Hilal

AI Content For Brands | Ex-Journalist/Ex-Lawyer / We are this planet's kidney stones.

[Hilal @MyRedSelf]

Just finished and submitted my new AI commercial “Time To Grow Up” for Runway’s Big Ad Contest for products that don’t exist. #runwaybigadcontest

[Hilal @MyRedSelf]

@deadsun_0 Oh, that automatically comes with the title anyway, don’t worry about it lol

[Hilal @MyRedSelf]

This just in!

The tools I used:
Midjourney
Nano Banana Pro
GPT Images 2.0

#breakingnews #aiartdaily #aiartist #midjourneyart

[Hilal @MyRedSelf]

If life gave you a sink, just swim in it. 🤭

The tools:
@midjourney
GPT Images 2.0
Canva

#midjourneyart #midjourneyartist #aiartist

[Hilal @MyRedSelf]

Doomscrolling is the new smoking.

Everyone knows it’s bad for you and keeps doing it anyway.

#midjourney + GPT Images 2.0 + Canva

#midjourneyart

[Hilal @MyRedSelf]

“It’s okay not to be okay.”

Made with #Midjourney + GPT Images 2.0

#midjourneyart

[Hilal @MyRedSelf]

“Just training.”

Made with #Midjourney + GPT Images 2.0 + Canva

#barbiecore #aiartist

[Hilal @MyRedSelf]

“Take a seat.”

#Midjourney + GPT Images 2.0

[Hilal @MyRedSelf]

Striving for happiness… 🌸

Made with @midjourney + GPT Images 2.0

#drinkpoppi #poppi #specad

[Hilal @MyRedSelf]

Money can’t buy happiness, but it can buy coffee. ☕️

#aiartdaily #aiartist #midjourneyart

[Hilal @MyRedSelf]

Everyone calls you crazy until…

[Hilal @MyRedSelf]

@martinleblanc @magnific Can’t wait! 👀

[Hilal @MyRedSelf]

I made a #LiquidDeath spec ad campaign "photoshoot", using Midjourney and GPT Images 2.0. (The whole set of images are on my LinkedIn page).

The challenges:

* This was the first time I worked with tattoos, and it was very stressful. Larger and less detailed tattoos could be relatively easier to regenerate. But it is almost never the same tattoos in
regenerations when you are dealing with smaller tattoos, like hand or finger tattoos.

* GPT Images (both 2.0 and 2.5) is still struggling with human anatomy, especially when it comes to "crossed legs". For my legs shot you can see below, I had to regenerate my shot dozens of times, despite using reference images. The legs were either too thin or too long, OR the leg that is behind was fragmented into irrational sections.

[Hilal @MyRedSelf]

@ichelpark @midjourney I think it mostly maintained it frame to frame.

[Hilal @MyRedSelf]

Here comes my #Starbucks spec ad celebrating the Pumpkin Spice Latte season and the beginning of fall.

What if PSL existed in the 90s? I believe it would have been a go-to morning drink for white-collar employees in corporate settings.

The tools I used:
@midjourney
GPT Images 2.0
Seedance 2.5 via Magnific

What’s your favorite fall drink?

#psl #pumpkinspicelatte #pumpkinspiceeverything

[Hilal @MyRedSelf]

@ichelpark @midjourney It was included both in the Midjourney and the Seedance prompts.

[Hilal @MyRedSelf]

The #PumpkinSpiceLatte season is in full swing. And as a #PSL fan, I had to create something to celebrate the beginning of the fall, as well as the ongoing PSL season.

I wanted to design a #Starbucks spec ad campaign with this question in mind: "What if PSL existed in the 90's?" Surely, it would have been a go-to morning drink for white-collar employees. So here’s some of the frames I made as part of it. This will be followed by a video version very soon.

The tools I used:
@midjourney
GPT Images 2.0

What is your favorite fall drink?

[Hilal @MyRedSelf]

This is the short version. You can find the full version on my LinkedIn page.

[Hilal @MyRedSelf]

I re-created the most viral #Pringles ad with my own characters and in a different style.

Here are some of my takeaways: (And here's a link to the original Pringles commercial for comparison: https://t.co/ZFV6xk0y7T )

* One of the biggest challenges in AI filmmaking is "acting," especially when subtle face expressions or humorous conversations are involved. In other words: Fine-tuning your prompts to make your characters "act" in the exact way you have in mind.

* It is still problematic to make your characters turn their eyes toward a certain direction without moving their heads. (Both image and video models struggle with this.)

* Even if you provide a reference image for positioning your character in a certain way, AI models may still need verbal descriptions in addition to that. In my experience, reference images were almost never enough. I was only able to get the results I wanted, after I openly described what was going on in my reference images.

* I got to test GPT Images 2.5 and quickly realized that it is unfortunately worse than GPT 2 in some aspects like creating vast areas (e.g., sky, forest, greenery, ocean, etc.)

The tools I used:

Midjourney
GPT Images 2 (via @magnific / formerly Freepik )
Seedance 2.0 (via @magnific / formerly Freepik )
Seedance 2.5 (via @magnific / formerly Freepik )
@ElevenLabs
@capcutapp

[Hilal @MyRedSelf]

@kitten_meow34 lol 😆