Prompting Guide
AI image generation is getting better by the day, however it still has its limits.
Character consistency is now easy on StoryboardHero, as you can create and save your characters and use them in your prompts.
On StoryboardHero, the AI will generate images based on the input provided in the prompt field. By default, the description field is copied to the prompt field for input. However, you can update the prompt field to define your scene as precisely as possible, thereby achieving satisfactory images and consistency.
Following these few tips, you’ll get better results when creating your images:
1. Be specific and concise
A shorter prompt will be easier and will be less likely to confuse the AI generator. Choose precise, definite and concrete words but do not multiply them. Adopt a synthetic style as follows:
There is an old man that runs while raindrops start to fall heavily as he raises his umbrella
An old man running under the rain.
2. Order your prompt
When writing your prompt, start with the most important thing of your shot, which is often what is in the foreground. You can follow this order :
– Subject : characters, animals, location
– Action : what the subject are doing and how
– Additional settings : lighting, colours, and framing.
For example:
With an umbrella, soft light, under the rain, an old man runs
An old man running under the rain, with an umbrella, soft light
3. Do not precise the style
You choose the style at the moment of generating the images. No need to add it in your prompt.
A black-and-white sketch illustration of an old man.
An old man.
Only if you have a specific need, not encompassed by our library of styles, you can choose the closest style in our library and may add specificities at the end of the prompt.
An old man, Van Gogh painting style.
4. Use our background prompt feature
You can separate the foreground and background prompt by using our background prompt feature. If some of your shots take place in the same location, create a background prompt and and repeat the prompt in each shot to achieve background consistency.
A dark, shady and narrow street with high residential buildings.
5. Reiterate defining features
Maintain consistency by reiterating defining features in each scene description. The AI generator does not keep memory of the previous image generated. If your character is crying in a shot, reiterate the information in the prompt of the following shot.
6. Use descriptive adjectives
As we’ve said, the prompt should feature a subject. You may add descriptors to this subject. Descriptors are most often adjectives, adverbs and verbs that answer questions like: what does the subject look like, what is happening, what is the subject doing, what’s happening around the subject?
Descriptive adjectives are your best chance at refining your images. They can also help confer a specific mood or a vibe to your images.
7. Use terms the AI understands
Use straight-forward and concrete terms. Avoid abstract and subjective nouns (love, hate, justice, infinity, joy) or adjectives (beautiful, free, fulfilled, open-minded). Abstract words often lead to surprising results. Instead, imagine the scene in your head and describe what is visible to the audience.
The AI generator also understands photographic terms, like: close-up, backlighting, mid shot, low angle,…
More generally:
– Use active rather than passive
– Use positive turn rather than negative
– Use parentheses to tweak the emphasis
8. Experiment
Depending on the image created by the AI, tweak your prompt to get what you really want. Emphasise elements that were overlooked by the AI and change your approach when the AI seems not to understand.