AI's Facial Expressions: A Mixed Bag of Success with Titan-g1
- 9 minutes read - 1795 wordsTable of Contents
In the realm of artificial intelligence, generating realistic facial expressions is a challenging task. This blog post delves into the results of an experiment that tested a generative AI model’s ability to capture facial expressions across diverse scenes. The model demonstrated a mixed bag of success, excelling in capturing the desired aesthetic style but struggling with accurately representing camera position and scene details. We’ll explore these findings in detail, providing insights into the model’s strengths and weaknesses, and highlighting the ongoing journey towards achieving truly expressive AI-generated imagery.
Created with: titan-g1
Lost in Thought: A Moment of Contemplation
A young woman finds solace in a warm cafe, her gaze lost in the distance. The soft lighting and delicate details create a sense of intimacy and quiet reflection. Is she lost in memories, or dreaming of the future? This image captures the essence of pensive contemplation.
Prompt
facial-expressions Gratitude: Contentment and appreciation for solitude ; Single woman; eye-level; Single Persons; cozy cafe with warm lighting; cinematic
Characteristic
Shot : A young woman is sitting at a table in a cafe, looking out the window. There are flowers in the foreground and a cup of coffee on the table.
Aesthetic Score : 0.7
Mood : calm, thoughtful, wistful
Quality
Entropy : 6.58
Noise : 100
Prompt Clip Score : 0.21
AI Evaluation
Likelihood of AI : 0.10
Image errors : The image has slight noise and blur in the background.
Firefighter’s Heroic Rescue Brings Hope Amidst the Flames
A heartwarming scene unfolds as a firefighter rescues a smiling child from a burning building. The contrast between the danger and the child’s safety creates a powerful dramatic effect, highlighting the bravery and compassion of the firefighter.
Prompt
facial-expressions Gratitude: Relief, gratitude for the hero’s bravery ; Firefighter rescuing a child from a burning building; wide shot; Heroes; smoke and flames in the background; cinematic
Characteristic
Shot : A firefighter is holding a young child in their arms, the child is smiling. There is a fire in the background.
Aesthetic Score : 0.7
Mood : hopeful, dramatic, heartwarming
Quality
Entropy : 6.94
Noise : 97
Prompt Clip Score : 0.31
AI Evaluation
Likelihood of AI : 0.20
Image errors : Slight blur on the edges of the image
Family Laughter: A Moment of Joy Captured
This heartwarming image depicts a family gathered around a table, radiating joy and connection. Their smiles and laughter create a warm and inviting atmosphere, capturing the essence of a festive occasion.
Prompt
facial-expressions Gratitude: Warmth, appreciation for family and connection ; Family having dinner together; eye-level; Normal People; warm, inviting kitchen; cinematic
Characteristic
Shot : A family gathering around a table, likely celebrating a holiday or special occasion. The table is set with candles and food, and the people are laughing and enjoying each other’s company.
Aesthetic Score : 0.6
Mood : joyful, warm, festive
Quality
Entropy : 6.75
Noise : 97
Prompt Clip Score : 0.21
AI Evaluation
Likelihood of AI : 0.20
Image errors : The image appears to be slightly overexposed, with some highlights blown out.
Victory Dance! Gamer Celebrates Triumph with Friends
A group of friends gather for a lively gaming session, with one young man in the foreground radiating pure joy as he celebrates a victory. His infectious smile and raised fist capture the energy and excitement of the moment, highlighting the competitive spirit and camaraderie of the group.
Prompt
facial-expressions Gratitude: Excitement, gratitude for the shared experience ; Gamer celebrating a victory with friends; close-up; Gamer; brightly lit gaming room with screens and controllers; cinematic
Characteristic
Shot : A group of friends are playing a video game. One of them is very excited and has his fists raised in the air.
Aesthetic Score : 0.7
Mood : joyful, excited, competitive
Quality
Entropy : 6.94
Noise : 103
Prompt Clip Score : 0.29
AI Evaluation
Likelihood of AI : 0.20
Image errors : Some noise and blurriness in the background
Silhouettes of Hope: A Man Finds Tranquility at Sunset
A solitary figure, clad in a plaid shirt, stands amidst a field, his gaze fixed on the horizon as the sun dips below the horizon. The warm hues of the setting sun paint the scene with a sense of tranquility and hope, capturing a moment of quiet contemplation.
Prompt
facial-expressions Gratitude: Awe, gratitude for the beauty of nature ; Man looking out at a beautiful sunset; eye-level; Single Persons; vast, open field with golden light; cinematic
Characteristic
Shot : A man in a plaid shirt is standing in a field looking off into the distance at the sunset.
Aesthetic Score : 0.7
Mood : calm, contemplative, hopeful
Quality
Entropy : 6.70
Noise : 96
Prompt Clip Score : 0.17
AI Evaluation
Likelihood of AI : 0.10
Image errors : The image is slightly blurry, particularly on the man’s face. This may be due to the camera’s focus, or potentially a problem with the lens.
Joyful Kitchen Moments: A Woman’s Infectious Smile
Capture the warmth and happiness of a kitchen scene with this image. A woman in a striped apron beams with joy, clapping her hands in anticipation. The bright lighting and her infectious smile create a mood of excitement and delight.
Prompt
facial-expressions Gratitude: Hope, gratitude for the doctor’s care ; A kind-hearted chef reassures a nervous contestant in a bustling cooking competition, offering words of encouragement in a brightly lit kitchen filled with gleaming appliances.; cinematic
Characteristic
Shot : A woman in a kitchen, wearing an apron, is clapping her hands with a big smile on her face.
Aesthetic Score : 0.7
Mood : joyful, happy, excited
Quality
Entropy : 6.93
Noise : 102
Prompt Clip Score : 0.22
AI Evaluation
Likelihood of AI : 0.10
Image errors : No noticeable errors.
Friends Enjoying a Sunny Picnic in the Park
A group of friends share laughter and joy during a picturesque picnic in a sun-drenched park. The image captures a sense of warmth, camaraderie, and carefree happiness, with a balanced composition and beautiful natural lighting.
Prompt
facial-expressions Gratitude: Joy, gratitude for friendship and good times ; Group of friends laughing together at a picnic; eye-level; Normal People; sunny park with green grass and trees; cinematic
Characteristic
Shot : A group of friends enjoying a picnic in a park on a sunny day.
Aesthetic Score : 0.7
Mood : joyful, happy, relaxed
Quality
Entropy : 6.84
Noise : 102
Prompt Clip Score : 0.24
AI Evaluation
Likelihood of AI : 0.20
Image errors : The image suffers from some minor artifacts and a slightly uneven exposure.
Victory Dance! Woman Celebrates Triumph with Joyful Laughter
This image captures the pure joy of victory as a young woman raises her fist in the air, laughing with a trophy just out of frame. The scene radiates a celebratory and triumphant mood, showcasing the genuine excitement of her accomplishment.
Prompt
facial-expressions Gratitude: Pride, gratitude for recognition and hard work ; Gamer receiving an award for their achievements; close-up; Gamer; stage with a crowd and flashing lights; cinematic
Characteristic
Shot : A young woman is celebrating a victory, raising her fist in the air and smiling broadly. She is holding a trophy, which is partially obscured by the edge of the frame. There is a blue background with a bright light source, possibly a spotlight.
Aesthetic Score : 0.7
Mood : joyful, triumphant, celebratory
Quality
Entropy : 6.37
Noise : 106
Prompt Clip Score : 0.22
AI Evaluation
Likelihood of AI : 0.10
Image errors : No visible errors in this image.
Lost in the Pages: A Moment of Tranquility in the Library
A woman finds solace and peace amidst towering bookshelves, bathed in soft, warm light. Her serene expression speaks of a world lost in the pages of her book, creating a scene of calm contemplation.
Prompt
facial-expressions Gratitude: Peace, gratitude for knowledge and escape ; Woman reading a book in a quiet library; eye-level; Single Persons; peaceful library with bookshelves and natural light; cinematic
Characteristic
Shot : A young woman in a dark blue sweater is reading a book in a library, with a bookshelf behind her. The focus is on the woman and the book, the bookshelf is slightly out of focus.
Aesthetic Score : 0.7
Mood : calm, contemplative, focused
Quality
Entropy : 6.52
Noise : 103
Prompt Clip Score : 0.21
AI Evaluation
Likelihood of AI : 0.10
Image errors : The image has minor compression artifacts in the background, and some slight blurriness in the bookshelf.
One Man’s Trash, Another’s Inspiration: Volunteer Makes a Difference on the Beach
A volunteer, clad in a blue ‘Volunteer’ shirt, picks up trash on a sunny beach, his smile reflecting the hope and optimism of his actions. The scene, with two other volunteers in the background, highlights the importance of environmentalism and inspires viewers to make a difference.
Prompt
facial-expressions Gratitude: Satisfaction, gratitude for making a difference ; Volunteer helping to clean up a beach; wide shot; Heroes; beautiful beach with clear water and blue sky; cinematic
Characteristic
Shot : A man in a blue volunteer shirt is picking up trash on a beach, two other people in the background are picking up trash too, the beach is sandy with blue sky and some green hills behind the beach
Aesthetic Score : 0.6
Mood : happy, helpful, hopeful
Quality
Entropy : 6.54
Noise : 97
Prompt Clip Score : 0.23
AI Evaluation
Likelihood of AI : 0.20
Image errors : There is no artifact or errors in the image.
Conclusion
The results of the analysis 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 0.15, 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 0.42, which is also below average. This indicates that the model didn’t fully understand the scene described in the prompt and didn’t create an image that accurately reflects it.
- Aesthetic Analysis: The model scored 0.09, which is considered very good. This means that the generated image closely matched the expected aesthetic style, despite the issues with camera position and scene understanding.
Overall, the model seems to be better at capturing the desired aesthetic than understanding the scene and camera position. This suggests that the model might need further training to improve its ability to interpret and translate prompts into accurate visual representations.
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://docs.aws.amazon.com/bedrock/latest/userguide/titan-image-models.html