What’s new
- Built around camera moves, not general-purpose video: DoP turns a still image into a shot using one of 50+ camera-motion presets, ranging from basic dolly, pan, tilt and zoom moves to more elaborate ones like bullet time and robo-arm shots.
- The preset sets the camera logic, not your wording: instead of describing the camera move in the prompt, you pick a preset — Higgsfield says preset selection directly determines the camera logic — and use the prompt for scene detail instead.
- Trained with reinforcement learning on top of diffusion: Higgsfield says it trained the model the way reasoning LLMs are trained, teaching it camera movement, lighting, lensing and scene structure specifically, so moves read more like something a real cinematographer shot.
- Starts from a single image: no footage or shot list needed — one sharp photo plus a preset is a complete input.
What it’s good for
- Turning a product photo or portrait into a short clip with real camera movement, for e-commerce or portrait-style content.
- Giving a concept image or poster a trailer-like feel through camera motion, without any physical camera rig.
- Adding movement to existing static art — illustrations, renders, photos — without reshooting or building a 3D scene.
Prompting tips
- Start from a sharp, well-composed image: Higgsfield’s own guidance says image quality drives output quality — a soft or poorly framed photo tends to produce a soft, blurry clip.
- Pick your preset by the type of move you want: presets are grouped into categories — Effects (VFX-driven), Basic/Epic Camera Control (simple to advanced camera moves), Catch the Pulse (a curated themed set), and Mix (combined presets) — choose the category first, then the specific preset.
- Let the preset handle the camera, let your words handle the scene: since the preset already defines the motion, use the prompt to describe what’s happening and the mood, not the camera move itself.
- Turn on Enhance if the shot barely moves or ignores your details: that’s Higgsfield’s own suggestion for when the output doesn’t reflect what you described.
- Pick 3 or 5 seconds: those are the only two duration options currently offered.
Upload a sharp wide shot of a harbor at dusk. Select the "Dolly In" preset under Camera Control.
Prompt: the camera slowly pushes toward a fishing boat at the center of frame, the sea lit orange by the sunset, a couple of gulls crossing over the bow, quiet and documentary-like.
Set duration to 5 seconds. If the move looks too subtle, turn on Enhance and regenerate.
Known limits
- Image-to-video only — there’s no way to generate from text alone; you always need a starting keyframe image.
- Only two duration options per generation, 3 or 5 seconds, with nothing longer currently available.
- Output quality depends heavily on the input image: Higgsfield’s own help docs say a soft or poorly composed source image tends to produce a soft, blurry result.
Using it in Nomi
On the generation canvas, add a video card, pick “Higgsfield DoP” in the model picker and choose a mode (First frame to video, First and last frame). Add your references, write the prompt and generate. In the storyboard you can pick it per shot. The first time, connect any one of the providers above under Settings → Models.
Sources
- How do I use DoP (Higgsfield official Help Center)
- Which Higgsfield AI model should you use (Higgsfield official Help Center)
- Higgsfield Camera Controls preset page (Higgsfield official)
- Introducing Higgsfield DoP I2V-01-preview (Higgsfield official blog, 2025-03-31)
- About Higgsfield (Higgsfield official site)
