Higgsfield Image to Video: A Still-to-Clip Workflow

Higgsfield turns a still image into a short clip by applying a camera preset and a prompt to a start frame. The generation itself takes a minute. What separates people who get usable clips from people who burn credits is everything around that minute: how the still was prepared, how the preset was chosen, how the prompt was written and how the results were reviewed and filed. This guide lays out a workflow you can repeat on every job, whether the output is a product clip, a fashion loop or a moving mood board.

The workflow in one view

1. Prepare the start frame. Sharp, correctly cropped, subject centred, margin around it.
2. Choose the camera preset. One move, matched to what the image can support.
3. Write the motion prompt. One sentence about the scene and subject; nothing about the camera.
4. Generate and review. Two or three takes, shortest duration, judged frame by frame.
5. Organise the takes. Name, file and record the settings of everything you keep.

Each step is small. Skipping any of them is what produces the familiar pile of near-misses.

Step 1: prepare the start frame

The start frame sets the composition, palette and subject for the entire clip, so fix it before you generate rather than after.

• Crop to the final aspect ratio. Decide whether the clip is vertical, square or widescreen, then crop the still to match. Letting the model reframe gives it an extra job.
• Centre the subject and leave margin. Push-in presets crop the edges; orbits need space to move around. A subject touching the frame edge will be lost.
• Remove text and logos. Lettering smears as soon as the camera moves. Add it back in your editor.
• Check sharpness and resolution. Soft or low-resolution stills produce soft, wobbly video. Upscale first if necessary.
• Prefer simple backgrounds for big moves. The more a move reveals, the more the model invents. A plain backdrop keeps that invention believable.

If you are generating the still inside Higgsfield with Soul, choose a preset and framing with the video move already in mind. The Soul presets guide covers that side.

Step 2 and 3: choose the preset, write the prompt

Pick one camera preset for each take. Dolly In and Dolly Out are the safest because they show little beyond the still; 360 Orbit, Bullet Time and FPV Drone are more dramatic and more likely to expose guesses about parts of the scene the image does not contain. Match the move to the job of the shot in your edit, not to what looks most exciting in isolation. The camera presets guide goes deeper on choosing and mixing.

Then write the prompt. Because the preset owns the camera, the prompt only has to say what happens in the scene:

• One sentence, one subject action. "She turns her head towards the light and smiles slightly."
• Add secondary motion. Fabric, hair, steam, dust, water. "Her scarf lifts in a light breeze."
• State what stays still. "Background and lighting remain constant."
• No camera words. Do not write "zoom", "pan" or "orbit"; the preset handles it and conflicting text produces confused motion.
• Pace words help. "slowly", "gently" and "subtle" reduce warping.

Settings at a glance

Generation is charged in credits that vary with model, duration and resolution. Check the platform’s current pricing page before a large batch.

Step 4: generate and review

Run two or three takes of the same setup. Presets are consistent but not identical between runs, and the cost of a second take is small compared with the cost of settling for a flawed first one. Review each clip in this order:

1. Play at full size once for overall impression. Does the move feel right for the shot’s purpose?
2. Scrub the first and last second. Artefacts cluster at the start, where the still becomes motion, and at the end, where the model runs out of information.
3. Check the subject’s edges and face. Warping shows there first.
4. Check what was revealed. For orbits and drone moves, look at the newly invented areas and decide whether they are acceptable.
5. Decide: keep, retry with one change, or change the preset. Change only one thing per retry so you learn what caused the difference.

Step 5: organise the takes

Clips without their settings are nearly worthless a week later, when a client asks for "the same move on the new colourway". Build the habit now:

• Name files consistently. Project, subject, preset, take number: summer-lookbook_coat-cream_dolly-in_t2.mp4.
• Keep the start frame with the clip. Same name, image extension. You will need it to regenerate or to try a different preset.
• Record the prompt, preset, model and duration in a text file or spreadsheet row beside each keeper. Copy the prompt exactly as submitted.
• Separate keepers from rejects at the moment of review. A folder of everything is a folder you will never open.
• Note what you learnt. "Orbit invents a wrong back pocket on this bag; use arc or dolly instead." One line saves a future credit spend.

For client work, this record also answers the questions that come up at delivery: which tool and model made the clip, what the prompt was and when it was generated.

Try it

Start image: a woman in a long camel coat standing on an empty pedestrian bridge, centred, vertical crop, no text.
Preset: Dolly In.
Prompt: She turns her head slowly towards the river and her coat moves gently in the wind. Background and lighting remain constant.
Duration: shortest available. Takes: 3.
File as: bridge-test_camel-coat_dolly-in_t1..t3

Once one take is clean, keep everything the same and swap the preset to a slow orbit or arc. Reviewing the pair side by side is the quickest way to learn how much invention your particular image can carry.