
Three prompts with the same style wording produced three different kinds of file. The notebook has 21 paths and six distinct fill values, the leaf has one path and one fill, and the parcel includes a gradient. The steps below take those downloads through recoloring and placement on a fictional notebook page.
The example below uses a fictional Fieldnote notebook. It is a layout demonstration, not a real product offer or a conversion-rate experiment.
Each subject was generated separately in the SVG creator dashboard on September 14, 2026. Every run used Classic, one SVG and a transparent background, with a displayed cost of two credits. The three initial generations used six credits in total. No image reference was attached.
This was the notebook prompt:
One simple outline icon of an open notebook for a fictional stationery shop
product badge. Square composition with generous padding, rounded line ends,
dark navy #18345F only, no lettering, no filled background, no gradients or
shadows. Transparent background.
For the other two runs, only the subject phrase changed: “an open notebook” became “a single leaf” and “a closed shipping parcel.” The style wording stayed the same.
Download the original notebook, leaf and parcel before editing them. The leaf dashboard screenshot and parcel dashboard screenshot show those separate results.
| Original file | Bytes | Paths | Distinct fill values |
|---|---|---|---|
| Notebook | 23,180 | 21 | 6 |
| Leaf | 3,726 | 1 | 1 |
| Parcel | 4,923 | 2 | 2, including a gradient reference |
All three contain paths and no image elements. They share a navy outline direction, but that does not make their construction, exact palette or apparent line weight identical. The notebook has considerably more internal detail than the other two.
The parcel's gradient is an example of a prompt constraint that was not met. The notebook's multiple fills also differ from the requested single exact navy. Inspect the exported markup; a small preview can make several close shades look like one color.
The measurement file records actual fills, bytes, path counts and hashes. These are three selected outputs, not a success-rate estimate for the generator.
The manual files replace each fill attribute with #18345F and set proportional fitting to xMidYMid meet. Original path coordinates are preserved. This replaces the parcel's painted gradient with a solid fill, but it does not redraw the icons into a perfectly matched set.

Use the manual notebook, manual leaf and manual parcel to reproduce the page. The notebook still looks denser. If the actual interface needs uniform visual weight, simplify it or choose a more consistent icon source before shipping that interface.
Open the working fictional product page. Its cards display the icons in 52-by-52-pixel boxes with proportional fitting. The labels are “Notebook details,” “Materials guide” and “Delivery options.”

Those labels describe information areas without claiming free shipping, certified materials or a product guarantee. A leaf drawing does not prove a sustainability claim. Use the actual product information when writing a real shop's labels.
The cover is a manually assembled HTML mockup using the earlier Fieldnote mark. It is not a product photo. The expandable asset note explains which parts came from generation and which were prepared manually.
Because the adjacent text explains each card, the badge images use empty alt text. If an icon becomes a standalone control, give that control a name that describes its action. Do not turn a visual badge into an unlabeled button.
Check the icons beside the real text, at normal zoom and on the intended backgrounds. Compare their apparent size and detail; equal CSS boxes do not guarantee equal visual weight. Keep enough space for labels to wrap on smaller screens.
The currentColor tutorial covers icons that must follow a control's CSS color. The logo variants guide prepares a mark for different backgrounds. For a standard navigation action, the Font Awesome comparison explains why a stock icon may be a better starting point.
These assets demonstrate a reproducible preparation workflow. They do not establish a revenue lift, faster checkout or increased trust. A real store would need its own product review, usability checks and appropriately designed measurement to answer those questions.