Stick figures obsolete: Qwen Image Edit, Flux multi-reference and AnyPose eat raw refs directly
cocktailpeanut · x · 2026-11-11
Developer cocktailpeanut explains how image control has evolved: Qwen Image Edit, Flux multi-reference, and AnyPose now process raw reference images directly through vision encoders and multimodal attention, with the text prompt defining each image's role — no intermediate stick figure is generated or injected as a control mechanism. The old stick-figure workflows have lost their reason to exist.
Related event: OpenPose Skeletons Fade as Reference-Based Image Editing Takes Over(3 posts)→
More from Multimodal
- ComfyUI Challenge Runner-Up Creativefield Open-Sources Free AI Photoshop Alternative — bfl_ai · 2026-11-11
- ComfyUI Challenge Winner ComfyArt: Open-Source Fully Local AI Design Canvas — bfl_ai · 2026-11-11
- Code-drawn portrait uses 7.29 million brushstrokes: the Fable 5.2 generative art piece — repligate · 2026-11-11
- The 'Boeing Bench': AI models race to build a Boeing 747, Fable 5.2 beats Gemini 4 Carbon — victormustar · 2026-11-11
- 300 random prompts on MiniMax H3 locally: ~23s per image on one RTX 5080 — JoNike · 2026-11-11
- Same prompt face-off: FABLE 5.2 vs Gemini 4 Argon draw an illuminated-manuscript magpie — teortaxesTex · 2026-11-11