Editor and logos

Image editor

The editor finishes a work off: crop it, add a caption, overlay a shape, correct the colour. All of that without leaving the service and without opening a graphics application. It has its own item in the header — "Editor".

Image editor

On the left — the same works drawer as in the generator, the canvas in the middle, the layers panel on the right.

No need to download and upload back. Press the editor icon on a work card and it opens right here. The same is available from the generator: the preview of an uploaded picture has an "Edit image" button.

Your own file from disk

You can also load a file from your computer, not just pick a work from the drawer. The limits: one file, up to 5 MB, PNG, JPEG or WebP.

An uploaded file gets two buttons: apply it to the canvas and remove it. Once the file is open in the editor, the button becomes a green tick — the sign that this is what is being edited.

If there are unsaved edits, switching pictures asks for confirmation — you cannot lose your work by accident.

Five sections

Adjustments · Filters · Annotate · Fine-tune · Figures. The editor opens on Annotate with the Text tool selected — which is why it feels like it is "about captions".

Annotate offers eight tools: text (with font, size, weight and colour), image, rectangle, ellipse, polygon, a pen for freehand drawing, line and arrow.

The Adjustments section has an Original button that unfolds a list of ready crop proportions — each labelled with both the pixel size and the aspect ratio. Next to it: Flip Horizontally and Flip Vertically.

Crop presets in the editor

The list of proportions is the same as in the generator. Handy when the picture is already finished but you need another format for a cover or a post: crop it here instead of redoing the generation.

The Filters section holds ready colour presets: Original, TikTok Pop, Beauty Glow, Ektar 100, Urban Night, Golden Desert, Cinematic Blue, Food Blogger, Nature Fresh, Cosy Coffee and more. The strip scrolls sideways, and clicking applies the preset right away.

The Filters section

Fine-tune holds sliders: brightness, contrast, HSV, warmth, blur and sharpness.

The Figures section has the same rectangle, ellipse, polygon and line as the annotation tools, just gathered separately: handy when you need a backdrop for text or a frame rather than a caption.

Above the canvas — the size in pixels, zoom with plus and minus, and undo/redo buttons.

A second mode: a blank canvas

The "Reset" button does not do what it looks like. With no edits it simply clears the canvas. But if there are edits, it asks for confirmation and puts a blank white sheet in place.

Effectively this is a second mode of the editor: not "fix a picture" but draw from scratch. Need a simple banner, a backdrop with text or a diagram — press "Reset" and draw right here, no generation required.

The buttons at the bottom

The set depends on where the editor was opened from. On its own page there are four: Reset · Upload · Download · Save to gallery. Opened from the generator the set is different: Cancel · Download · Save to gallery · Apply.

ButtonWhat it does
Resetclear the canvas — or put a blank white sheet in place
Uploadtake a file from disk instead of a work from the drawer
Downloadtake the file for yourself; the work does not go to the gallery
Save to gallerycreate a new work from the edited image
Applyhand the result back to wherever the editor was opened from
Cancelleave without saving

"Save to gallery" creates a new work and the original stays — you cannot ruin the source. The result is saved as WebP at maximum quality (the picture is processed at a raised resolution) and lands at the top of your works, appearing in the drawer without a page reload.

The editor library's own save button has been removed — looking for "Save" inside its panels is pointless: only the buttons at the bottom save anything.

If the editor was opened from the generator, "Apply" returns the picture to where you came from — "Custom Image", an additional picture or the face slot; and if you opened the editor just to look, there is nowhere to return it. The generation size adjusts itself to the edited picture, if the model has such a size.

Edits are remembered between openings — close the editor and you will continue from the same place, not from a blank sheet. Closing it is not scary.

An uploaded picture is not a work

A file you uploaded yourself — into the editor or into "Custom Image" — does not become a work: it is stored separately, as a source for generation, and never appears in your feed of works.

Hence the answer to "I uploaded a picture and it isn't in my works". That is how it should be: to turn something uploaded into a work, run it through a generation or save it with "Save to gallery".

What you save from the editor starts private

A work created with "Save to gallery" appears for you only — it does not go into the shared feed. It is also flagged as adult, so in your own feed it may look blurred if you have turned such works off in profile settings. Both flags can be switched later on the work's card: Your works.

Logo generator

It lives in the "More" menu rather than next to the ordinary generator — hence the question "where are the logos".

Logo generator, step 1

Four steps

Each step in the bar carries two lines — a name and a caption under it:

StepNameCaption
1Brand information"provide the details"
2Composition and mood"pick a layout"
3Editing"adjust the layout"
4Styling"AI processing"

Step one is simple: the "Brand name" field with an example hint, a choice of layout size — and the "Create layouts" button, inactive until a name is entered.

Steps cannot be skipped. Until the previous one is finished, the next says "this section will become available once the previous one is complete" and asks you to go back. But you can move through completed steps — the step bar is clickable.

There is a separate restriction: a step closed by your plan. It looks different from the usual plans window — a card right inside the step, with an "Open plans" button. Don't confuse the two: in one case you need to finish the previous step, in the other to change plan.

The main reason people give up halfway

  • on step 2 — "pick a composition and a mood, never mind the icon for now, you will be able to replace it later";
  • on step 3 — "remember that all of this will still be styled by the neural network, this is not the finished version";
  • on step 4 — "the composition will be preserved, while the icon and the font are improved".

On steps 2–3 you assemble a skeleton — what goes where. The picture is made from it by the neural network on step 4. People give up on the second step deciding "it came out ugly" — and it is supposed to be ugly there.

On the first step, besides the brand name field, there is "Or upload your own logo" — by dragging or picking a file, PNG, JPG or WEBP up to 5 MB.

So the logo generator also works as "polish an existing logo": upload yours → fix the layout → hand it to the network for styling. Nothing on the screen names this route.

What the layout editor has

Add text, the text itself, font, weight, colour, icons, plus save, duplicate and delete for an element. The layout gallery switches between three views: tiles, large, preview.

The styling step

Here you get the usual generation settings: a styling model with its own list, a styling prompt, a "Negative" field, a random-prompt dice and collapsed additional settings.

The additional settings are not translated — they show technical parameter names in Latin script. If you don't know what they are, don't touch them: the defaults are chosen sensibly.

The styling prompt changes only the appearance. The composition and the layout stay as you assembled them on steps 2–3. Don't redo the layout with a prompt — go back to the "Editing" step.

Where to go next

  • Edit a fresh work right after generating it — Generator.
  • Find the work you want to fix — Your works.
  • Learn which formats are available for download — the same page.
Edit on GitHub