, and special character and be at least 8 characters long, my regex code would be – and this is not a joke – “^(?=.*[A-Z])(?=.*[a-z])(?=.*\d)(?=.*[@$!%*?&])[A-Za-z\d@$!%*?&]{8,}$”… so in those cases, it’s a no-brainer to simply ask ChatGPT to write the requirements.
These models tend to be very good at anything that has been written before and very bad at things that have never been written before. In that advertising class I mentioned earlier, a great method for coming up with a headline is to write 50 variations knowing that 99% or all of them will be bad. The trick is to get the obvious stuff out of the way. If you know that all of ChatGPT’s suggestions will be obvious, you can quickly eliminate those and write 50 even better ones!
That’s not even to mention certain AI tools being included in production software. I think it’s pretty lazy to use GPT to generate an entire scene for you. But let’s say, as has happened to me, you are given a partial .PNG file and need to recreate the other half; Photoshop’s generative features will decimate that recreation time. Third-party tools like Klutz GPT will generate After Effects scripts and expressions based on text descriptions.
If something only needs to be passable, I do find myself caving to the siren song of GPT. However, if something needs to WOW, then I avoid GenAI like the plague.
Obviously, the ethics of the data scraping used to create these models are dubious as best and outside of the scope of this newsletter. However, for the time being, my suggestion to creatives would be to learn more about the limits and functions of these new GenAI tools. Keep a watchful eye on them, but rest easy at night knowing that any company that fires their creative department and replaces them with “Prompt Engineers” is going to be churning out some absolutely laughable slop shortly thereafter.
Production: Shooting a Real Estate Video For Cheap (On iPhone)
The specter of Apple is truly haunting this newsletter today.
It’s been a big first week for CaJu Creative! We signed on recently to do some real estate videography and photography and thought it’d be a good opportunity to test a proof of concept: can the job be done solely on an iPhone? The short answer is yes! The longer answer is that you need to be familiar with the kinds of shots done with much better equipment first.
The main trick is to plan your shots ahead of time and know what your phone can do. For instance, we shot on my wife’s iPhone 15 in 4k, 60fps. That’s better than my DSLR shoots. Couple that with the new Action Mode and we were able to pull off some really great sweeping shots that required almost no stabilization in post.
If you are using a DSLR or mirrorless camera to shoot photos, but you’re lacking wide-angle lenses, Photoshop and Lightroom both have photo-merging tools that can create really high-quality wide-angle shots so you can get entire rooms at once. Don’t have a nice camera, your iPhone has a panorama mode!
If you ever find yourself needing to shoot a real estate video and you’re short on equipment, leverage what you have. Can’t afford a drone? Use Google Earth Studio! Do a walkthrough of the entire house at a high frame rate and then use time ramping in post to slow down the shots when you get into a new room. Hide cuts with pillars and walls or use them to do mask wipes.
The biggest tricks to creative editing, in general, are 1. pre-planning and 2. continuous shots. That’s making a shot list so you hit everything you’re getting paid to and then thinking creatively about how the movement or content of one shot can move into another. Did you end the last shot panning up to the ceiling? Start the next one panning up from the floor. Water in this scene and the next? End the shot with a close-up and start the next with a similar closeup in the new scene.
Just be cautious about symbolic interpretation when it comes to creative cuts. Don’t shoot a child’s playroom and then smash cut to a fireplace, for instance. A further exploration of the unconscious symbolism of creative editing is planned for a future article!
We’re also planning another article exploring how we made that real estate video when it’s finished so look out for that!
Until then, make sure you’re subscribed to us on all of our channels as we’ve got a few other fun videos in the works that will be releasing soon! If you have an idea for the next newsletter, shoot us a message; I’d love to hear it!
Thanks for being the founding subscribers as well. Up up and away from here on end.
Until next time, stay fresh.
- Casey