AI's Facial Expressions: A Step Forward, But Still Room for Growth with Imagen-v2
- 9 minutes read - 1841 wordsTable of Contents
Facial expressions are a powerful tool for conveying emotions and intentions. In the realm of artificial intelligence, the ability to generate realistic and expressive faces is a crucial step towards creating more engaging and believable virtual experiences. This blog post examines the performance of a generative AI model in capturing facial expressions, highlighting its strengths and weaknesses, and exploring the potential for future advancements in this area.
Created with: imagen-v2
Radiant Smile, Vibrant City
A close-up portrait captures the joy and energy of a young woman with afro hair, her bright smile radiating against a blurred urban backdrop. The warm lighting and vibrant yellow jacket amplify the feeling of happiness and optimism.
Prompt
facial-expressions Happiness: Joyful, carefree ; Single person; eye-level; Single Persons; A bustling city street with vibrant colors and people going about their day.; cinematic
Characteristic
Shot : A young woman with an afro hairstyle is smiling broadly and looking up. The background is blurred, but shows she is likely walking in a city with buildings on either side of her.
Aesthetic Score : 0.8
Mood : joyful, positive, carefree
Quality
Entropy : 6.72
Noise : 76
Prompt Clip Score : 0.24
AI Evaluation
Likelihood of AI : 0.20
Image errors : There are no visible image errors.
Conquering the Summit: A Moment of Triumph
A muscular man stands victorious atop a rocky mountain peak, silhouetted against a breathtaking sunset. His raised arms and determined expression embody the spirit of achievement and the power of human resilience.
Prompt
facial-expressions Happiness: Triumphant, proud, relieved ; Hero; eye-level; Heroes; A hero standing triumphantly on a mountain peak, with a breathtaking sunset behind them.; cinematic
Characteristic
Shot : A muscular man stands on a mountaintop with his arms raised in victory, the setting sun behind him.
Aesthetic Score : 0.6
Mood : triumphant, powerful, epic
Quality
Entropy : 6.78
Noise : 118
Prompt Clip Score : 0.27
AI Evaluation
Likelihood of AI : 0.70
Image errors : Some artifacts and blurring are present, particularly in the background.
Friends Sharing Laughter and Joy at a Park Picnic
A group of friends gather around a picnic table, their laughter filling the air. The table is laden with food and drinks, creating a scene of joyful camaraderie. The lush green background adds to the sense of lightheartedness and friendship.
Prompt
facial-expressions Happiness: Warm, intimate, joyful ; Normal people; eye-level; Normal People; A group of friends laughing and sharing a meal at a picnic table in a park.; cinematic
Characteristic
Shot : A group of friends enjoying a picnic outdoors, sitting at a wooden picnic table and laughing. It appears to be a sunny day, as there is a lot of natural light and shadows. There is a forest in the background.
Aesthetic Score : 0.7
Mood : joyful, friendly, relaxed
Quality
Entropy : 6.69
Noise : 63
Prompt Clip Score : 0.27
AI Evaluation
Likelihood of AI : 0.10
Image errors : The image is slightly overexposed, causing the background to be washed out and losing detail.
Caught in the Moment: A Man’s Shocked Reaction
A close-up portrait captures a man’s intense reaction, his mouth wide open in a shout of surprise and excitement. The headphones on his head suggest he might have just heard something unexpected, leaving him caught in the moment.
Prompt
facial-expressions Happiness: Excited, exhilarated, triumphant ; Gamer; close-up; Gamer; A gamer’s face lit by the screen, eyes wide with excitement as they celebrate a victory.; cinematic
Characteristic
Shot : Close-up portrait of a man wearing headphones, with a wide open mouth, seemingly in shock or excitement.
Aesthetic Score : 0.3
Mood : surprised, excited, intense
Quality
Entropy : 6.06
Noise : 64
Prompt Clip Score : 0.30
AI Evaluation
Likelihood of AI : 0.70
Image errors : The image has some minor artifacts, particularly around the edges of the headphones and the hair. There is a slight blurriness in the background.
Golden Hour Glow: A Moment of Joy in a Field of Flowers
Capture the essence of happiness with this stunning image. A woman with long brown hair stands bathed in the warm light of the setting sun, her smile radiating joy and carefree spirit. The dramatic lighting and gentle breeze create a sense of warmth and beauty, making this a truly captivating scene.
Prompt
facial-expressions Happiness: Free, joyful, carefree ; Single person; eye-level; Single Persons; A woman dancing freely in a field of wildflowers, bathed in golden sunlight.; cinematic
Characteristic
Shot : A young woman with long brown hair is standing in a field of flowers, looking off to the side with a happy expression. The sun is setting in the background, casting a warm glow on the scene.
Aesthetic Score : 0.8
Mood : happy, warm, romantic
Quality
Entropy : 6.74
Noise : 65
Prompt Clip Score : 0.24
AI Evaluation
Likelihood of AI : 0.80
Image errors : There are a few minor artifacts in the hair, particularly around the edges. The image appears to have been slightly oversharpened.
Silhouettes of Hope in the Desert Sunset
A lone figure walks towards a majestic rock formation, their silhouette stark against the fiery hues of a desert sunset. The scene evokes a sense of adventure, hope, and dramatic beauty.
Prompt
facial-expressions Happiness: Brave, heroic, selfless ; A lone hiker, silhouetted against the setting sun, races across a vast, windswept plain, determined to reach a distant, towering rock formation before a sudden storm breaks.; cinematic
Characteristic
Shot : A lone figure walks across a desert landscape at sunset with a large rock formation in the background.
Aesthetic Score : 0.7
Mood : epic, vast, mysterious
Quality
Entropy : 6.70
Noise : 102
Prompt Clip Score : 0.27
AI Evaluation
Likelihood of AI : 0.20
Image errors : The image has some minor artifacts around the figure’s silhouette, suggesting it was potentially edited or processed in post-production. These artifacts are not extremely noticeable, but they detract slightly from the overall image quality.
Campfire Nights: Friends, Stars, and Warmth
A cozy scene of four friends gathered around a campfire under a starlit sky. The warm glow of the fire and the soft light of the stars create a sense of camaraderie and wonder. This image evokes feelings of warmth, friendship, and the simple joys of life.
Prompt
facial-expressions Happiness: Warm, cozy, loving ; A group of friends gathered around a campfire, sharing stories and laughter under a starlit sky.; cinematic
Characteristic
Shot : Four young adults are huddled around a campfire under a starry night, they seem to be enjoying each other’s company and the warmth of the fire.
Aesthetic Score : 0.7
Mood : cozy, warm, adventurous
Quality
Entropy : 6.41
Noise : 110
Prompt Clip Score : 0.31
AI Evaluation
Likelihood of AI : 0.80
Image errors : The image appears to have some minor artifacts, especially in the background. The starry night sky looks a bit too blurry and the texture of the ground looks somewhat artificial.
The Focus of a Champion
A young man, eyes glued to the screen, navigates a virtual world with unwavering determination. The soft, warm lighting adds to the intensity of the moment, capturing the pure focus of a gamer in the heat of the game.
Prompt
facial-expressions Happiness: Focused, determined, absorbed ; Gamer; close-up; Gamer; A gamer’s hands deftly navigating a game controller, with a look of intense focus and concentration.; cinematic
Characteristic
Shot : A young man with glasses is playing a video game, holding a controller in his hands.
Aesthetic Score : 0.7
Mood : intense, focused, determined
Quality
Entropy : 6.48
Noise : 60
Prompt Clip Score : 0.26
AI Evaluation
Likelihood of AI : 0.80
Image errors : The image has some artifacts in the background and on the man’s skin. The lighting is a bit artificial, making the image less natural
Golden Hour Reflections
A man, lost in thought, gazes towards the setting sun in a bustling park. The soft light casts a warm glow, creating a sense of calm and nostalgia. His blue jacket blends with the fading sky, adding to the serene atmosphere.
Prompt
facial-expressions Happiness: Peaceful, content, nostalgic ; Single person; eye-level; Single Persons; A man sitting on a bench in a park, watching children play, with a gentle smile on his face.; cinematic
Characteristic
Shot : A man standing in a park, looking off to the side with a soft smile on his face. The background is blurred and out of focus, suggesting a warm and inviting atmosphere.
Aesthetic Score : 0.75
Mood : nostalgic, contemplative, hopeful
Quality
Entropy : 6.85
Noise : 71
Prompt Clip Score : 0.22
AI Evaluation
Likelihood of AI : 0.25
Image errors : The image is slightly overexposed and the colors are a bit too saturated. There are some artifacts visible in the background, especially around the trees.
Wonder Woman Triumphant: A Moment of Hope and Celebration
A powerful image captures Wonder Woman’s radiant smile as she basks in the glow of victory. Red petals cascade around her, symbolizing a triumphant moment amidst a cheering crowd. The shallow depth of field draws the viewer into her joy, creating a sense of intimacy and shared celebration.
Prompt
facial-expressions Happiness: Triumphant, victorious, celebrated ; Hero; wide shot; Heroes; A hero standing tall, surrounded by cheering crowds, after achieving a great victory.; cinematic
Characteristic
Shot : A woman, likely a superhero, is smiling triumphantly in the midst of a battle. Red petals are falling from the sky around her. It looks like a celebration of victory.
Aesthetic Score : 0.7
Mood : triumphant, celebratory, joyful
Quality
Entropy : 6.50
Noise : 60
Prompt Clip Score : 0.23
AI Evaluation
Likelihood of AI : 0.90
Image errors : The image has some artifacts, particularly in the background and the petals, which look a bit blurry and artificial. The woman’s hair seems to have some unnatural texture.
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 0.375, which is below the “good” range of 0.5 to 0.75. This suggests that the model didn’t fully capture the intended camera position described in the prompt.
- Shot Analysis: The model scored 0.59, falling within the “good” range. This indicates that the model was able to understand the scene and create a shot that was generally consistent with the prompt.
- Aesthetic Analysis: The model scored 0.15, which is outside 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 demonstrated a decent understanding of the scene and shot composition, but struggled to achieve 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://deepmind.google/technologies/imagen-2/