Test: ChatGPT's New Image Model Shows Strong Character Consistency
omooretweets · x · 2026-08-20
Comparing experiments from last year, there are now many more video models capable of syncing audio and facial movements. ChatGPT's new image model performs well in both generation and editing. Notably, character consistency saw a huge step up, allowing the same avatar to remain consistent across different outfits and environments.
Related event: $200 AI Influencer Gains 1M Views in One Week(4 posts)→
More from Multimodal
- KRAFTON releases Raon-OpenTTS-1B, an open zero-shot voice cloning model — victormustar · 2026-08-20
- GPT-Image2 Workflow: Dynamic Brand Posters for Drinks and Pastries — GCWebDesigner · 2026-08-20
- GPT Image 2 Tip: Transform Logos into Living Botanical Installations — GCWebDesigner · 2026-08-20
- Midjourney v8.2 Creation: Tree of Life — chrisfirst · 2026-08-20
- Testing Claude's ability to control CustomUse, Blender, and Three.js — majidmanzarpour · 2026-08-20
- Fairground AI Creator TV launches on Amazon Prime Video and Xumo Play — DavidmComfort · 2026-08-20