Computational Photography and AI Imaging
How phones and software make images: the image signal processor, burst merging, HDR and tone mapping, computational depth and bokeh, night modes, super-resolution and zoom, AI denoising and upscaling, how diffusion models generate images, professional AI workflows and disclosure, and content credentials, detection and Canadian law on deepfakes and AI authorship, ending in a pipeline and provenance policy capstone.
PHOT-420 Year 4 3 credits 12 lessons · 6 h
Offered by HSM University · Photography
This is the Canadian version of this course. See the American version.
What you'll learn
- Explain why phone cameras depend on computation, describe the stages of a computational photography pipeline and classify operations as measurement, interpretation or invention.
- Model sensor noise, apply demosaicing, white balance and colour correction algorithms, and trace common artefacts to pipeline stages.
- Calculate the noise benefit of burst merging and explain coarse-to-fine alignment and robust merging that prevents ghosting.
- Measure dynamic range in stops, merge bracketed exposures into radiance maps and compare global, local and learned tone mapping, including HDR gain maps.
- Relate optical depth of field to focal length and aperture, compute depth from disparity and evaluate synthetic bokeh and its artefacts.
- Plan low-light captures by balancing motion, read noise and frame count, and apply calibration frames and stacking blend modes.
- Explain sampling limits, multi-frame super-resolution, crop zoom and multi-camera fusion, and test resolution claims critically.
- Compare classical and learned denoising, sharpening and upscaling, and interpret PSNR, SSIM and the perception-distortion trade-off.
- Describe how GANs and diffusion models generate images, how prompts and guidance steer them, and how inpainting, fine-tuning and training data affect results.
- Apply an AI intervention ladder to professional workflows, contracts, disclosure, editorial, contest and stock rules, privacy and quality control.
- Explain detection limits and C2PA content credentials, and summarize Canadian and US law on synthetic images and AI authorship.
- Design and test a computational photography pipeline and write an AI-use and provenance policy for a photography practice.
Course content
12 lessons · 6 h total · final exam
- 1 From optics to computation 31 min · quiz
- 2 The image signal processor: demosaicing, white balance and noise 32 min · quiz
- 3 Burst photography and multi-frame alignment 28 min · quiz
- 4 HDR imaging and tone mapping 29 min · quiz
- 5 Computational depth, bokeh and portrait modes 30 min · quiz
- 6 Night modes and low-light stacking 30 min · quiz
- 7 Super-resolution and computational zoom 30 min · quiz
- 8 AI denoising, sharpening and upscaling 29 min · quiz
- 9 Generative AI fundamentals for imaging 30 min · quiz
- 10 Generative editing workflows and professional practice 29 min · quiz
- 11 Authenticity, provenance and content credentials 31 min · quiz
- 12 Capstone: a computational photography pipeline and provenance policy 31 min · quiz
- E Final exam 20 questions
Requirements
Recommended background. You can start any course at any time.
Description
How phones and software make images: the image signal processor, burst merging, HDR and tone mapping, computational depth and bokeh, night modes, super-resolution and zoom, AI denoising and upscaling, how diffusion models generate images, professional AI workflows and disclosure, and content credentials, detection and Canadian law on deepfakes and AI authorship, ending in a pipeline and provenance policy capstone.
Part of the Film & Photography stream, typically taken in year 4.