The Basics, step-by-step
Last updated
Last updated
Here's a little guide that might help you understand prompting on Ideogram, step by step.
The prompts that will be used will always be written like this:
Prompt:
A rough pencil sketch featuring a boy with sandy hair and blue eyes, wearing a hoodie checked shirt and jeans, playing with a friendly Golden Retriever, in a park on a sunny day.
The text in green means this text is new or was added to a prompt previously used.
Strikethrough purple text means this text has been removed or replaced from a prompt previously used.
For all examples below, we'll use the following settings: • Magic Prompt set to OFF • Aspect ratio of 1:1 • Model v1.0 or 2.0 as specified.
Ready? Let's start!
Let's start with a very basic prompt:
Prompt: A boy on a neutral grey background.
Nothing complicated here. The image of a boy and a neutral grey background as requested. Now let's add more details...
Let's add some specific details like hair and eye colour, and maybe he's wearing a checked shirt, and how about a more interesting background.
Prompt:
A boy on a neutral grey background with sandy hair and blue eyes, wearing a checked shirt, against a backdrop of wooden planks.
This one is more beautiful. Perhaps a little too static. Let's try to make it a bit more dynamic...
OK, that previous prompt and the image it generated were still ordinary. Perhaps changing the angle and/or the point of view would produce a more interesting image?
Prompt: A boy with sandy hair and blue eyes, wearing a checked shirt, left-sided partial profile pose, against a backdrop of wooden planks.
Almost a studio photo isn’t it?
Note: there are also other terms used to achieve this pose, such as three-quarter pose, or slightly turned to the left, etc.
Why not add a few elements to make the overall image more attractive?
Prompt: A boy with sandy hair and blue eyes, wearing a checked shirt and jeans, sitting cross-legged on the floor, beside a friendly dog, between them both there is a house plant in a pot, against a backdrop of wooden planks.
Note how well the AI has rendered all the requested elements and their relative positioning in the image composition. Now let's change the mood next...
Until now, images have been rather static. Let's change that.
Prompt: A boy with sandy hair and blue eyes, wearing a checked shirt and jeans, playing with a friendly Golden Retriever, in a park on a sunny day.
The addition of an environment contributes greatly to the atmosphere of a picture. And everyone knows that the Golden Retriever is a child's best friend. 😉
So far, the AI has provided us with photograph-like images without us even asking. There are also predefined style tags, as shown below, that can be selected and added to the prompt to specify the style of image we want to generate.
These are really just words added at the end of the prompt to specify the style you want, nothing more. So if you select 'illustration', it is the very same as if you've added the word 'illustration' at the end of your prompt.
Many of these 'tags' are less effective, or even ineffective, when you use the v1.0 model, depending on the subject of the prompt. They are still very useful in older versions v0.1 and v0.2. Version 1.0 has a different “language” approach compared to it’s two predecessors.
Prompt: A boy with sandy hair and blue eyes, wearing a checked shirt and jeans, playing with a friendly Golden Retriever, in a park on a sunny day, illustration.
Prompt: A boy with sandy hair and blue eyes, wearing a checked shirt and jeans, playing with a friendly Golden Retriever, in a park on a sunny day, 3d render.
Prompt: A boy with sandy hair and blue eyes, wearing a checked shirt and jeans, playing with a friendly Golden Retriever, in a park on a sunny day, anime.
With Ideogram version 1.0, it's generally more efficient to integrate the desired style directly into the prompt.
Let's try this next...
Let's put some stylistic details in the original prompt and see the results.
Original Prompt: A boy with sandy hair and blue eyes, wearing a checked shirt and jeans, playing with a friendly Golden Retriever, in a park on a sunny day.
Prompt:
Washed-out soft watercolour painting featuring a boy with sandy hair and blue eyes, wearing a checked shirt and jeans, playing with a friendly Golden Retriever, in a park on a sunny day.
Prompt:
A rough pencil sketch featuring a boy with sandy hair and blue eyes, wearing a checked shirt and jeans, playing with a friendly Golden Retriever, in a park on a sunny day.
Prompt:
An origami art featuring a boy with sandy hair and blue eyes, wearing a checked shirt and jeans, playing with a friendly Golden Retriever, in a paper-art park on a sunny day.
Of course, you can mix and match all the styles to your heart's content.
Prompt: Soft, faded and minimalist watercolour painting with black ink pen outlines, featuring a boy wearing a checked shirt and jeans, playing with a friendly Golden Retriever, in a park on a sunny day.
Let's say you want to give your image a title, as if you were making a book cover.
Prompt: Washed-out soft watercolour painting featuring a boy with sandy hair and blue eyes, wearing a checked shirt and jeans, playing with a friendly Golden Retriever, in a park on a sunny day. At the top of the image, the book title is: "Michael and his dog" in a child-book type of title typography. At the bottom, the author's name: "John Johnson"
Not everyone is going to make children's books, so let's try other subjects.
Prompt:
A sleek and modern illustration of a logo for "PetPalace" featuring a minimalist stylized image of a regal cat wearing a crown, sitting next to a luxurious palace.
Prompt:
Close up photo of a beer can with a big modern sailor's anchor inspired logo named "Mariner's Brew" in big letters, a full glass of beer in the background, on the counter of an Irish pub, dynamic view, studio photography, food photography.
Prompt:
A cool cat with wearing a hat and sunglasses playing the saxophone with his paws on the stage of a jazz club, a neon sign in the background saying: "Cool Cat Jazz". Intense atmosphere, dark ambience, dark tones, jazz club genre, 40's and 50's look.
Enclosing text in a phrase between quotes “ ” seems the best way to make the AI understand it needs to render the text in the image.
