AI Struggles with Facial Expressions: A Study in Aesthetics with Midjourney
- 9 minutes read - 1800 wordsTable of Contents
Facial expressions are a powerful tool for conveying emotions and intentions. In the realm of generative AI, accurately capturing these expressions is a significant challenge. This blog post delves into the performance of a generative AI model in understanding and generating images with specific facial expressions. We’ll explore the model’s strengths and weaknesses, analyzing its ability to capture camera position, shot composition, and aesthetic elements. Through this analysis, we’ll gain insights into the current state of AI’s ability to portray human emotions visually.
Created with: midjourney
Lost in the City Lights
A young woman with curly hair stands silhouetted against the vibrant cityscape, her gaze lost in the twinkling lights. The moody atmosphere and dramatic contrast create a sense of mystery and intrigue, inviting you to explore the hidden stories of the urban night.
Prompt
Agreement pensive, resigned: melancholy, contemplative ; person; eye-level, close-up; Single Person; background a bustling city street at night; cinematic
Characteristic
Shot : A young woman with curly hair is standing in the middle of a city street at night, looking up at the brightly lit buildings. The city lights create a colorful bokeh effect in the background.
Aesthetic Score : 0.7
Mood : melancholy, hopeful, contemplative
Quality
Entropy : 6.36
Noise : 102
Prompt Clip Score : 0.24
AI Evaluation
Likelihood of AI : 0.20
Image errors : Slight chromatic aberration in the city lights. The bokeh effect is a bit too artificial and overdone.
Heroic Silhouette Against the Flames
A dramatic image of a silhouetted superhero standing on a rooftop overlooking a burning city. The figure is illuminated by the flames and smoke, creating a sense of urgency and heroism in the face of destruction.
Prompt
Agreement focused, unwavering: determined, resolute ; A superhero standing tall; eye-level, low-angle; Hero; background a cityscape with a burning building in the background; cinematic
Characteristic
Shot : A lone figure in a dark suit stands on a rooftop, silhouetted against a cityscape engulfed in flames.
Aesthetic Score : 0.6
Mood : dramatic, heroic, ominous
Quality
Entropy : 6.77
Noise : 91
Prompt Clip Score : 0.29
AI Evaluation
Likelihood of AI : 0.70
Image errors : The flames appear somewhat artificial and the overall texture of the image is a bit blurry.
Family Dinner: A Moment of Warmth and Togetherness
A family of four gathers around a dimly lit kitchen table, bathed in warm light that creates a sense of intimacy and togetherness. Their happy expressions radiate warmth and love, capturing the essence of a cozy family dinner.
Prompt
Agreement smiling, relaxed: peaceful, content ; A family gathered around a dinner table; eye-level; Normal People; a cozy kitchen with warm lighting; cinematic
Characteristic
Shot : A family is having dinner together at a table in a dimly lit kitchen. The scene is warm and inviting, with a focus on the family’s happiness and togetherness.
Aesthetic Score : 0.7
Mood : warm, inviting, cozy
Quality
Entropy : 6.50
Noise : 93
Prompt Clip Score : 0.26
AI Evaluation
Likelihood of AI : 0.10
Image errors : There are no visible artifacts or errors in the image.
Lost in the Code: A Man’s Intense Focus Under Neon Lights
A close-up shot captures a bearded man in a black beanie, his eyes glued to a computer screen bathed in vibrant blue and pink neon light. The intensity of his gaze and the dramatic lighting create a sense of suspense and focus, hinting at a world of digital possibilities.
Prompt
Agreement concentrated, determined: excited, engaged ; A gamer intensely focused on a screen; eye-level, close-up; Gamer; a dimly lit room with neon lights reflecting on the screen; cinematic
Characteristic
Shot : A close-up shot of a young man with a beard wearing a beanie and glasses. The man is looking intently at a computer screen, which is illuminated by a pink light. The scene is lit with vibrant blue and purple lights, creating a futuristic and somewhat mysterious atmosphere.
Aesthetic Score : 0.6
Mood : mysterious, futuristic, intense
Quality
Entropy : 6.74
Noise : 101
Prompt Clip Score : 0.23
AI Evaluation
Likelihood of AI : 0.40
Image errors : No significant errors in the image, except for possible minor smoothing of the skin in the image, which is a potential indication of artificial intelligence.
Lost in Thought: A Moment of Melancholy Beauty
A young woman with fiery red hair and freckles gazes upwards, her expression a blend of contemplation and longing. Leaning against a white brick wall, she becomes a focal point against a softly blurred background, creating an intimate and vulnerable atmosphere. The lighting and composition evoke a sense of dreamy melancholy, capturing a fleeting moment of introspection.
Prompt
Agreement thoughtful, wistful: reflective, introspective ; A woman ; eye-level, close-up; Single Person; background a row of old, brick buildings with faded paint; cinematic
Characteristic
Shot : A young woman with freckles and red hair is looking up and to the side, with a white brick wall behind her.
Aesthetic Score : 0.7
Mood : pensive, wistful, introspective
Quality
Entropy : 6.74
Noise : 101
Prompt Clip Score : 0.22
AI Evaluation
Likelihood of AI : 0.10
Image errors : No significant errors.
Man Faces the Storm’s Fury
A solitary figure stands defiant against a raging storm, illuminated by flashes of lightning. The dramatic scene evokes a sense of intensity and somber reflection.
Prompt
Agreement angry, determined: powerful, defiant ; A hero raising; eye-level, high-angle; Hero; a dark, stormy sky with lightning flashing in the background; cinematic
Characteristic
Shot : A man is standing in a stormy sky with lightning and rain. The man’s face is contorted in a grimace, as if he is in pain or fear.
Aesthetic Score : 0.6
Mood : dramatic, intense, fearful
Quality
Entropy : 6.39
Noise : 90
Prompt Clip Score : 0.24
AI Evaluation
Likelihood of AI : 0.80
Image errors : The rain is repetitive and appears unrealistic. The lightning is also a bit too artificial. The man’s expression is exaggerated, and the color grading makes the scene look a bit flat. The image has a bit of noise, which is most noticeable in the dark areas.
Sunlight and Laughter: A Moment of Joy in the Garden
Three friends share a moment of pure joy in a sun-drenched garden. The warm light filters through the leaves, creating a sense of intimacy and seclusion as they laugh together. This heartwarming scene captures the essence of friendship and carefree happiness.
Prompt
Agreement laughing, smiling: joyful, carefree ; A group of friends laughing together; eye-level; Normal People; a sunny park with trees and flowers; cinematic
Characteristic
Shot : Three young adults laughing together, surrounded by lush green foliage and vibrant flowers.
Aesthetic Score : 0.7
Mood : joyful, carefree, vibrant
Quality
Entropy : 6.64
Noise : 84
Prompt Clip Score : 0.24
AI Evaluation
Likelihood of AI : 0.20
Image errors : The image has some slight blurriness, particularly around the edges, and some noise in the darker areas.
Victory Dance! Gamer Celebrates Triumph in a Shower of Confetti
This image captures the pure joy of victory as a young gamer, headphones on, erupts in celebration from his gaming chair. Confetti swirls around him, adding to the excitement and creating a dynamic and celebratory atmosphere. The lighting and composition enhance the dramatic effect, showcasing the raw emotion of the moment.
Prompt
Agreement excited, jubilant: triumphant, ecstatic ; A gamer celebrating a victory; eye-level, close-up; Gamer; a brightly lit room with confetti and streamers; cinematic
Characteristic
Shot : A young man in headphones is celebrating a victory with confetti falling around him. He is sitting in a gaming chair, with a computer screen in the background.
Aesthetic Score : 0.7
Mood : joyful, energetic, triumphant
Quality
Entropy : 6.80
Noise : 102
Prompt Clip Score : 0.29
AI Evaluation
Likelihood of AI : 0.50
Image errors : The confetti seems slightly artificial and the lighting is slightly overexposed, resulting in a slightly artificial feel.
Lost in Thought: A Moment of Melancholy in the Park
A young man, shrouded in shadow, sits amidst fallen leaves in a park. His posture and the low angle of the image evoke a sense of loneliness and contemplation, capturing a somber mood amidst the overcast sky.
Prompt
Agreement sad, resigned: lonely, melancholic ; A man sitting; eye-level, mid-shot-or-medium-shot; Single Person; background a deserted park with fallen leaves; cinematic
Characteristic
Shot : A young man sits alone in a park, surrounded by fallen leaves, with the path leading into a blurred background of trees.
Aesthetic Score : 0.7
Mood : melancholy, contemplative, introspective
Quality
Entropy : 6.90
Noise : 109
Prompt Clip Score : 0.29
AI Evaluation
Likelihood of AI : 0.20
Image errors : No visible artifacts or errors.
Silhouette of Solitude: A Figure Lost in the City Lights
A lone figure stands on a rooftop, their silhouette stark against the vibrant cityscape. The mood is dark and contemplative, hinting at a sense of isolation and mystery. The dramatic effect of the scene evokes a feeling of loneliness and introspection.
Prompt
Agreement confident, resolute: determined, hopeful ; A hero on a rooftop; eye-level, high-angle, over-the-shoulder; Hero; background a panoramic view of a city skyline at night; cinematic
Characteristic
Shot : A lone figure stands on a rooftop overlooking a city skyline at night. The city is illuminated by countless lights, creating a dazzling display. The sky is cloudy and dark, adding to the moody atmosphere.
Aesthetic Score : 0.8
Mood : dramatic, urban, melancholic
Quality
Entropy : 6.05
Noise : 85
Prompt Clip Score : 0.25
AI Evaluation
Likelihood of AI : 0.20
Image errors : Slight graininess and some noise in the shadows.
Conclusion
The results show that the generative AI model performed well in understanding the camera position and shot composition, but struggled with the aesthetic aspect.
Here’s a breakdown:
- Camera Position: The model scored 0.45, which is considered below average. This suggests that the model didn’t accurately capture the intended camera position described in the prompt.
- Shot Analysis: The model scored 0.58, which is considered average. This indicates that the model was able to understand the scene and shot composition in the prompt to a reasonable degree.
- Aesthetic Analysis: The model scored 0.07, which is considered poor. This means that the generated image’s aesthetic significantly deviated from the expected aesthetic described in the prompt.
Overall, the model demonstrated a decent understanding of the scene and shot composition, but struggled to accurately capture the intended camera position and aesthetic.
Sources:
- https://dramaresource.com/storytelling/
- https://seedsoftellers.eu/resources/the-body-language-for-young-tellers/
- https://digitalcollections.sit.edu/cgi/viewcontent.cgi?article=1288&context=sandanona&filename=1&type=additional
- https://citeseerx.ist.psu.edu/document?doi=7f842882e9bb1fa2c0e96939bc8d2c37e34e17c0&repid=rep1&type=pdf
- https://www.twinkl.co.uk/search?q=drama+facial+expression
- https://midjourney.com