Images

Describe a picture and Firelent creates it: product shots, illustrations, posters, social posts, mood images for your website. Then refine it in plain words — “warmer”, “closer”, “less busy” — until it's right.

How it works#

Each image — and each change — costs 3 credits. If something goes wrong, you get them back automatically.

Write a great image prompt#

Describe the picture in full sentences, roughly in this order:

	Example
Subject	A barista pouring latte art
Setting	in a small café with a big window
Light	warm morning sun, soft shadows
Camera or style	close-up photo, shallow depth of field — or “flat illustration”, “watercolour”, “3D render”
Colour and mood	natural colours, calm, cosy
Shape	wide landscape, tall portrait or square

A close-up photo of a barista pouring latte art in a sunlit café, warm morning light through the window, shallow depth of field, natural colours, wide landscape format.

Rules of thumb#

• Say what you want, not what you don't. “An empty street at dawn” works better than “a street with no people”.
• One main subject. Pack in five ideas and some get dropped or blended together.
• Use photography words for photos: close-up, wide shot, from above, golden hour, soft light, shallow depth of field.
• Name the medium for illustrations: line drawing, flat vector, watercolour, paper cut-out, clay, 3D render.
• Say what it's for. “A banner for a bakery website” gives Firelent useful context.

Shape and size#

There's no size setting — describe the shape you need in your prompt. In the example above, “wide landscape format” came back as a 16:9 image, 1376 × 768 pixels.

For	Ask for
A website banner or slide	Wide landscape (16:9)
An Instagram post	Portrait (4:5) or square
A story, reel or phone wallpaper	Tall portrait (9:16)
A profile picture or icon	Square

Images are made for screens. For print, you need about 300 pixels per inch: an image 1,376 pixels wide prints sharply at roughly 11 to 12 cm wide.

Text in images#

Short text — a word or a short headline — can work. Put the exact words in quotes, keep them short and check the spelling. For anything longer, like a menu or a flyer full of details, create the image without text and add the words in your design tool or on your website.

Refine by asking#

Every follow-up in the same chat edits your newest image:

• Change one thing at a time. “Make it warmer” first, then “move the cup to the left”.
• Say what to keep. “Keep the barista and the window — only change the light.”
• Be clear about what matters. Small requests can come out subtle, like the steam above. If it's important, say so: “clearly visible steam”.
• Start fresh for something new. To make a different image, open a new chat with Create → Image — a follow-up always edits the latest one.

Use your image#

• Download it from the full-size view, or later from Library with the Images filter.
• Put it on your website: open your website project, attach the image in the chat and say where it goes — “use this as the hero image”.
• Use it in a deck: insert it into your downloaded slides in PowerPoint, Keynote or Google Slides.

Good to know#

• One image per request. For variations, ask again — sets like “three poster versions” come out as a single image.
• Real people, brands and logos: only create images of people and brands you have the right to use, and don't present a created image as a real photo of a real event.
• Hands, small text and fine details are where images most often go wrong. Look closely before you publish.

FAQ#

Can I edit an older version?#

Not yet — follow-ups always edit the newest image. Download any version you like as you go.

Can I choose the exact size?#

Not yet. Describe the shape in your prompt, then crop in any image editor if you need exact pixels.

Why doesn't my image have the text I asked for?#

Text is the hardest part for image models. Keep it to a few words in quotes, or add it afterwards.

Next steps#