Upscale any video of any resolution to 4K with AI. (Get started for free)

"How can I fix new Topaz Video AI faces that appear strange or exaggerated?"

**Compression artifacts**: Faces appear distorted or exaggerated due to compression artifacts, which are errors introduced during video compression that can cause facial features to appear unnatural.

**Proteus model limitations**: The Proteus model, a deep learning-based super-resolution model, can sometimes produce unrealistic or distorted faces, especially when used with low-quality input videos.

**Manual processing required**: Manual processing and previewing at a lower resolution (e.g., 150x or 125x) can help identify and correct facial distortions before upscaling to the final resolution.

**Model limitations**: AI models like Iris, designed for face enhancement, may not always produce perfect results, especially with low-quality or noisy footage.

**Human sensitivity to faces**: As humans, we are uniquely sensitive to seeing faces that look slightly unnatural, which is why facial distortions can be particularly noticeable.

**Face detection and alignment**: Accurate face detection and alignment are crucial for face enhancement models like Iris, as misalignment can cause facial distortions.

**Noise and compression**: Noise and compression in the original video can lead to facial distortions and artifacts, even with advanced AI models like Iris.

**Context-aware processing**: Context-aware processing, which takes into account the surrounding area of the face, can help improve facial enhancement and reduce distortions.

**Multiple model combinations**: Using a combination of multiple AI models, such as Proteus and Iris, can help reduce facial distortions and produce more natural-looking faces.

**Model updates and improvements**: Regular updates to AI models like Iris can improve their performance and reduce facial distortions, as seen in the latest Topaz Video AI v33 release.

Upscale any video of any resolution to 4K with AI. (Get started for free)