Dancing Barry
Sector
AI Creative
Type
Concept
Role
Creative Director
Skills
Concept Development AI Video & Audio Editing
Objective
To test whether a fully AI generated non human character could maintain a coherent identity across multiple images, scenarios and motion clips using Flux Kontext and a multi tool generative workflow.
Dancing Barry was a playful but technical investigation into character identity within generative workflows. The project tested how far a single AI defined character could be pushed across variations in pose, style and context while still remaining unmistakably himself.
Dancing Barry was a playful but technical investigation into character identity within generative workflows. The project tested how far a single AI defined character could be pushed across variations in pose, style and context while still remaining unmistakably himself.
Character foundations.
The project began by designing a wholly unique caveman character in Midjourney. The aim was to create a figure with enough visual specificity to withstand transformation. Texture, bone structure, hair details and clothing motifs were intentionally crafted to give the character a recognisable signature that tools could repeatedly interpret.
With this foundation established, the character was introduced into Flux Kontext to test its ability to retain identity under variation. I generated multiple poses, expressions and environmental scenarios, analysing how the system dealt with perspective shifts, wardrobe changes and emotional nuance. This phase centred on understanding the limits of character coherence and how context influences the perception of continuity.
Character foundations.
The project began by designing a wholly unique caveman character in Midjourney. The aim was to create a figure with enough visual specificity to withstand transformation. Texture, bone structure, hair details and clothing motifs were intentionally crafted to give the character a recognisable signature that tools could repeatedly interpret.
With this foundation established, the character was introduced into Flux Kontext to test its ability to retain identity under variation. I generated multiple poses, expressions and environmental scenarios, analysing how the system dealt with perspective shifts, wardrobe changes and emotional nuance. This phase centred on understanding the limits of character coherence and how context influences the perception of continuity.
Motion and expression.
Once the still imagery demonstrated reliable consistency, the workflow expanded into motion. Images were brought into Kling 2.1 to generate video clips that preserved both the physical traits and personality of the character. The focus was on retaining recognisable shape language and behavioural cues even as movement increased visual complexity.
An original audio track was then created in Udio to complement the visual tone and enhance the expressive quality of the character. Editing the full sequence in Davinci Resolve allowed me to refine pacing, synchronise audio and unify the visual aesthetic so the final piece presented a coherent, living version of the character across every medium.
Motion and expression.
Once the still imagery demonstrated reliable consistency, the workflow expanded into motion. Images were brought into Kling 2.1 to generate video clips that preserved both the physical traits and personality of the character. The focus was on retaining recognisable shape language and behavioural cues even as movement increased visual complexity.
An original audio track was then created in Udio to complement the visual tone and enhance the expressive quality of the character. Editing the full sequence in Davinci Resolve allowed me to refine pacing, synchronise audio and unify the visual aesthetic so the final piece presented a coherent, living version of the character across every medium.
