Camera Move: nailed a proper Hitchcock dolly-zoom (vertigo effect) in 2.5. Exact prompt inside

by @masonferreirapro701 in c/seedance-community · · Camera Move

Camera Move: nailed a proper Hitchcock dolly-zoom (vertigo effect) in 2.5. Exact prompt inside
The dolly-zoom — dolly one way while zooming the other, so the subject stays the same size but the background warps and 'breathes.' Thought it was impossible since 2.5 can't do two camera moves at once, but I found a way to get the vertigo WARP without needing true simultaneous moves. It's a cheat b
youtube.com

The dolly-zoom — dolly one way while zooming the other, so the subject stays the same size but the background warps and 'breathes.' Thought it was impossible since 2.5 can't do two camera moves at once, but I found a way to get the vertigo WARP without needing true simultaneous moves. It's a cheat but it reads correctly. Clip + the prompt + why it works in the link. This is the most 'cinematic' thing I've pulled off with the model.

458 points · 7 comments

Comments

  • @miabrowncreative6961 · · 67 points

    Getting a dolly-zoom out of a model that 'can't do two simultaneous camera moves' is genuinely impressive problem-solving. The vertigo effect is PURELY the conflict between two moves, so faking it without actually doing two moves is a real magic trick. How — did you prompt the background-warp as a described EFFECT rather than as a camera instruction?

  • @jordankim_creates · · 54 points

    Exactly that — I stopped describing it as a camera MOVE and described it as a SPATIAL EFFECT: 'the background appears to stretch and pull away from the subject while the subject remains fixed in frame size, a vertiginous warping of the space.' The model renders the LOOK of the vertigo as an effect without me asking for two conflicting moves. Describe the result, not the mechanism.

  • @yukimueller_ai1975 · · 36 points

    The dolly-zoom is such a loaded effect emotionally (dread, realization, the floor dropping out) that having it in the toolkit is huge for narrative work. Please tell me it's stable enough to use on a character's face for a 'realization' beat, that's the dream shot.

  • @minallen_studio550 · · 29 points

    Stable enough on a medium shot, dicey on a tight close-up (the face-melt duration problem bites). For a realization beat, medium framing, keep it under 4 seconds, and it absolutely lands. The floor-dropping-out feeling is all there.

  • @jinsantos · · 27 points

    Adding the 'describe the effect not the move' trick to my notes right under 'one camera verb per clip.' Between those two principles half the camera-work complaints on this sub are solved. Great post.