AI and Creative Media

• The presentation is about AI for Creative Industries, specifically Module 2: AI, ML, and Data Science

Goal of Section:

• Introduce multimedia and creative content
• Introduce current AI use for creative industry
• Apply generative AI to creating media
• Introduce current issues with generative AI

What is Multimedia?

• Uses a combination of different forms of media
• Purpose is to convey information or provide entertainment

What is Media?

• Media is communication to deliver information, entertainment, and other forms of content to a wide audience

Components of Multimedia

• Text
• Audio
• Images
• Video
• Interactive Media

Current AI Use in Creative Industry

• Machine Learning focuses on supervised learning, unsupervised learning, deep learning, generative models, and multimodal learning
• Supervised learning involves a model trained on a labeled dataset to accurately predict labels for new data
• Unsupervised learning trains models on unlabeled data to identify underlying structures or patterns
• Deep Learning applies neural networks to model complex patterns in large datasets
• Generative Models generate new, similar data points
• Multimodal Learning integrates and processes information from multiple data types simultaneously

Rendering With AI

• Upscaling uses AI to increase the resolution of images
• AI can be used in 3D rendering and video rendering

AI for 3D Animation

• Speech Animation synchronizes lip movements and facial expressions with speech
• This is used in character animation and virtual avatars
• Automated facial animation can derive from audio
• Speech is analyzed and translated into virtual muscle activations within vocal tract simulation

Image Processing with AI

• AI can enhance images or videos

Generative AI

• It applies generative models to create new data points similar to existing data
• Generative AI can create stories, images, audio, video, and 3D models
• With generative AI, artists can create artwork, and use AI as a software tool
• Artists can create prompts to ask AI to generate content, while editing techniques enhance the result
• AI tools assist in artwork creation
• AI can generate content with image editing
• AI can be used in video editing using AI-generated video, audio, and voice over

Technical Side: How Generative AI Works

• AI fixes original image by adding noise, then removing it to repair the image
• Image generation applies a diffusion model for this

Image Generation with Diffusion Model

Typical steps include

• Noise is added to Image 1 to create Image 2
• Input A prompts a desired image (e.g., cat), with image 2 as a starting point
• The diffusion model works to generate output B, (in this case, a cat)

Prompt Engineering

• Is the practice of designing inputs for AI tools to produce optimal outputs
• First step is to have a clear image of the target picture
• Good Prompt Structure:
• Type of expected image (Realistic, Semi-Realistic, Illustrator, etc.)
• Subject, action, object
• Light and camera setting

Prompts

• Using Subject, action, and object e.g. 1 GIRL,WORKING LOOK, BLACK SUIT, FORMAL SUIT, HOLDING A RABBIT DOLL SIMPLE BACKGROUND, WHITE BACKGROUND
• Additional light and camera settings e.g. garden, morning light, rim light, view from above
• Additional details e.g., 1 GIRL, 25 YO, LONG HAIR, BROWN HAIR,BROWN EYE,WORKING LOOK, BLACK SUIT, FORMAL SUIT, SMILE, HOLDING A RABBIT DOLL garden, flowers, morning light, rim light, view from above, look at viewer, Olympus em5,25 mm f 1.4,bokeh,

AI and Image Generation

• To get started with Dreamina, register for an account, and agree to the Terms of Service/Privacy Policy
• Dreamina can create 450 free credits per day
• Generate images, videos, or music by adding text or an image
• Quality and aspect ration can be adjusted to improve image

Practices

• Practice 1 - generate image of favorite pet inside dreamed home, follow good prompt structure, submit to learning
• Practice 2 - Generate image for product advertisement; must select product to sell, presenter and have presenter look like youself, presenter must hold product, submit to learning Practice 3 AI Image Inpainting – select one out of three provided image on ICT eLearning, use AI inpainting to decorate the selected image in Valentine Theme, submit to eLearning

Flaws with AI-Generated Images

• Text presentation (AI often jumbles letters and wording)
• With AI generated art it has difficultly with the amount of fingers, legs, or tails

AI Inpainting

• Generative AI can detect objects inside an image, and can be used to alter parts of the image
• Select the source image and download, got to Canvas in Dreamina, upload selected image onto canvas
• Enter image editing mode by selecting the image and then the inpainting tool
• User is able to select the are with a "brush" tool, and then enter a prompt to create an editted version of the image

Generative AI for Audio

• Generative AI can be used in a variety of ways for audio
• Can use text prompt to generate lyric
• Text prompt to generate melody
• Select from provided singer sound
• Has a limited length and style if free version

Generative AI for video

• Has Text to video option that will generate a video with a prompt
• Limitations: limited length and movements

Adobe Creative Cloud

• MU Software download, free for Mahidol students and staff Steps for Software download
• Be on MU Wi-fi (Not MUIC Wifi)
• Go to https://muit.mahidol.ac.th/
• Software download &Manual > Software Download
• Log in with Mahidol account
• Can then download the software to personal Machine

Copyright

• Is a legal concept that grants creator of the original work with exclusive rights to use and distribute it
• Protect creators' intellectual property and provide control over how their work is used, grant for how their work is shared
• Only applies for a tangible medium of expression
• Does not protect facts, ideas, systems, or methods of operation
• Covers literary, dramatic, artistic works such as art, movies, music, computer software and architecture
• Governs making copies, distribution, preparing derivatives based on originals, performing etc etc

Public Domain (PD)

• Creative works not protected by copyright
• Free for anyone to use, copy, modify, distribute
• Types of PD
  • Expiration of copyright
  • Works created by the govenrment
  • Voluntary release
  • No copyright protection

Fair Use

• A legal doctrine that allows limited use of copyrighted material without getting permission of rights holders
• Particular use qualifies if
  • adds new expression, meaning to the origin material
  • parody
• a non-commercial use for educational, research, or for personal use

Terms of Use

• Legal agreements between the service provider and the users and shows the responsibilites relating
• Often cover usage rights, restrictions, intellectual property rights etc

Copyright and Generative AI

• LAION is considered to be an open AI for copyright
• Ongoing debates about copyright and AI with no definitive answer
• Material created by AI tools do not receive copyright protections in the US
• Generated artworks infringe copyright
• The US Copyright office has provided guidance saying that while AI assisted inventions aren't categorically unpatentable, the inventorship analysis should focus on human contributions

Test your knowledge of AI in creative media! Explore supervised/unsupervised learning, multimedia, and generative models. Understand AI's role in creative tasks.