Our rigorous testing methodology:
Transparency creates trust
When evaluating generative AI tools in ethically sensitive niches, the highest degree of objectivity and transparency is essential. Our mission is to introduce you not only to the most powerful platforms, but above all to the safest and most ethically responsible ones.
Each platform listed here undergoes a standardised, multi-week practical test by our team of experts. Our overall rating (which you can see in the tool tiles) is made up of the weighted sum of the following five core criteria:
Our 5 core test criteria for AI generators
Criterion 1: Image quality and model stability (expertise)
This criterion evaluates the technical performance of the core model:
Prompt adherence (prompt drilling): How exactly does the tool convert complex, multi-part text entries? We use Identical test prompts across all platforms.
Realism and consistency: Assessment of anatomical correctness, attention to detail (hands, eyes) and the ability to create consistent characters across different generations.
Hallucinations & artefacts: Frequency and severity of technical errors (e.g. missing limbs, faulty textures).
Criterion 2: Cost transparency and performance (trustworthiness)
Monetisation through credits or subscriptions must be clear and fair.
Credit ratio: How many credits does a high-quality, high-resolution generation cost? We calculate the cost per image in EUR.
Transparency: Are subscription traps or non-transparent renewals used?
Performance per euro: How quickly does the tool generate images at the maximum quality setting?
Criterion 3: Ethical strength and safety filter (Safety & Trustworthiness)
As this is sensitive content, the responsible handling of misuse is a key evaluation criterion.
NCDI ban (Non-Consensual Deep Imagery): If the creation of content without express consent in the terms of use explicitly and unequivocally prohibited?
Filter strength: How effectively does the system prevent the generation of illegal or extremely abusive content (even in the case of targeted circumvention attempts by the tester)?
Reporting procedure: Is there a clear, easily accessible procedure for reporting abuse?
Criterion 4: Data protection and data storage (trustworthiness)
Transparency about the storage of user data and uploaded images is essential for credibility.
Deletion periods: How long are uploaded source images or generated results stored? Platforms with short or clear cancellation periods receive a higher rating.
Anonymity/payment: Are anonymous payment or usage methods possible?
Encryption: Is sensitive data (e.g. uploaded faces) stored in encrypted form?
Criterion 5: User-friendliness (UX) and features (experience)
This criterion evaluates everyday use and the available functions.
Intuitiveness: How easy is it for new users (beginners) to familiarise themselves with the system?
Variety of features: Does the tool offer advanced functions (inpainting, face-swap, video generation, batch processing)?
Customer support: Quality and speed of response to technical enquiries during the test period.
Our commitment to objectivity
Our ratings are quarterly updated to reflect the rapid development of AI models. The results of our tests influence our recommendations and the Scorecards in the tool tiles. Any affiliate relationship has no influence on the awarding of stars or the placement in the ranking.
Authorised: Julian Weber (Certified Prompt Engineer)
