How to Use Three Free AI Images a Month Well

A quota is a constraint, and constraints are useful

Every CartoonMakerAI account gets three AI-generated images a month at no cost, resetting on the first. That's deliberately small. It's enough to settle a character, test a thumbnail direction, or board the opening of a story — and not enough to generate idly, which turns out to be the point.

Unlimited generation encourages a particular bad habit: rerolling. You type a vague prompt, dislike the result, and press the button again with barely any change, hoping the model lands somewhere better. Thirty images later you have one you tolerate and no idea why it worked. Three images a month makes that approach impossible, and the alternative — deciding what you actually want before you spend a slot — is a skill that pays off when you move to video, where every generation costs real credits.

Spend the first image on the hardest question

Whatever you're making, one decision is doing most of the work. For a story channel it's usually the main character. For an explainer it's usually the object being explained. For a thumbnail it's the single element that has to be readable at small size.

Put your first image there. Not on a nice-looking establishing shot, and not on a background you could describe in a sentence — on the thing that, if you get it wrong, makes everything after it wrong too. If the character is right, the rest of the series inherits it. If the character is vague, every later frame inherits that instead.

Write nouns, not adjectives

The single highest-leverage change to how most people prompt is replacing evaluative words with concrete ones. "A beautiful, whimsical fox character" tells the model almost nothing it can draw. "A round-faced fox with a rust-red coat, small round glasses, and a green scarf" gives it four things to hold onto, and — critically — gives you four things to repeat next month when you need the same fox again.

This matters more under a quota than it does without one. A described character can be regenerated. A lucky character cannot, and once you've used a slot discovering one you can't reproduce, you've spent the month's budget on something you can't build on. The character consistency guide goes further into why this happens.

Save the prompt, not just the image

When an image works, the image is the smaller half of what you made. The text that produced it is the reusable part: it's what lets you generate the same character in a different pose next month, or hand the same visual direction to a video generation run.

Keep a plain text file. One block per character, one per recurring setting. It takes a minute and it compounds — by the third month you're not starting from scratch, you're starting from a description you already know works.

Decide the aspect ratio before you generate

A 1:1 image cropped to 16:9 is not the same as a 16:9 image. The model composes for the frame it's given — subject placement, negative space, how much room the background gets. Cropping afterwards throws away those decisions and usually cuts through the part of the composition that made the frame work.

If the image is for a thumbnail, generate at 16:9 or 9:16 depending on where it's publishing. If it's a storyboard frame for a video, generate at the video's ratio. Getting this right costs nothing; getting it wrong costs a slot.

When to stop and switch to video

Three images is roughly the right amount to answer "what does this look like?" It is not enough to answer "does this move well?" Once the look is settled — character described, palette chosen, framing decided — the next questions belong to video, and they cost credits rather than quota.

That's the intended shape of the workflow: stills are the cheap place to be indecisive, and motion is where you commit. Spending the free three on decisions rather than on rerolls is what makes the expensive step shorter.

What happens when you run out

Past the third image, each one costs a single credit from the same balance your videos draw on, and the quota resets on the first of the next month regardless of whether you used it. A failed generation doesn't consume a slot — only images that actually render count.

If you find yourself consistently wanting more than three, that's real signal rather than a nuisance: it usually means you've moved from experimenting to producing, and the plans are priced for that.