STYLE GUIDE · Pizu AI
A cartoon avatar, a family illustration, and a pet portrait need different things from the same effect. An avatar has to read at a glance. A family picture has to keep people distinct. A pet portrait may depend on one ear shape or a patch of white fur.
Choose the details that make the subject recognizable before choosing how much to simplify them. This guide compares three visual directions, gives you a prompt for each, and shows what to inspect before you accept the result.
Choose a style for the size and use of the image
| Direction | Visual treatment | A useful starting point for | Main tradeoff |
|---|---|---|---|
| Flat 2D | Solid color areas, clear outlines, few small details. | Profile pictures, simple pet icons, small social graphics. | Fine markings and subtle facial features may disappear. |
| Soft 3D character | Rounded forms, smooth shading, visible volume. | Playful character portraits and greeting-card concepts. | Changes to the eyes, cheeks, or head shape can weaken likeness. |
| Ink illustration | Visible linework with limited color and optional paper texture. | Portrait prints, illustrated memories, editorial-style images. | Fine lines can become hard to read when the image is reduced. |
These are visual descriptions, not fixed presets or promises about an output. A generated flat-color image is not automatically a vector file. Raster images contain pixels; vectors describe shapes mathematically. Check the actual file format before planning to scale it for print. Adobe’s raster and vector explanation.
If you are undecided, start with the intended display size. A detailed ink portrait may look appealing full-screen but lose its expression as a small circular avatar. Conversely, a very simple icon may feel sparse in a large framed print.
Choose a reference that reveals what you want to preserve
Use the source photo to answer three questions: can you see the face or main form, can you distinguish the important markings, and is anything essential hidden?
- For a person: look for visible eyes, a clear face outline, and the hairstyle or glasses you want retained. A filter cannot show an obscured feature without inventing it.
- For a pet: include the ears, muzzle, and distinctive fur patches. If a white chest patch matters, choose a photo where it is actually visible.
- For a group: prefer a photo where faces do not overlap. Decide whether the full scene matters or whether a closer crop would make the people easier to distinguish.
Crop away irrelevant space before uploading if you already know the framing you want. Keep your original file: the crop is a preparation step, not a replacement for the original photograph.
Try three style briefs for the same pet portrait
Imagine a reference photo of a seated black dog with floppy ears and a white chest patch. The examples below keep those three identifying details constant while changing the rendering. They are editable prompt examples, not a gallery of tested results.
Flat 2D: prioritize the silhouette
Turn the uploaded dog photo into a flat 2D illustration. Retain the floppy ears, white chest patch, and seated pose. Use bold clean outlines, solid black and white shapes, and a plain pale blue background. Keep the whole head and both ears inside the frame. Simplify fine fur detail.
Check: reduce the image to its intended avatar size. Can you still distinguish the ears and chest patch? If the patch disappears, request a clearer white shape rather than adding more fur texture.
Soft 3D: add volume without replacing the subject
Restyle the uploaded dog photo as a softly shaded 3D cartoon character. Retain the dog's floppy ears, white chest patch, muzzle proportions, and seated pose. Use rounded forms, soft studio lighting, and a plain pale blue background. Keep the expression gentle and the eyes close to their original relative size.
Check: compare the muzzle, eye spacing, and ear length with the photo. A technically polished image may still look like a different dog. If it does, ask for less exaggeration before changing the lighting or background.
Ink illustration: make linework the main treatment
Turn the uploaded dog photo into an ink illustration with restrained black and pale blue color. Retain the floppy ears, white chest patch, and seated pose. Use confident contour lines and sparse hatching to suggest fur. Keep the face clearly readable, with a mostly plain background and subtle paper texture.
Check: look for dark hatching that fills the white patch or obscures the eyes. Request lighter shading in those areas if the defining features become difficult to read.
For people and groups, write a different preservation brief
The pet prompts do not translate to portraits simply by replacing “dog” with “person.” Identify what matters in that particular photograph.
For a portrait, that might be a short fringe, round glasses, a broad smile, and the angle of the face. For a group, it may be the number of people, left-to-right order, and clothing colors. Name a few visible details rather than asking to preserve “everything perfectly.”
Restyle this group photo as a clean ink-and-color illustration. Keep the same number of people in the same left-to-right order. Retain their clothing colors, visible glasses, and main hairstyles. Simplify the background, keep the faces distinct, and use restrained shading.
Count the people in the result. Then compare each face and check hands, arms, and overlapping shoulders. Keeping the group arrangement does not guarantee that every person's likeness survived the transformation.
Diagnose the mismatch before generating again
| Problem | First adjustment to try |
|---|---|
| The result is attractive but no longer recognizable. | Reduce the requested exaggeration. Name the two or three features that changed most. |
| The cartoon looks busy at small size. | Remove background details and ask for larger, simpler color areas. |
| The result mixes flat shapes and glossy 3D shading. | Choose one rendering direction and remove the competing instructions. |
| A group member is missing or merged with another. | Try a reference with clearer separation. Restate the count and arrangement, then inspect again. |
| The style barely changes. | Replace vague wording such as “more cartoon” with the actual treatment: solid fills, clear outlines, rounded volume, or visible ink lines. |
Use the prompt as a style brief in Pizu AI
Open Photo to Cartoon, upload your reference, and review the prefilled prompt. Replace the style instructions with the direction you want to test instead of appending a contradictory style to them. Generate, compare the defining features, and download the version you want to keep.
If you are inventing a character instead of transforming a photograph, use the page's text option. That is a character-design task: describe appearance, pose, expression, and setting without claiming resemblance to a person or pet you have not supplied.
A useful final check is to view the image twice: once beside the reference, and once at the size where it will be used. The first checks identity. The second checks whether the style actually serves the job.