Another week another generative AI comparison. Today it’s reigning champ versus the offering from a little upstart named Google called ImageFX.  I’m still shocked they went with the most generic name possible for their most advanced work.

ImageFX comes out of Google’s AI Test Kitchen.  One interesting feature of ImageFX are the option for changing the image.  You add a prompt, ImageFX quickly shuffles it for better ordering and then provides dropdown menus for the key features.  So you can quickly change an oil lamp to an electric one or update the style of a hat without leaving or creating a new prompt.  See below.

Onto the comparison.  With the commercial work, I grade on how easily fooled would I be if the art was used in their commercial work.

---

Prompt 1: Victorian-Era London

• Description: Illustrate a Victorian-era street in London, shrouded in thick fog. The cobblestone streets are wet from recent rain, and gas lamps emit a soft, yellowish glow, casting long shadows through the mist. People in period clothing—top hats, coats, and dresses—move through the fog, some with umbrellas, others hurriedly walking past. A horse-drawn carriage is parked on the side of the street, and shop windows with ornate signage are partially visible through the haze. The scene should capture the atmosphere of a cold, foggy night, with a sense of mystery and anticipation in the air, as if something is about to happen.
• Midjourney:
               
• ImageFX:
  
  

Comparison: Midjourney captures the gothic atmosphere with more realism.  While ImageFX looks like the cover of the Magic Treehouse book.  s

---

Prompt 2: Epic Fantasy Battle

• Description: Illustrate a large-scale historical battle taking place on an open field. The armies should resemble those from the medieval period, with knights in armor, archers, and cavalry. However, add fantasy elements such as dragons flying overhead, wizards casting spells from the sidelines, and magical creatures like griffins and trolls participating in the fight. The landscape is rugged with rolling hills and a castle in the distance under siege. Smoke and fire from the battle obscure parts of the scene, and the overall mood is intense and chaotic, capturing the epic scale of the conflict.
• Midjourney:
  
  
• ImageFX:
  

Comparison: Both tools create a dramatic scene, but Midjourney looks more like an accurate historical painting of a battle while again.  ImageFX looks very middle-school.

---

Prompt 3: Adidas Soccer Ball

• Description: A realistic close-up image of an Adidas soccer ball resting on the grass of a professional soccer field. The ball’s black and white design is clean and new, with the Adidas logo prominently displayed. The background shows the stadium seats blurred out, with a few players visible in the distance, warming up for a game. The grass is lush and perfectly manicured, and the scene is lit by bright stadium lights, emphasizing the importance of the match. The image captures the anticipation and excitement of soccer.
• Midjourney:
  
• ImageFX:
  

Comparison: ImageFX nails the realism with impeccable detail on the ball and grass, perfect for commercial use. Midjourney’s version avoids people in the background for the most part and is serviceable but boring.

---

Prompt 4: Starbucks Coffee on a Desk

• Description: A realistic image of a Starbucks coffee cup sitting on a busy work desk. The cup is a standard tall size, with the green Starbucks logo clearly visible on the white cup. The desk is cluttered with work essentials like a laptop, notepads, pens, and a pair of glasses. A small succulent plant and a desk lamp add to the cozy, productive atmosphere. The background includes a modern office space with natural light streaming through large windows, creating a productive and focused environment.
• Midjourney:
  
• ImageFX:
  

Comparison: Midjourney looks 90% like an actual photograph while ImageFX looks maybe 40%.

---

Prompt 5: Toyota Camry in Suburban Neighborhood

• Description: A realistic image of a Toyota Camry parked in a quiet suburban neighborhood. The car is a new model, with a shiny, polished exterior and the Toyota emblem prominently displayed on the grille. The background includes a row of well-maintained houses, green lawns, and a tree-lined street. The car is positioned in a driveway, with the sun setting in the background, casting a warm glow over the scene. The image conveys reliability and comfort, typical of the Toyota brand.
• Midjourney:
  
• ImageFX:
  

Comparison: Midjourney provides an almost photographic quality, perfect for advertisements, while Midjourney delivers a artistic rendering.

---

Prompt 6: Coca-Cola Can at a Picnic

• Description: A realistic image of a classic Coca-Cola can sitting on a wooden picnic table during a sunny afternoon. The can is cold, with condensation droplets visible on its surface, glistening in the sunlight. The background shows a lively outdoor setting, with a red-and-white checkered picnic blanket, a basket filled with snacks, and trees providing shade. Nearby, a group of friends is laughing and enjoying the day, adding to the feeling of summer fun. The Coca-Cola logo is clearly visible, and the overall scene radiates warmth and nostalgia.
• Midjourney:
  
• ImageFX:
  

Same as with the Starbucks and Toyota.  Midjourney looks real but not compelling while ImageFx looks rendered but with a story to tell.

Prompt 7: Depression Bear

• Description: Design a stuffed animal that represents the concept of depression. The toy should have a somber and melancholic appearance, with droopy eyes and a slouched posture. Its fur is a muted, dark gray or blue, conveying a sense of heaviness. The stuffed animal’s fabric should look slightly worn, as if it has been hugged tightly and often. Its expression should be subtly sad, with downturned features and an overall sense of loneliness and fatigue. The design should evoke empathy and understanding, symbolizing the emotional weight and isolation often associated with depression.
• Midjourney:
  
  
• ImageFX:
  
  

Prompt 8: 1927 Yankees take on the 1999 Diamondbacks

• Description:  Create a historically accurate black-and-white photograph depicting the 1999 Arizona Diamondbacks playing against the 1927 New York Yankees at the old Yankee Stadium. The scene should capture the essence of the era with the players wearing their respective vintage uniforms. The photograph should show a moment during the game, such as a pitch being thrown or a player sliding into a base. The stadium should be filled with period-appropriate details, such as old-fashioned billboards, fans in 1920s attire, and the classic architecture of Yankee Stadium. The overall image should look like an authentic historical photograph, capturing the drama and intensity of the game in black and white.
• Midjourney:
  
  
• ImageFX:
  

Comparison: I could replay this prompt all day and never get what I’m looking for.  AI doesn’t understand baseball which is probably why it’s less popular than before, it’s complicated.  There aren’t any historical players, but ImageFX got the finer points of old Yankee Style right.

Google could take Midjourney down the road.  But right now Midjourney still rules.