Skip to content
OpenDream AI

AI Art for Beginners: Your First Steps

A gentle, hands-on introduction: make a first image, grow it one detail at a time, use simple templates and sidestep the mistakes almost everyone makes early on.

4 min read · Updated September 2026

You do not need to be an artist

Making images with AI is closer to directing than to drawing. You describe what you want, look at what comes back, and steer. The skills that help are ones most people already have: noticing what they like, describing it in words, and being willing to try again. This guide covers the first steps, a few simple templates and the mistakes almost everyone makes at the start.

Your first session: keep it small

For a first attempt, pick something simple with a single subject: a cup of tea on a windowsill, a cat on a stack of books, a lighthouse at sunset. Simple subjects let you see clearly how each word in your prompt changes the result.

Write one plain sentence:

A cup of tea on a sunny windowsill

Generate it. Then add one detail at a time and generate again:

A cup of tea on a sunny windowsill, morning light, steam rising
A cup of tea on a sunny windowsill, morning light, steam rising, soft watercolor illustration

Watching the image change with each addition teaches you more than any list of tips. You will quickly notice which words have a strong effect (light, medium, style) and which barely register (vague adjectives).

Three beginner templates

Templates give you structure without having to remember rules. Fill in the brackets.

Scene template

[Subject] in [place] at [time of day], [type of light], [art style], [mood]

Example: a red bicycle leaning against a stone wall in a quiet village at dusk, warm streetlights, soft gouache illustration, peaceful.

Portrait template

[Shot type] portrait of [person description] wearing [clothing], [background], [light], [photo or art style]

Example: close-up portrait of a smiling elderly fisherman wearing a knitted cap, harbor in the background, soft overcast light, natural documentary photograph.

Object template

[Object with material and color] on [surface], [background], [lighting], [style], [composition]

Example: a hand-thrown blue ceramic vase on a light oak table, plain white wall, soft window light, minimalist product photograph, centered.

The templates section has many more, including ones with fill-in variables so you only change the parts that matter.

Picking settings without overthinking

When you are starting out, only a few settings really matter:

  • Aspect ratio: square for general use, vertical for portraits and phone wallpapers, wide for landscapes and banners.
  • Number of images: generating a few at once lets you compare and pick a favorite.
  • Model: if you are unsure, let an automatic mode choose for you. You can explore specific models later.

Advanced options can wait until you have a feel for the basics.

How to improve a result

When an image is almost right, resist the urge to rewrite the whole prompt. Ask yourself exactly what is wrong and change only that.

  • Subject too small? Add "close-up" or "medium shot".
  • Too dark or too flat? Describe the light: "bright morning sun from the left".
  • Wrong feel? Change a mood word or the style.
  • Something unwanted appeared? Describe the scene more precisely so there is no room for it, or remove it with an edit.

If you like most of an image but one thing is off, editing is usually faster than regenerating. A plain instruction such as "remove the second cup" or "make the sky more orange" often does the job.

Common beginner mistakes

Using praise instead of description. "Stunning, amazing, best quality" does not tell the model what to draw. Replace each word with a visible property.

Packing too much in. A dragon, a castle, a knight, a storm, a rainbow and a village in one prompt gives the model too many competing priorities. Start with one or two elements.

Mixing contradictory styles. "Photorealistic cartoon oil painting" forces awkward compromises. Pick one.

Expecting long text to come out perfectly. Short words in quotation marks work much better, and some models handle lettering far better than others.

Giving up after one try. Generation involves randomness, so the same prompt can produce a weak image and a strong one. Try a few before changing the words.

Forgetting the shape. A tall subject in a wide frame looks cramped. Match the aspect ratio to the subject.

Borrowing names and characters. Describe the qualities you like instead of naming living artists or copying famous characters. It keeps your work your own.

Build a small personal library

As you experiment, save the prompts that work. A simple note with the prompt, the model and the aspect ratio saves time later. Over a few sessions you will build a set of reliable starting points: favorite lighting phrases, styles that suit your taste, and compositions that fit what you make.

Browsing the gallery and the prompt library is also a good way to learn. When you see an image you like, read its prompt and notice which words seem to be doing the work.

A gentle first week

  • Day 1: Make five simple images, each with a single subject.
  • Day 2: Take one subject and try it in five different styles.
  • Day 3: Take one prompt and change only the lighting, five times.
  • Day 4: Try the three templates above with your own ideas.
  • Day 5: Pick your favorite image so far and improve it with edits.
  • Day 6: Make a vertical and a wide version of the same scene.
  • Day 7: Create a small set of three images that share one style.

By the end of the week you will understand how subject, light, style and format interact, which is most of what separates a beginner from a confident user.

Keep learning

Put it into practice

Try what you just read in the Create Studio.

Start creating