Editable Layer Export for Listing Images

A Graflio image arrives as a flat 1024px square, like every AI image does. What happens next is the feature: open it in the Studio, run Make Text Editable for half a credit, and every headline, caption, badge word and unit of measurement lifts off the picture into a real text object over a repainted, text-free background. From there retyping is free. This page is about exactly which parts become objects, which parts stay pixels, and what each export keeps.

What a layer is here

The image model returns one flattened raster. Opening it in the Studio costs nothing and changes nothing: the picture becomes the canvas background of a 1080px square document, and you can already add your own text, shapes and pointer lines on top of it.

Make Text Editable is the step that changes the file's nature. It sends the flattened canvas to Ideogram's layerize-text model, which returns two things: a background plate with every letter erased and repainted, and one block per piece of text carrying its position, bounding box, font size, colour, alignment, line height and the name of the typeface it was set in. Graflio rebuilds each block as a text object, maps the returned font name to a family, loads that face, and places it on the plate at the coordinates it came from. The result looks identical to the image you started with and behaves like a document.

One call handles the whole image, so the cost is predictable: 0.5 credit, about 17 cents at pack pricing. The prompt behind it explicitly describes a text-dense infographic, because without that instruction the model tends to lift the headline and leave the small print behind. On a busy image it can still miss a block; clicking the button again captures the rest.

Editable, flattened, and the line between them

The honest summary: text becomes text, and everything else stays a picture. That is not a limitation of the export format, it is what the separation step actually does. A tool that told you the arrows and icons were vector objects after a single AI pass would be describing something else.

Editing text once it is text

Text objects take the usual handling — retype, move, rotate, resize, recolour, change typeface or size. Fonts come from the Google Fonts library plus any font files you have uploaded to your brand kit, so a licensed brand face is available rather than approximated.

The part worth knowing about is the layer panel. Objects carry a role, and the panel groups them by role: headline, subheadline, callout titles, callout text, numbers, cards, badge text, badge shapes, closeup labels, footer, lines, measurements, and free text and shapes you added. Picking a group selects every object in it and applies a font, size or colour change to all of them at once. On a six-image set this is the difference between fifteen minutes and two hours, because listing images fail as a set, not one at a time.

Two AI actions reach the parts the text pass cannot: a highlight-and-prompt spot edit repaints a marked region for 0.5 credit, and Translate redraws the whole image in another language for 1 credit, keeping layout, colours and framing and changing only the words. Thirteen languages are offered in the picker. Undo and redo cover the manual edits, and the canvas autosaves to the product, so a design reopens as a design rather than as a JPEG.

What each export keeps

The PSD is built from the canvas: a background layer, then one raster layer per object, named in stacking order after its role and the first 28 characters of its text, so the panel reads 02 headline · Fits any 12 mm hose rather than Layer 14. It is separable and maskable, not retypeable. The SVG is the one that keeps live type. If you archive one file per image, archive the SVG: vector text can always be rasterised later, and rasterised text can never be un-rasterised.

On resolution, note the difference between the download and the export. The generated file in the gallery is 1024px square, which clears Amazon's stated 1,000px floor for zoom but sits under the 1,600px Amazon recommends. The Studio exports at 2160px, comfortably above both. If an image matters, export it rather than saving the gallery file.

Who else does this, and what is actually different

Layered output is no longer rare. Pixii ships a Convert to Layers editor and DesignLumo exports layered PSD, so a page claiming Graflio is the only tool with editable layers would be false and you would catch it in one search. The distinction worth arguing is narrower and more useful: what the layout was derived from, and what an edit costs after the first one. Graflio's layout comes from a researched brief built on your live listing, its reviews and the category best sellers, and every text edit after the 0.5-credit separation is free and instant, because it is a text box.

Against prompt-only generators the gap is wider and simpler. There, changing four words means regenerating the image, and the model moves things you did not ask it to move. The sixth revision is where that architecture stops being cheap.

When you do not need any of this

If the set is right the first time, export the PNG and stop. Layer separation earns its half credit when copy is likely to change: a spec correction, a legal qualifier, a price on the image, a second marketplace in another language, a claim your category reviewer objects to. If none of those apply to your product, do not spend the credit. And if the change you need is to the photograph rather than to the words, layers do not help at all — that is a spot edit, or a better source photo.

Common questions

What does making an image editable cost?

0.5 credit per image, charged once. After that, retyping, restyling, moving and recolouring text costs nothing. Failed actions are refunded automatically.

Can I edit the product photo or the background?

Not as objects. The photograph, scene, icons and graphic elements stay in one flattened plate. You can repaint a marked region with a highlight-and-prompt spot edit for 0.5 credit, or cover an element with your own shape, but you cannot detach the product and move it.

Which export keeps the text editable?

SVG. It writes the 2160px canvas with text still as text elements and references the webfont stylesheet, so Illustrator opens it with live type. The PSD gives you one named raster layer per object, which is separable but not retypeable. The PNG is flat.

Does the layout have to be regenerated after an edit?

No, and that is the point. Text edits are local: the words change, nothing else moves, and no AI call is involved. Regeneration is only needed if you want a different image altogether.

Is Graflio the only tool with layered output?

No. Pixii has a Convert to Layers editor and DesignLumo exports layered PSD. What differs is where the layout comes from and how much an edit costs once the file is open.