How Our Face Rating Report Works
The report is designed to explain a photo impression, its limits and a few practical changes. Here is what the system evaluates and what it leaves out.
What is being evaluated?
The submitted image is the evidence. The report considers visible presentation: lighting, framing, expression, image clarity and the impression of visible facial balance. It produces a broad subjective range and observations written in ordinary language.
It does not perform identity matching, measure a person's social value, diagnose health, calculate a verified symmetry percentage or place someone in a population percentile.
How your photo is processed
- Prepare the photo. Your browser checks the image format and size. You confirm that you are an adult submitting your own photo.
- Check the photo. OpenAI checks for one clearly visible, fully clothed adult and rejects unsuitable or unclear photos. These checks are not reliable age or identity verification.
- Create the report. Google processes the photo through Cloudflare to suggest scores and changes you can try. Paid plans also compare estimated audience perspectives. Safety checks apply to the submitted photos and the response.
- Check the result. A report is shown only if it passes the required format and content checks. If a check fails, you receive an error instead of a score.
Current status: Photo analysis and one-time purchases are available.
The visible samples show recorded tool results on generated adult portraits. Limited integration checks do not establish model accuracy or safety across real-world portraits. Broader evaluation remains incomplete.
How are scores displayed?
All ratings and score variations use 0.5-point steps: for example, 6.5 / 10 with a +0.5 variation gives 7.0 / 10. The overall range uses the same steps. A range describes a subjective impression; it does not show how certain the result is. The same image may receive different scores on separate runs.
Audience insights
The $4.99 report gives each photo a base score and estimated variations across seven ethnic / cultural audience perspectives, with a short reason for each. These are AI estimates. The audience labels describe perspectives used for the assessment; they do not identify your ethnicity.
These are AI style simulations, not measured demographic preferences.
Your new look
Choose from 11 appearance settings, or use Surprise me and adjust the combination. Your selected settings and up to five reference photos guide personalized suggestions and one AI-generated portrait. The first photo sets the composition. You can save the resulting image; facial-fullness changes are visual simulations.
How to make a fair photo comparison
Use your own recent photos with similar lighting and framing. Choose a context before comparing them. The detailed report should explain which aspects support that context and which changes are worth trying; it should not claim to know how an entire demographic group will respond.
What the recorded examples show
This review covers the 12 generated adult portraits and their styling previews in our sample reports. The scores below come from the saved reports. The most recent report was recorded on .
In this set, 11 previews scored higher, 1 scored the same and 0 scored lower than their original photo. These are selected examples, not a random sample of users. The counts do not estimate your chance of receiving a higher score.
| Example | Original / 10 | Preview / 10 | Change |
|---|---|---|---|
| Example 01 | 5.0 | 6.0 | +1.0 |
| Example 02 | 5.5 | 6.0 | +0.5 |
| Example 03 | 5.0 | 6.0 | +1.0 |
| Example 04 | 5.5 | 6.0 | +0.5 |
| Example 05 | 6.0 | 6.5 | +0.5 |
| Example 06 | 6.0 | 6.0 | 0.0 |
| Example 07 | 5.5 | 6.0 | +0.5 |
| Example 08 | 5.5 | 6.0 | +0.5 |
| Example 09 | 5.5 | 6.0 | +0.5 |
| Example 10 | 5.5 | 6.0 | +0.5 |
| Example 11 | 5.0 | 6.0 | +1.0 |
| Example 12 | 6.0 | 6.5 | +0.5 |
The original photo and preview are different images. Hair, clothing and other details can change together. A difference in score does not identify which change caused it. These saved results show the kind of feedback the tool returns; they do not prove that a styling change improves real-world attractiveness.
What still needs evaluation
| Question | Current evidence |
|---|---|
| Does the same photo receive a stable score? | These examples do not test repeated runs of an unchanged photo under the same settings. We have not published a repeatability estimate. |
| What happens when only the light or crop changes? | The styling examples change several details at once. They do not isolate lighting, camera position or crop. |
| Do audience scores match real people's preferences? | No human audience panel was used for these examples. The scores are model simulations, not survey results. |
| How often does screening accept or reject the wrong photo? | Successful sample reports do not measure screening error rates. Broader evaluation on varied, consented adult photos remains incomplete. |
Evaluation priorities
- Safety screening and rejection of unsuitable or ambiguous photos.
- Consistency when the same consented adult photo is processed again.
- Whether a change in lighting or crop changes the advice sensibly.
- Whether suggestions remain respectful, specific and useful across varied adult portraits.
- Failure handling, privacy controls and paid-report delivery.
There is no published accuracy percentage yet. We will describe the evaluation set, method and observed limits before claiming one.
Before you upload
Read the privacy notice for photo handling and the report access limits before choosing a plan.
Maintained by MyFaceRating · Reviewed