When Prompts Can't Get the Shot: AI Video Creators Struggle With Camera Control
Fearless-Brain8928 · reddit · 2026-09-25
A follow-up Reddit post details the core pain of AI video generation: when you need exact camera moves and character blocking, prompting falls apart — drift, swapped characters, near-miss framing, and burned credits. Blender blocking fixes staging but not faces. The author is building a tool to gain shot control without learning Blender and is surveying others on their workarounds.
More from Multimodal
- Midjourney ships new --tile and inpainting, previews real-time models — midjourney · 2026-09-25
- AI art piece 'One Stone a Visit' made with Fable, Grok and Opus — techartist_ · 2026-09-25
- Fill It Black: Open-Source LoRA Brings One-Step Inpainting to Krea-2 — cierpliwy · 2026-09-25
- Chaining MiniMax Video Clips in ComfyUI: Easy to Append, Hard to Insert — HRH_Duke_Morbid · 2026-09-25
- New open-source ComfyUI extension runs any OpenRouter model via 13 nodes — TheMoonMidas · 2026-09-25
- YuE2 music model with full CoT runs on iPhone in just 1.7GB of memory — Acceptable-Cycle4645 · 2026-09-25