AI's Facial Expressions: A Step Forward, But Still Room for Growth with Dall-e-3
- 10 minutes read - 2032 words
Table of Contents
Facial expressions are a powerful tool in storytelling, conveying emotions and adding depth to characters. In the realm of generative AI, the ability to create realistic and expressive faces is a crucial step towards generating truly compelling images. This blog post examines the results of a recent experiment where a generative AI model was tasked with creating images based on prompts that included specific facial expressions. The results reveal a mixed bag, with the model demonstrating a good understanding of scene and camera position, but struggling to capture the desired aesthetic in facial expressions. We’ll delve into the model’s performance, analyzing its strengths and weaknesses, and discuss the potential for future improvements.
Created with: dall-e-3
Drowning in Chaos: The Weight of Clutter
A man stands amidst a sea of disarray, his head in his hands, reflecting the overwhelming stress and anxiety of a life consumed by clutter. The image captures the suffocating feeling of being trapped in a chaotic environment.
Prompt
facial-expressions Frustration: Overwhelmed and defeated ; A single person; eye-level; Single Persons; A cluttered apartment with overflowing laundry baskets and takeout containers.; cinematic
Characteristic
Shot : A man stands in a messy room, looking distressed with his hands on his head. The room is filled with clutter, including trash bags, clothes, and cardboard boxes. The scene is set in a home or apartment, with a washing machine and other appliances visible.
Aesthetic Score : 0.2
Mood : overwhelmed, distressed, chaotic
Quality
Entropy : 6.82
Noise : 100
Prompt Clip Score : 0.23
AI Evaluation
Likelihood of AI : 0.20
Image errors : The image has some minor errors, such as a few blurry areas, slight noise in the darker areas, and some unnatural lighting.
Shadowed Figure in the Alley: A Moment of Suspense
A man cloaked in darkness, his face obscured by the shadows of a dimly lit alley. The dramatic lighting and his intense expression create a palpable sense of anticipation and mystery. What secrets lie hidden in the depths of this urban labyrinth?
Prompt
facial-expressions Frustration: Powerless and angry ; A superhero; close-up; Heroes; A dark alley with flickering streetlights, the hero’s cape billowing in the wind.; cinematic
Characteristic
Shot : A man in a superhero costume stands in a dark alley lit by street lamps. The man’s face is serious and his gaze is intense, suggesting that he is about to embark on a dangerous mission.
Aesthetic Score : 0.7
Mood : dark, mysterious, intense
Quality
Entropy : 6.17
Noise : 94
Prompt Clip Score : 0.23
AI Evaluation
Likelihood of AI : 0.90
Image errors : The image appears to be AI generated, some artifacts in the background and the man’s hair suggest this.
Chaos on the Rails: Man Caught in a Sea of Fury
A tense scene unfolds on a crowded train, where a man in a suit is the target of a hostile crowd. The dark, claustrophobic atmosphere is amplified by the angry expressions and shouts of the passengers, creating a palpable sense of aggression and violence.
Prompt
facial-expressions Frustration: Impatient and stressed ; A businessman; eye-level; Normal People; A crowded train with people pushing and shoving, the businessman trapped in the middle.; cinematic
Characteristic
Shot : A man in a suit is surrounded by people on a train. He is yelling and looks angry, while the others are either scared or angry as well.
Aesthetic Score : 0.6
Mood : intense, chaotic, aggressive
Quality
Entropy : 6.82
Noise : 100
Prompt Clip Score : 0.27
AI Evaluation
Likelihood of AI : 0.60
Image errors : Some of the characters are a bit blurry, and the lighting is uneven.
In the Shadow of the Screen: A Woman’s Tense Vigil
A solitary figure bathed in a spotlight, a woman hunches over a computer, her face etched with tension. The darkness surrounding her amplifies the intensity of the moment, hinting at a future filled with suspense and uncertainty.
Prompt
facial-expressions Frustration: Focused but frustrated ; A gamer; close-up; Gamer; A dimly lit room with a computer screen displaying a frustratingly difficult level, the gamer’s hands shaking on the keyboard.; cinematic
Characteristic
Shot : A young woman, possibly a gamer, is seen in profile, intensely focused on a computer screen, likely playing a video game. The setting is a dimly lit room, possibly a bedroom or a gaming den, with multiple computer monitors and other gaming equipment visible in the background. The lighting is dramatic, highlighting the woman’s face and the screen, while the rest of the room is shrouded in shadow.
Aesthetic Score : 0.7
Mood : intense, focused, mysterious
Quality
Entropy : 6.39
Noise : 93
Prompt Clip Score : 0.21
AI Evaluation
Likelihood of AI : 0.90
Image errors : The image appears to have some minor artifacts and blur, particularly around the edges of the subject and the background objects. The overall image quality is slightly grainy and could benefit from sharpening.
Lost in Thought: A Moment of Melancholy in the Park
A young woman sits alone on a stone path, her posture and expression conveying a deep sense of sadness and loneliness. The low camera angle emphasizes her vulnerability, creating a poignant image of despair. A bench in the background adds to the feeling of isolation.
Prompt
facial-expressions Frustration: Lonely and isolated ; A young woman; eye-level; Single Persons; A deserted park bench, the woman staring blankly at the ground, her phone lying forgotten beside her.; cinematic
Characteristic
Shot : A young woman is sitting on the ground in a park, looking sad and defeated. She is wearing casual clothing and her phone is lying on the ground next to her. There is a bench to the right of the frame.
Aesthetic Score : 0.5
Mood : sad, lonely, dejected
Quality
Entropy : 6.70
Noise : 92
Prompt Clip Score : 0.25
AI Evaluation
Likelihood of AI : 0.20
Image errors : There are no obvious image errors or artifacts.
Firefighter Battles Blaze, Desperate to Open Door
A firefighter, caught in the heart of a burning building, struggles to open a door, smoke and flames billowing through the opening. The image captures the intensity and suspense of the moment, highlighting the firefighter’s determination in the face of danger.
Prompt
facial-expressions Frustration: Urgent and desperate ; A firefighter; close-up; Heroes; A burning building with smoke billowing out, the firefighter struggling to open a door.; cinematic
Characteristic
Shot : A fireman in a burning building, yelling, with a fire visible through a window behind him. The scene is lit with a dramatic orange and blue light, and there is smoke in the air.
Aesthetic Score : 0.6
Mood : intense, dramatic, heroic
Quality
Entropy : 6.54
Noise : 105
Prompt Clip Score : 0.31
AI Evaluation
Likelihood of AI : 0.80
Image errors : The image is slightly blurry and the smoke effect looks artificial
The Weight of Knowledge: A Student’s Intense Focus
A young man sits amidst a bustling library, his gaze locked directly on the viewer. His serious expression and the surrounding students create a palpable sense of intensity and focus, capturing the pressure and dedication of academic pursuit.
Prompt
facial-expressions Frustration: Overwhelmed and anxious ; A student; eye-level; Normal People; A crowded library with students hunched over books, the student staring at a blank page, their pen hovering over the paper.; cinematic
Characteristic
Shot : A young man is sitting in a library, surrounded by other students. He looks stressed and tired, staring directly at the viewer. His eyes seem unnatural and out of place in the otherwise realistic image.
Aesthetic Score : 0.3
Mood : tense, stressed, uncomfortable
Quality
Entropy : 6.81
Noise : 85
Prompt Clip Score : 0.27
AI Evaluation
Likelihood of AI : 0.90
Image errors : The man’s eyes look unnaturally large and bulging. This creates an unsettling and distracting effect that disrupts the image’s aesthetic.
In the Zone: Gamer’s Intensity Captures the Screen
A woman, headphones on, is completely immersed in her video game. Her focused expression and the dramatic lighting create a powerful image of dedication and determination.
Prompt
facial-expressions Frustration: Focused and intense ; A gamer; close-up; Gamer; A brightly lit gaming tournament stage, the gamer staring at the screen, their controller gripped tightly in their hands.; cinematic
Characteristic
Shot : A woman, wearing a headset, is intensely focused on playing a video game, holding a controller in her hands. She’s positioned in a dimly lit, potentially competitive gaming environment, likely a tournament stage or a dedicated gaming space. The lighting highlights her face and the controller.
Aesthetic Score : 0.7
Mood : intense, focused, determined
Quality
Entropy : 6.77
Noise : 95
Prompt Clip Score : 0.27
AI Evaluation
Likelihood of AI : 0.80
Image errors : Some minor blurring around the edges of the image, particularly near the woman’s hair and the background. This could be a result of the lighting or the image processing.
Overwhelmed and Distressed: A Woman Struggles with Paperwork
A woman sits at a cluttered kitchen counter, her posture slumped and her face etched with worry as she reads through paperwork. The dim lighting and the messy surroundings amplify the sense of gloom and stress she’s experiencing.
Prompt
facial-expressions Frustration: Exhausted and defeated ; A single mother; eye-level; Single Persons; A messy kitchen with dishes piled high in the sink, the single mother staring at a pile of bills, her shoulders slumped.; cinematic
Characteristic
Shot : A woman is looking at documents with a worried expression. She is standing in a kitchen, leaning on a counter with a sink full of dirty dishes behind her. The scene is lit with a soft blue light that gives it a dramatic and moody feel.
Aesthetic Score : 0.4
Mood : dramatic, somber, lonely
Quality
Entropy : 6.69
Noise : 90
Prompt Clip Score : 0.26
AI Evaluation
Likelihood of AI : 0.30
Image errors : The image is slightly over-sharpened, resulting in a slightly unnatural look. The lighting is uneven, with some areas appearing too dark and others too bright. The subject’s hair appears somewhat unnatural and slightly pixelated.
Doctor’s Worried Gaze Reflects Patient’s Uncertain Future
A dimly lit hospital room, a doctor’s concerned expression, and a patient lying passively in bed create a palpable sense of urgency and tension. The scene evokes a feeling of seriousness and concern, highlighting the gravity of the situation.
Prompt
facial-expressions Frustration: Concerned and helpless ; A doctor; close-up; Heroes; A hospital room with a patient hooked up to machines, the doctor looking at a medical chart with a furrowed brow.; cinematic
Characteristic
Shot : A doctor, a young woman, looks concerned at the patient in the hospital bed. There is a medical monitor showing a heartbeat in the background.
Aesthetic Score : 0.6
Mood : serious, concerned, somber
Quality
Entropy : 6.87
Noise : 86
Prompt Clip Score : 0.27
AI Evaluation
Likelihood of AI : 0.20
Image errors : No visible artifacts or errors. The lighting is a little dark and there are some shadows on the faces.
Conclusion
The results show that the generative AI model performed well in understanding the scene and camera position, but struggled with the aesthetic aspect. Here’s a breakdown:
- Camera Position: The model scored 0.31, which is below the “good” range of 0.5 to 0.75. This indicates that the model didn’t fully capture the intended camera position in the prompt.
- Shot Analysis: The model scored 0.58, which falls within the “good” range. This means the model was able to understand the scene in the prompt reasonably well.
- Aesthetic Analysis: The model scored 0.24, which is significantly higher than the “very good” range of -0.2 to 0.1. This suggests that the generated image’s aesthetic deviated significantly from the expected aesthetic described in the prompt.
Overall, the model shows promise in understanding scene and camera position, but needs improvement in generating images that match the desired 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://openai.com/index/dall-e-3/