Skip to content
DreaminaHub

Writing Dreamina prompts that land first time

Every failed generation costs credits. The formula below is ordered by how much weight Dreamina gives each part — get the first two elements right and most prompts work.

Last reviewed

Short answer

What is the prompt formula?

Subject → style → camera → lighting → mood → detail, in that order, separated by commas. The subject is what is physically in frame and must come first, because terms earlier in a prompt carry more weight. Everything after it modifies. Add motion terms only for video, and keep negative prompts for artefacts you have actually seen rather than ones you are guessing at.

Anatomy

The six parts, and what each one controls

Each slot answers a different question. Leaving one out is fine; putting them in the wrong order is not.

PartAnswersExampleSkip it when
SubjectWhat is in the frame?a lone hiker on a ridge at dawnNever — this is the prompt
StyleWhat kind of picture is it?cinematic film stillYou want the model default
CameraHow is it framed?wide 24mm establishing shotFraming genuinely does not matter
LightingWhere does the light come from?golden hour backlightThe style already implies it
MoodHow should it feel?calm and minimalThe subject carries the mood
DetailWhat finish?photorealistic, natural skin textureWorking in an illustrated style

Two before five

If you only change two things about how you prompt: put the subject first, and describe the light. Those two account for most of the gap between a render that works and one that nearly works.

Tool

Build one now

Pick from each slot and the prompt assembles in the correct order. Copy it straight into Dreamina.

Prompt builder

Subject first, then modifiers in the order Dreamina weights them most heavily.

Negative prompt

Your prompt

a woman in her thirties laughing, freckles, curly hair, cinematic film still, shot on 85mm lens, shallow depth of field, golden hour backlight, photorealistic, natural skin texture

Negative: blurry, low resolution, extra fingers, deformed hands

Worked examples

Prompt library

Five prompts that follow the formula. Swap the subject and keep the rest.

Product on a surface

a matte black coffee flask on wet slate, studio product photograph, shot on 85mm lens, shallow depth of field, soft diffused window light, clean minimal background, highly detailed

Natural portrait

a woman in her thirties laughing, freckles, curly hair, editorial photograph, shot on 85mm lens, golden hour backlight, photorealistic, natural skin texture

Atmospheric street scene

a night market street with steam rising from food stalls, cinematic film still, wide 24mm establishing shot, neon rim lighting, cyan and magenta, moody and cinematic, soft film grain

Image-to-video motion

the subject turns slowly towards camera and smiles, slow dolly in, handheld feel, natural daylight, locked exposure

Flat marketing graphic

a bold sale poster for a running shoe brand, flat vector illustration, high key studio lighting, bright and energetic, clean minimal background, generous empty space on the left

Iteration

Changing one thing at a time

A generated image alongside the prompt that produced it
A render and its prompt. When output is close but wrong, resist rewriting the whole prompt — change one clause and regenerate.

The expensive mistake is rewriting an entire prompt when one element is off. You lose the parts that were working and cannot tell which change helped. Alter a single clause, run it again on the cheapest model, and only move to a high-fidelity model once the composition is right. A prompt that took six cheap iterations and one expensive final render costs far less than three expensive guesses.

FAQ

Prompting questions

Does prompt order actually matter in Dreamina?
Yes. Terms earlier in the prompt carry more weight. Putting the subject first and modifiers after is the single change that most improves adherence — a prompt that opens with "cinematic, 8k, masterpiece" and mentions the subject at the end will drift.
How long should a prompt be?
Long enough to be specific, short enough that every term earns its place. Roughly fifteen to forty words is the productive range. Past that, later terms get diluted and you lose control rather than gaining it.
Do quality words like "8k, masterpiece, award winning" help?
Very little on current models, and they cost you prompt weight that a concrete description would use better. "Shot on 85mm, shallow depth of field" changes the image; "masterpiece" mostly does not.
What should go in the negative prompt?
Only artefacts you are actually seeing. A negative prompt is a corrective tool, not a checklist — loading it pre-emptively with twenty terms constrains the model in ways you did not intend. Start empty, add terms as problems appear.
Why does the same prompt give a different result every time?
Each generation uses a different random seed. If you need the same character or scene across several renders, supply a reference image instead of relying on the prompt text to reproduce it.
Should I write prompts in my own language?
You can, and it usually works. English prompts consistently track closer to intent though, because that is what the models saw most of during training. If a prompt is not landing, translating it to English is worth trying before rewriting it.

Next

Keep reading