Generative AI — Image, Audio & Video Models

Course outline and learning expectations

This is a tutor-led course. The outline shows what your tutor will cover; teaching materials and examinations are provided directly to enrolled students.

What you will learn

• Distinguish generative models from predictive and discriminative models
• Explain diffusion model training, denoising and inference in clear technical terms
• Design and evaluate text-to-image, image-to-image, inpainting and controlled-generation workflows
• Adapt an image model using LoRA or another parameter-efficient method while managing data and compute constraints
• Build and assess introductory generative audio workflows for speech, music and sound
• Prototype generative video workflows and explain temporal consistency, cost, safety and quality limitations
• Apply copyright, licensing, consent, privacy, provenance, bias and disclosure practices relevant to Uganda and Africa
• Develop, test and present an AI-assisted marketing content application for a small Ugandan business

Modules

1. 1

   Generative media foundations

   Differentiate generative and predictive modelling and map multimodal workflows

   Generative versus predictive modelsDiscriminative and generative objectivesLatent representationsTransformers, GANs, VAEs and diffusion overviewModel selection and evaluation
2. 2

   Diffusion models for image generation

   Explain how text-to-image diffusion works from noise to image

   Forward and reverse diffusionDenoising networksLatent diffusionText conditioning and guidanceSchedulers and samplingQuality, diversity and common failure modes
3. 3

   Image generation and model adaptation

   Build and evaluate image workflows and perform lightweight fine-tuning

   Prompt designNegative prompts and generation parametersImage-to-imageInpainting and outpaintingControl methodsDataset preparationLoRA and parameter-efficient fine-tuningEvaluation and documentation
4. 4

   Generative audio systems

   Create and assess introductory music, speech and sound-generation workflows

   Audio representationsText-to-speech and voice synthesis conceptsMusic and sound generationVoice consent and impersonation risksAudio editing and quality checksAfrican-language support and limitations
5. 5

   Generative video systems

   Prototype short video workflows while accounting for technical limitations

   Text-to-video and image-to-videoShot and storyboard planningMotion and temporal consistencyLip-sync and voice integrationEditing and exportCompute, cost and current limitations
6. 6

   Responsible AI, copyright and data protection

   Apply ethical, legal and governance controls to generative media projects

   Uganda copyright and licensing basicsConsent and personal dataDeepfakes and disclosureBias and cultural representationDataset provenanceWatermarking and content credentialsPlatform safety policies
7. 7

   Africa-focused business applications

   Design practical creative workflows for Uganda, East Africa and African markets

   Small-business marketing contentTourism and hospitality promotionE-commerce product visualsEducation and public-information mediaDesign mockupsLocalization and accessibilityHuman review and approval
8. 8

   Capstone application

   Build and present an AI-assisted marketing content generator for a small Ugandan business

   Problem definitionUser and brand requirementsImage and caption workflowResponsible-use checklistPrototype implementationTesting with realistic promptsEvaluation and iterationDemo and documentation