Image & Character Model Creation FAQ
1. What makes a good base image for character creation?
For the best results when creating an AI character model, start with a clear, high-quality image where the face is well-lit and looking directly at the camera or nearly so. Avoid photos with sunglasses, heavy shadows, or obstructions over the face. Both AI-generated images and real photographs can work well as a base.
2. How long does it take to train a model?
Training your unique AI character model on KreatorFlow is impressively fast! Typically, the process completes within a few minutes. Once trained, your model is ready immediately for you to start generating images and video content.
3. Why did my model creation fail?
Model creation can occasionally fail due to a few reasons. Common ones include the uploaded base image not meeting quality guidelines (e.g., too small, face unclear, multiple faces detected) or exceeding file size limits. Less frequently, a temporary technical issue on our platform might occur. If you encounter persistent failures, please check your image against our recommendations, and if the problem continues, don't hesitate to contact our support team with details so we can assist you.
4. Can I edit a model after it's been created?
Currently, core model retraining or direct editing of a trained character model's base characteristics isn't supported after the initial training. However, you have extensive control over how the model appears in generated content through prompting, pose selection, and generation settings. If you need a fundamentally different base look, creating a new model with a different base image is recommended.
5. Why is my model not looking like the uploaded photo?
Achieving high likeness is key! If your generated character model doesn't closely resemble the uploaded base photo, ensure the base image was clear, front-facing, and well-lit without obstructions. Our AI strives for accuracy, but the quality of the input image is crucial. Trying a different, higher-quality base photo often resolves this. You can also adjust "Face Strength" settings during generation for finer control (where available).