Tutorial · Image AI · Votel

Tutorial: Image AI in Votel

Votel's Library has an AI image tool built in. Most people type two words and get a random picture. Here is how to write a prompt that gets the picture you were imagining, with every example generated live in Votel and shown right next to its prompt.

5Real prompts, real images
7Aspect ratios to pick from
$0.075Per image on Nano Banana 2
Generated in a live Votel account on November 21, 2026. Labels, models and prices can change.
AI-generated photo of a smiling plumber kneeling at a kitchen sink with a wrench
16:9 website hero
AI-generated latte with heart latte art on a wooden table in a coffee shop
Best prompt
AI-generated busy coffee shop interior from a two-word prompt
Two-word prompt
01

Find the tool

In Votel, open Library and click the AI Image Gen tab. The Library also has tabs for Files, Snapshots, Websites, Emails, AI Brand Kit, AI Video and AI Audio. Inside AI Image Gen there are two switches at the top left: Image and GIF. This guide uses Image.

Everything you make lands in one grid, newest first. Your balance sits at the top right, and the prompt box is pinned to the bottom of the screen.

Votel AI Image Gen prompt bar with Add image, a Describe the image you want box, the Nano Banana 2 model, 1:1 aspect ratio, brand kit and Generate button showing 0.3 credits
The whole control panel: add image, prompt, model, shape, brand kit, price and Generate.
02

Pick model, shape and brand kit

Three dropdowns sit under the prompt box. Set them before you write, because the shape changes how you describe the picture.

ControlWhat you get
ModelNano Banana Pro, Nano Banana 2, GPT Image 2, GPT Image 1.5, Ideogram 4.0, Ideogram 3.0. Price per image shows next to each.
Aspect ratio1:1 Square, 16:9 Wide, 9:16 Tall, 4:5 Social, 3:2 Photo, 3:4 Portrait, 4:3 Landscape.
Brand kitNone, or your brand kit. It adds your logo as a reference image.
Add imageReference images, on the Nano Banana and OpenAI models. Ideogram is prompt only.
Test cheap, then upgrade. I made every image here on Nano Banana 2 at 0.3 credits. Get the prompt right on the cheap model first. Only then try a pricier one with the same words.
Votel model picker listing Nano Banana Pro, Nano Banana 2, GPT Image 2, GPT Image 1.5, Ideogram 4.0 and Ideogram 3.0 with credit prices
The model list, with a price on every row.
03

What goes in a prompt

An AI image model does not read your mind. It fills every blank with its own guess. A good prompt closes the blanks you care about. I use the same seven parts every time.

PartWhat it answersExample
SubjectWhat is the one main thing?"a single oat milk latte with heart latte art"
SettingWhere is it?"a small independent coffee shop, exposed brick"
LightWhen is it and where does light come from?"warm golden morning light from a large window on the left"
CameraHow close, what angle, how blurry is the background?"low three-quarter angle, shallow depth of field"
StylePhoto, illustration, 3D? What feel?"photorealistic editorial photo, subtle film grain"
CompositionWhere does the subject sit? Where is the empty space?"empty space in the upper right"
Avoid listWhat must not appear?"No text, no logos, no watermark"
Order does not matter, specifics do. If a part is missing, the AI invents it. That is fine when you do not care and a problem when you do.
Same idea, three prompts

Weak, better and best, side by side

I wanted a picture for a local coffee shop. Same model (Nano Banana 2), same square shape, three levels of effort. Nothing here was cherry picked. This is the first image each prompt gave me.

AI-generated busy coffee shop interior with a barista, many customers and a chalkboard menu
Weak · 2 words

coffee shop

A full room: barista, a dozen customers, chalkboard, laptops. Nothing wrong, but nothing chosen on purpose.
AI-generated latte in a speckled ceramic cup on a wooden table by a window
Better · 1 sentence

A latte in a ceramic cup on a wooden table inside a cozy coffee shop, soft morning light from a window.

Now there is a hero: the latte. The window light is right. The camera and background were still the AI's call.
AI-generated close-up of a heart latte art cup with steam, blurred cafe and customers behind
Best · full brief

Photorealistic editorial photo of a single oat milk latte with heart latte art in a speckled cream ceramic cup on a worn wooden table, shot from a low three-quarter angle with shallow depth of field. A small independent coffee shop softly blurred in the background: exposed brick, plants on a window sill, two customers chatting. Warm golden morning light from a large window on the left, gentle steam rising from the cup. Natural warm color, subtle film grain. Clean composition with empty space in the upper right. No text, no logos, no watermark.

Steam, heart art, blurred brick and customers, warm light from the left. Ready for an ad.
It still missed a little. I asked for empty space in the upper right. The blurred people and a menu board landed there instead, so a headline would need a darker overlay. Read your result against your prompt, line by line, and fix the one thing that missed.
Read the difference

What each addition did

What I addedWhat changed in the picture
"a latte in a ceramic cup" (subject)The scene stopped being "a room" and became "a cup". The AI had one thing to focus on.
"soft morning light from a window"Light became warm and directional instead of flat.
"single", "heart latte art", "speckled cream"The cup is specific: one cup, art on top, a speckled glaze.
"low three-quarter angle, shallow depth of field"The camera dropped to table level and the background went soft.
"exposed brick, plants, two customers chatting"The blurred background has a story instead of random extras.
"gentle steam rising"Steam appears. Small details only show up if you ask for them.
"empty space in the upper right"Partly followed. Treat composition requests as a strong hint, not a guarantee.
"No text, no logos, no watermark"No fake signage or garbled words in the frame.
Prompts for real work

Three real jobs, with the shape that fits

Pick the aspect ratio for where the image will live, then describe where the empty space goes so you can add a headline or caption later.

AI-generated website hero photo of a smiling plumber kneeling at a kitchen sink with a wrench

Website hero (16:9 Wide)

A local service business needs a hero that shows the work, with room on the left for a headline. The trick is telling the AI which third of the frame the person gets.

  • Nano Banana 2
  • 16:9
  • 0.3 credits
Copy this prompt

Photorealistic wide website hero photo of a friendly plumber in a navy work shirt kneeling beside a kitchen sink and smiling, holding a wrench, bright modern home kitchen, soft natural window light, camera at eye level. The plumber sits on the right third of the frame and the left side is clean, calm kitchen space where a headline could go. Candid, natural color. No text, no logos, no watermark.

AI-generated photo of a woman meditating on a mat in a bright yoga studio with plain wall above her

Social post (4:5 Social)

For Instagram and Facebook feeds, 4:5 takes up more screen than a square. Ask for a plain top third and your caption has somewhere to go. Color words like "sage green accents" keep a whole month of posts looking like one brand.

  • Nano Banana 2
  • 4:5
  • 0.3 credits
Copy this prompt

Photorealistic vertical social media photo for a local yoga studio: a woman in a calm seated pose on a mat in a bright, minimal studio, soft morning light through tall windows, warm neutral tones with sage green accents, shallow depth of field. She fills the lower two thirds of the frame and the top third is plain light wall, leaving room for a caption. Peaceful, authentic, not staged. No text, no logos, no watermark.

AI-generated heart latte art cup with steam on a wooden table in a coffee shop

Product or menu shot (1:1 Square)

One hero object, one lighting direction, one blurred story behind it. This is the best prompt from the comparison above, and it is ready for an ad, a menu board or a "now open" post.

  • Nano Banana 2
  • 1:1
  • 0.3 credits
Copy this prompt

Photorealistic editorial photo of a single oat milk latte with heart latte art in a speckled cream ceramic cup on a worn wooden table, shot from a low three-quarter angle with shallow depth of field. A small independent coffee shop softly blurred in the background: exposed brick, plants on a window sill, two customers chatting. Warm golden morning light from a large window on the left, gentle steam rising from the cup. Natural warm color, subtle film grain. Clean composition with empty space in the upper right. No text, no logos, no watermark.

These are AI images. The plumber and the yoga student are not real people and the businesses are made up. For a client, generate images that fit their trade, and never present a generated person as a real customer.
Fill in the brackets

Copy-paste templates

These are the recipes behind the examples. Swap each bracket for your client's real details and run it on the cheapest model first.

Website hero · 16:9

Photorealistic wide website hero photo of [subject doing something] in [setting], [light], camera at [angle]. [Subject] sits on the [right/left] third of the frame and the other side is clean [background] where a headline could go. Candid, natural color. No text, no logos, no watermark.

Social post · 4:5

Photorealistic vertical social media photo for a [type of business]: [subject and pose] in [setting], [light], [color palette], shallow depth of field. The subject fills the lower two thirds of the frame and the top third is plain [wall/sky/background], leaving room for a caption. [Mood]. No text, no logos, no watermark.

Product shot · 1:1

Clean product photo of [product] on [surface], [light direction] studio light, soft shadow, [background color] background, shot straight on with the product centered and lots of empty space around it. Sharp focus, true-to-life color. No text, no logos, no watermark.

Flat illustration · any ratio

Flat editorial illustration of [idea or scene] in [2 to 3 named colors], simple shapes, clean lines, generous whitespace, [composition: centered / left-aligned]. No people, no text, no watermark.

Stop burning credits

Habits that save credits

  • Change one thing per run. If you change the light and the subject together, you will not know which one helped.
  • Use Reuse prompt and settings. Click any image, then that button, and your last recipe is back in the box.
  • Always end with an avoid list. No text, no logos, no watermark. Add "no people" when you want an empty scene.
  • Look at hands, faces and signs. Zoom in before you post. Fix the prompt or run it again rather than editing around an error.
  • Name the shape before the words. A 16:9 hero and a 9:16 story need different compositions. Pick the ratio first.
  • Save winners as templates. The best prompt for a client becomes their reusable brand recipe.
Want it to move? The same prompt recipe works for video. Part 2 of this series, how to use AI video in Votel, shows weak, better and best video prompts with the clips they made.
Quick answers

FAQ

Which AI image model should I start with in Votel?

On November 21, 2026 the model picker listed Nano Banana Pro (0.6 credits per image), Nano Banana 2 (0.3), GPT Image 2 (from 0.7), GPT Image 1.5 (1.2), Ideogram 4.0 (0.4) and Ideogram 3.0 (0.4). Every image in this guide came from Nano Banana 2, the cheapest one on the list, so you can test a prompt for pennies before you spend more on a bigger model.

How much does one AI image cost in Votel?

The Generate button shows the credit cost before you click. On Nano Banana 2 it showed 0.3 credits, about $0.075, for the square, wide and 4:5 images in this guide. Other models cost more, and the price can change, so read the button each time.

Can I edit or restyle a photo I already have?

The picker marks Nano Banana and OpenAI models as taking reference images, and the prompt box has an Add image button. Ideogram is prompt only. I did not test photo edits for this guide, so I only generated from text. Try it on a copy of a photo you can afford to lose.

Why did the AI add things I never asked for?

It fills every gap with a guess. My one-line prompt, coffee shop, came back as a busy room with a barista, a chalkboard menu and a dozen customers. Naming the subject, the setting and the light takes the guessing away.

How do I reuse a prompt I liked?

Click any image in the Library to open it. The panel shows the full prompt, the aspect ratio and the date, with a Reuse prompt and settings button that puts it back in the prompt box so you can change one thing and run it again.

Can I use generated people as customers or testimonials?

No. A generated person is not a real customer. Use them for mood, backgrounds and illustration, and keep reviews, case studies and before and after claims to real people and real results.

Your turn

Make your first three images tonight.

Pick a client, copy one template, fill the brackets, and run it. Compare the first result to your prompt line by line. Votel has a 14-day trial, then plans are $97 or $297 a month.

All Things Votel is independent and not operated by or affiliated with Votel.ai. Links to Votel are affiliate links. Images in this guide were generated in a live Votel account on November 21, 2026 and may look different when you run the same prompts.