I created my first Dark super hero AI short film : OBSCURO — EP.1 : She Says Hello
Hey Reddit,
I spent the last few weeks pushing Kling AI to its limits — complex characters, dialogue, lip sync, action sequences. The question was simple: can you make a genuinely cinematic superhero short film with consistent characters across every scene?
The answer is yes. And it looks like this: https://www.youtube.com/watch?v=O8kZgI0JO7w
━━━━━━━━━━━━━━━━━━━━━━━━━━━
THE CONCEPT
OBSCURO — EP.1 : She Says Hello
A person goes missing. A hero is hired to help. A villain involved — why?
1min30 of dark cinematic action. Dialogue. Confrontation. Energy weapons. Dissolution. And a final shot that reframes everything.
━━━━━━━━━━━━━━━━━━━━━━━━━━━
THE WORKFLOW
→ Character sheets: Midjourney V7 & 8 + GPT Image 2 for just a couple of shots → Start frames: Midjourney → Video generation: Kling Video 3.0 for almost all shots (dialogue/lip sync) + 2.6 (only for one shot) → Character binding: Kling Elements system → Music: Suno AI → Final edit: Filmora
The key discovery: Kling's Element binding system is the real unlock for consistent characters across an entire film. Same face, same costume, same villain — every single shot. That's the hardest problem in AI filmmaking and Kling solves it better than anything else right now.
━━━━━━━━━━━━━━━━━━━━━━━━━━━
WHAT WORKED
Kling 3.0 lip sync is genuinely usable for cinematic work. Not perfect — but good enough to build dramatic scenes around.
"Camera slowly pull in" on a static character is the single most powerful prompt instruction I found. Pure tension, zero cost.
Custom Multi-Shot for scenes with different characters per shot. Standard Multi-Shot for static dialogue. Know the difference — it saves credits and time.
━━━━━━━━━━━━━━━━━━━━━━━━━━━
WHAT DIDN'T WORK
Complex physical movement — diving, running, fighting. Kling still struggles with precise human body action. Workaround: cut around it, use reaction shots.
Generating this character in detention contexts — both MJ and GPT Image block it hard. Solution: describe without captivity vocabulary.
━━━━━━━━━━━━━━━━━━━━━━━━━━━
Happy to answer questions about prompts, Element binding, Multi-Shot workflow, or anything else.
Drop them below 👇
— ObscuroFilms
━━━━━━━━━━━━━━━━━━━━━━━━━━━
SPECS: Resolution: 1920x1080 — 24FPS Images: Midjourney V7 & 8 + GPT Image 2 ; Video: Kling Video 3.0 / 2.6 Music: Suno AI Edit: Filmora