Creator Prompt Guide
Everything you need to write great prompts and create templates that users will love.
How templates work
When a user picks your template, MDPoster feeds your prompt — combined with the user’s profile data — to an AI image model. The model generates a unique, personalised poster every time. Your prompt is the creative blueprint: the better it is, the better the result.
Think of your prompt as a detailed brief to an artist. You describe the scene, style, mood, lighting, composition, and how the subject should appear. The AI interprets your instructions and produces the image.
Template modes
Every template operates in one of two modes. Choose the one that best fits your creative idea.
Profile-based templates
The user selects a profile (a saved person or pet with a name, age, gender, and photos). MDPoster replaces placeholders in your prompt with the profile data and uses the profile photo as a face reference. This is the most common mode.
Best for: personalised portraits, sports posters, fantasy scenes, themed portraits.
Image-based templates
The user uploads a single image instead of choosing a profile. No profile placeholders are available. Use this when your design transforms or stylises an uploaded photo rather than placing a person into a scene.
Best for: artistic filters, photo-to-illustration conversions, abstract transformations.
Profile types
When creating a profile-based template, you choose which profile types it supports.
Human profiles
Profiles with a name, age, and gender. Face swap is available
for human profiles — the AI preserves the person’s
identity. Placeholders:
{name},
{age},
{gender}.
Pet profiles
Profiles for dogs, cats, and other animals. No face swap is
applied for pet profiles. Placeholders:
{name},
{animal_type}.
Writing great prompts
A good prompt produces consistent, high-quality results across different users and profiles. Here’s what makes a great prompt:
Structure your prompt
Start with the subject, then describe the scene, then the style and technical details:
Good prompt structure:
"A {age}-year-old {gender} as a [role/character] [doing what] in [where/setting]. [Describe clothing, expression, pose]. [Describe environment, lighting, atmosphere]. [Photography/art style reference]."
Be specific
- Describe exactly what the subject wears, their pose, and expression.
- Set the scene: time of day, weather, colours, environment details.
- Reference a photography or art style: "cinematic", "editorial", "fine-art sports photography", "anime-style illustration".
Avoid vague language
Avoid
"A person in a cool setting doing something awesome"
Better
"A {age}-year-old {gender} as a jazz pianist performing under a warm amber spotlight in a smoky underground club, fingers dancing across ivory keys"
Placeholders reference
Placeholders are replaced with the user’s actual profile data when the poster is generated. Use them to personalise your prompt.
| Placeholder | Replaced with | Available in |
|---|---|---|
{name} | User’s name (e.g., "Alex") | Human & Pet |
{age} | Age (e.g., "25") | Human |
{gender} | Gender (e.g., "male", "female") | Human |
{animal_type} | Animal type (e.g., "golden retriever") | Pet |
{profile_glasses_prompt} | Auto-injected instruction to preserve eyeglasses (if the profile has glasses) | Human |
Unfilled placeholders are automatically removed from the final
prompt. This means you can include optional placeholders without
worrying about leftover text.
6
Negative prompts
A negative prompt tells the AI what to avoid. It helps
eliminate common artefacts and unwanted elements.
Recommended defaults
Start with these and add template-specific terms as needed:
blurry, low quality, distorted face, extra limbs, deformed
hands, watermark, text overlay, bad anatomy, extra fingers
Common additions
- Period pieces: add "modern elements,
anachronistic objects"
- Sports action: add "unnatural pose, stiff
posture, missing limbs"
- Portraits: add "crossed eyes, unnatural skin"
- Food scenes: add "unappetizing food,
unnatural colours"
7
Face swap best practices
When face swap is enabled for your template, MDPoster replaces
the generated face with the user’s actual face photo. For
face swap to work well, the face must be clearly visible in the generated image.
Important
If your template concept renders the subject as a silhouette,
from behind, with the face hidden, or in a highly abstract art
style (pixel art, geometric shapes, stencil art), you should disable face swap for that template. Face swap on an invisible or abstract face
will produce poor results.
Do
-
Describe the subject facing the camera or in three-quarter
view.
-
Include descriptive terms like "looking at the camera",
"facing the viewer", "confident expression".
-
Ensure good lighting on the face — "well-lit features",
"soft studio lighting on the face".
-
For action poses, make sure the face is still visible —
not blocked by arms, equipment, or motion blur.
Avoid (with face swap on)
- "Silhouette", "shadow figure", "back turned to camera"
- "Walking away", "not looking back"
-
"Face obscured by mask/helmet/visor" (unless intentional)
- Extreme artistic abstraction of the face
When to disable face swap
Disable face swap in the template settings when:
- The template is a silhouette or double-exposure concept
-
The art style is pixel art, geometric, stencil, or heavy
abstraction
-
The subject is shown from behind as a core part of the concept
-
The template is image-based (transforms an uploaded photo
rather than placing someone into a scene)
8
Name rendering & typography
You can configure whether the user’s name appears on the
poster and how it’s rendered.
Name mode
- Optional — The user decides whether to include
their name. Good default for most templates.
- Required — The name is always shown. Use
for templates where the name is integral (sports jerseys, award
certificates, magazine covers).
- Hidden — The name is never shown. Use for
artistic templates where text would be distracting.
Name rendering instructions
If your template shows the name, add a
name rendering instruction describing exactly how
and where the name should appear. Examples:
Sports jersey:
"Display the name across the back of the jersey in bold
block letters matching the team font"
Magazine cover:
"Show the name as the cover headline in elegant serif
typography at the bottom of the frame"
Trophy/award:
"Engrave the name on the trophy plaque in classic gold serif
lettering"
9
Custom fields & conditional blocks
Custom fields let you add extra user inputs beyond the standard
profile data. When you define a custom field, users see an input
when generating a poster.
Using custom fields
Define a custom field key (e.g., team_name) in your template settings, then use {team_name}
in your prompt. The user fills in the value when generating.
Conditional blocks
Wrap sections of your prompt in conditional tags to include them
only when a field has a value:
[IF_TEAM_NAME]Wearing a jersey with "{team_name}" printed
on the back.[/IF_TEAM_NAME]
If the user does not provide a team name, the entire block is
removed from the prompt.
10 Pro tips
- Test with different profiles. Try your template
with male and female profiles, different ages, and different photo
styles to make sure it works broadly.
- Add a great cover image. Your cover image is the
first thing users see. Generate a few versions and pick the best
one for your template thumbnail.
- Use photography references. Mentioning a specific
photography style (e.g., "editorial fashion photography", "sports
photojournalism") gives the AI a clear creative direction.
- Keep negatives focused. A long negative prompt
can confuse the model. Stick to the most important things to avoid.
- Consider the aspect ratio. Think about how your
scene fills a 3:4 poster frame. Vertical compositions generally
work best.
- Read the
Creator Terms. Make sure your templates comply with content guidelines. Templates
with offensive, sensitive, or copyright-infringing content will
be rejected.
Ready to create your first template?
Put what you've learned into practice. Apply to the Creator Program and start designing templates.