A clear photo, a short instruction, and a suitable image shape are enough to start. This walkthrough uses the South Park Character Creator on this site, which turns one photo into one generated South Park-style avatar at a time. You do not need to write a long prompt or choose individual eyes, hair, and clothing from menus. The screenshots and results below were captured on our live generator on October 17, 2026.
1. Prepare a photo the generator can read
Choose a picture with one clearly visible subject. Keep the face, hairstyle, and any accessory you want to preserve in the frame. A bright, front-facing portrait gives you a better starting point than a dark group photo or a screenshot of a screenshot.
The upload accepts JPG, PNG, or WebP files up to 10 MB. If your phone provides a HEIC file, export a JPG or PNG copy before uploading; changing the filename alone does not convert it. Crop out other people or identifying background details using your photo editor.
For lighting, framing, and likeness problems, use the separate photo preparation and troubleshooting guide.
2. Upload your picture
Open the homepage generator. Make sure From a photo is selected, then select Upload a photo and choose the prepared file. Check that the correct picture appears; use Change photo if you selected the wrong one. You can also choose Try a sample photo to explore the flow.
The portrait in this screenshot is a synthetic photo made for our demo, not a real person.
3. Add one or two optional details
The instruction field allows up to 200 characters. Use it for a visible detail you want to keep or change, such as glasses or a jacket color.
These are starting points to adapt, not promises of an exact result:
| What you want | Example wording |
|---|---|
| Keep a recognizable accessory | Keep my black rectangular glasses and navy zip hoodie. |
| Keep a hairstyle and outfit | Keep the long wavy auburn hair, freckles, and mustard sweater. |
| Change the setting | Keep the curly hair and red puffer jacket; snowy background. |
Choose the most important request first. Conflicting instructions make it harder to tell which change helped. The tool does not offer individual facial-feature sliders or an editable layer file.
No photo? Switch to Describe an OC to create an original character from text alone. The OC guide covers that mode.
4. Choose the image shape
Use 1:1 for a square profile picture, 3:2 for a wider composition, or 2:3 for a taller portrait. Shape controls the canvas, not a guaranteed crop of every subject. Leave some space around the head if the result will later appear inside a circle.
The generator uses Gemini 2.5 Flash Image at a 1K tier. A square result is 1024 × 1024; other aspect ratios have different pixel dimensions.
5. Generate once
Select Generate Now once and keep the page open while it works. In our tests, a result took about 20 seconds; timing depends on the request and service availability.
Your first image as a guest is free and does not need an account. After that, sign in and check the allowance shown beside the tool. Each completed image uses 5 credits, including a new image made with Generate again. Allowances are limited, not unlimited; see pricing for the current plans, renewal period, and expiry rules.
6. Review before trying again
Look at the face, hair, accessories, and edges of the image. If one detail is wrong, change that detail in the instruction or try a clearer source photo. Regeneration can also change details that were already correct, so download a result you want to keep before replacing it.
Do not repeatedly submit a failed request. Follow the message shown: an unsupported file needs a different file, an allowance limit requires waiting or checking your account, and a temporary service error calls for trying later. If credits were charged for a generation that failed, they are returned.
Real before-and-after results
Each pair below is the unedited first result for the instruction shown, generated October 17, 2026, at 1:1. The inputs are synthetic photos made for our demo.
| Photo and instruction | Result |
|---|---|
 "Keep my black rectangular glasses and navy zip hoodie." |  |
 "Keep the long wavy auburn hair, freckles, and mustard sweater." |  |
 "Keep the golden fur, white chest, and red collar." |  |
What we noticed: glasses, hair shape, clothing colors, and the dog's collar carried over in every case. Fine texture such as knit patterns and individual curls was simplified into flat shapes, which is expected for the style. Likeness varies between runs.
7. Download and check the saved file
When your image is ready, select Download. Review the watermark setting before saving. Open the downloaded file to confirm that it is the generated result and that it matches your choice.
Save your own copy while the result is available; do not treat an open preview or a temporary image link as permanent storage. If the download fails but the preview remains, try downloading again before paying for another generation.
8. Prepare the profile picture
Keep the original download, then make a cropped copy for your social profile. Preview the crop at a small size: hair, eyes, and one strong clothing color usually matter more than tiny background details.
The PFP guide explains circular crops and coordinated avatars. If you want to choose every component yourself instead, compare Picrew and AI before starting a second workflow.
Author
