Creating and sending

Images and AI generation

Uploading, dragging in, or generating images with AI — and the Media library that keeps them.

Every image you upload into an email, or generate with AI, is kept in your brand's Media library — so you can find it again instead of re-uploading or, worse, paying to regenerate something you already made.

Adding an image to a block

Select an Image block and you have four options:

Choose from media — every image this brand has already used, including ones generated with AI. Search runs over the prompts a generated image was made from, and you can widen it to every brand on the account.
Click the upload area and choose a file.
Drag an image straight onto the upload area from your desktop.
Paste a URL.

Choosing an existing image replaces only the picture. The alt text, link, width and alignment you set on the block stay as they were.

Uploads must be PNG, JPG, GIF or WebP, and under 4 MB. They are stored on our own infrastructure and served from your email, so they keep working even if your website is down.

Images are optimised for you

Every image is resized and re-compressed on the way in — down to at most 1600 px on its longest edge, and converted to WebP. A photo straight from a phone typically comes out around 90% smaller with no visible difference at the size an email displays it.

This matters more than it sounds. An email image is downloaded by every recipient: a 4 MB photo sent to 10,000 people is 40 GB of transfer, a slow-loading email, and a real cost to anyone reading on mobile data.

Animated GIFs are left alone — re-compressing one would keep only the first frame.
An image that is already small is left as it is, never made larger.
Photos taken on a phone are rotated the right way up automatically.
You do not need to prepare or resize anything before uploading.

Cropping

With an image in the block, Crop image opens a cropping tool — as does Crop on a landing page's image field. Drag inside the selection to move it, drag a corner to resize, or pick a ratio to lock it — including 4:1, which is useful for thin banners.

Cropping never destroys the original

The crop is saved as a new image and the block points at it. The original stays in your Media library, so a crop you regret later costs you nothing.

Cropping works on images stored in Nectazo. An image you pasted in from another website usually cannot be cropped here — browsers block reading pixels back from another site's image — so upload it first.

Generating an image with AI

The Image block has a Generate with AI button, and the Media page has the same one. So do the image, gallery and background fields on a landing page, where it is the sparkle beside the media button. They all open the same panel: describe what you want, choose a shape and a quality, and generate.

1Write a prompt. Be specific about subject, lighting, mood and colour — vague prompts give generic results.
2Pick a shape: square, landscape, portrait, wide, tall, or a 3:1 banner.
3Pick a quality. Standard suits almost all pictures; High is for anything with text on it.
4Generate. It usually takes well under a minute. If it is not right, adjust the prompt and try again.
5Click Use this image to place it in the block.

Whether you use it or not, every generated image is saved to your Media library.

Editing an image

The same panel edits an existing image instead of inventing one. Give it a starting image and the prompt stops describing a picture and starts describing a change — “put it on a warm neutral background”, “remove the text”, “make it feel like evening”.

An Image block that already has a picture opens straight into editing it, whether you press Generate with AI in the settings or the sparkle in the block's toolbar. To start again from nothing instead, remove the image from the panel and it goes back to generating from scratch.

There are three ways in, and they all land in the same place:

Before you generate. Open Generate with AI and either upload an image or press Choose from media. With no image, it generates from scratch as usual.
Straight after generating. Edit this takes the result you are looking at and makes it the starting point for the next change, keeping the shape, quality and brand setting you already chose.
From the Media page. Open any image and press Edit with AI.

An edit never overwrites the original

Editing produces a new image and saves it alongside the one you started from. The original stays in your Media library untouched, so an edit you dislike costs you nothing but the tokens.

An edit costs the same as a generation, because it is the same work.

Why there is no 4:1 option here

The model will not generate anything wider than 3:1. If you need a thinner banner, generate at 3:1 and then crop it to 4:1 — cropping has no such limit.

Use brand styles

If you have set a visual identity for the brand, the panel shows a Use brand styles checkbox, ticked by default. With it on, your brand's colours, font style and visual direction are added to the prompt, so generated images look like they belong to you rather than to nowhere in particular.

Set this up under Settings → Brand → Visual identity:

Field	What it does
Primary, secondary and accent colour	The palette the image should be built from. Nectazo describes each one to the model by name as well as hex — "amber (#ffc200)" — because a name steers the result better than a code alone.
Font style	How any text in the image should look. A description, not a font file: image models cannot load a typeface, but they respond well to "a bold geometric sans-serif".
Other visual direction	Free text for anything else — photography style, materials, mood, motifs.

The checkbox only appears when the brand actually has a visual identity set, and each brand has its own — which is the point, if you run more than one.

It is guidance, not a guarantee

The model is being asked to work in your palette, not constrained to it. Expect a strong resemblance rather than exact colour matching, and check the result before sending.

What the quality settings do

Both use the same model, at the same pixel size — the shape decides that, not the quality. What changes is how much work the model does on the image, which is why one takes longer and costs more.

Quality	Best for	Tokens
Standard	Almost all email imagery. The right default.	5
High	Detail that will be looked at closely. Noticeably slower, and on large shapes it can take long enough to time out.	20

The difference that matters is text. Legible, correctly spelled words are the hardest thing an image model does, so a banner with a headline on it is worth High. For a photograph shown at the width an email displays it, the two are very hard to tell apart — use Standard and keep the tokens.

What an image costs

Everything AI does is paid for from one pool of tokens, but not everything costs the same: an image costs considerably more to produce than a paragraph of text. A text generation costs 1 token, a Standard image 5, a High image 20. How many you get each month depends on your plan — see Plans and limits.

A generation that fails costs nothing.

Check what you are sending

AI image generation can produce text that is misspelled, hands that are wrong, and details that do not match reality. Look properly at anything before it goes to your list, and never use a generated image to depict a real product you sell — show the actual product.

The Media library

Media in the sidebar lists everything, newest first. You can:

Filter to just AI-generated or just uploaded images.
Switch between the brand you are in and all brands in your account.
Click any image to see its details — and, for generated images, the prompt that made it.
Copy its URL to paste into another image block.
Delete it.

Deleting is not retroactive

Emails already sent are unaffected — recipients have their own copy of what was delivered. But any draft still pointing at a deleted image will show a broken image, so check before removing something.

Something unclear or wrong? Tell us and we'll fix it.