Full social media campaign for Myanmar — 4 Facebook post series, 2 infographics, a comic post, and 2 animated campaign videos. Produced in one month using Midjourney, NanoBanana, Seedance AI, and After Effects.
Geneva Call is an international humanitarian organization that engages armed actors to respect international law. The Myanmar campaign needed to communicate the Principles of Humanity (impartiality, neutrality, independence) across social media platforms. I received PDF reference materials from the client and produced 4 Facebook post series, 2 infographics, a comic post, and 2 campaign videos within one month.
Recreating an established illustration style using AI tools requires knowing which outputs to reject and why, and knowing when to stop generating and start editing manually. Campaign scale, cultural sensitivity, and a one-month deadline all applied pressure at the same time.
Four post series covering different Principles of Humanity, 5–6 slides each. All Burmese text was set in Figma (Adobe tools do not render Myanmar script correctly — more on this in Section 04), exported as images, and placed into the final artwork. The client reviewed every slide for linguistic accuracy before final export.
Infographic 1 — Circular layout, equal weight for each principle
Infographic 2
Comic-format post — lighter visual register
NanoBanana (via Artlist.io) generated all start and end frames. The workflow ran in two steps per scene: generate a start frame, generate an end frame, then feed both into Seedance AI to animate the transition. This two-frame method produced more stable animations than prompting Seedance directly. Midjourney handled the static editorial illustrations for the social posts.
Over the course of this campaign, roughly 40% of generated videos were rejected. The criteria were specific: consistent object geometry frame to frame, neutral camera movement, appropriate emotional register for a humanitarian context. Good generation workflow is mostly elimination.
Too brutal — wrong emotional register
Stable, appropriate tone — soldier interaction
Objects disappear mid-clip, excessive zoom
Stable geometry, controlled camera, correct weight
A non-obvious technical issue that cost significant time — documented here because no documentation warned about it.
The problem: Adobe tools (Illustrator and After Effects) do not render Myanmar script correctly. Characters appear in the right order on screen but export with symbols rearranged or substituted. The tools were not confused about the font file but about the Unicode rendering order for a script that combines characters in ways Latin-based tools are not built for.
The fix: Set all Burmese text in Figma, which handles Myanmar script rendering correctly, then export each text block as an image. Those image files were imported into After Effects and Illustrator, placed as visual elements rather than live type. It added a step to every asset but made the output reliable.
A second issue: In Burmese, words are not separated by spaces. A space character functions like punctuation, closer to a comma or a full stop. This required a separate review loop with the client to verify that every line break and spacing decision read correctly to a native speaker. The client checked every slide of every series before final export.
Two finished campaign videos were delivered: one on humanitarian worker access, one on education protection. 20+ Seedance AI clips were produced, approved, color-corrected, and assembled in After Effects — adapted for social media formats and delivered with subtitles.
Video 1 — Let Humanitarian Workers
Video 2 — Protect Education
The campaign ran on Geneva Call's social media channels for the Myanmar audience. AI-assisted production cut per-asset time by approximately 4x. The Figma-as-text-renderer workaround resolved the Myanmar script issue and became the standard method for all subsequent Burmese-language assets in the project.