A developer tested the Eyecandy visual techniques library to adapt stylistic effects for generative AI art and video pipelines in production.

Selected effects

The author identified 3 effects from the library that are suitable for integration into image2video and mixed scene workflows.

• Slit-scan — a temporal slicing technique used to create warped motion artifacts across frames and extended exposures in video sequences.
• Morphing — transformation of one object into another, applicable to faces, props, sketches and controlled object replacements in scenes.
• Long Exposure — streaking and blending of moving elements to convey continuous motion with minimal dataset requirements and strong stylistic result.

Technical considerations

For the Slit-scan effect the developer plans to reproduce the technique in Blender and AE to compare algorithmic outputs and model-driven results.

The Morphing use case assumes a clear start and end image, but the goal is to morph a single object inside an otherwise static scene without altering background elements.

The Long Exposure approach prioritizes style over heavy datasets, so experiments will focus on prompt engineering and motion-guided conditioning for video models.

Workflow and next steps

Planned experiments include pipeline integration, controlled datasets, and qualitative comparisons between handcrafted compositing and model-generated variants for each effect.

The author intends to report back with visual results and technical notes on Friday after implementing the initial set of tests and comparisons.