A: We have a custom trained model. You can see the image quality of background generation (just try describe image and you will feel the difference between say SD 1.5 or SD 2 and this. With the right prompts it's same or better than midjourney as per some power users. We are exploring Dall-e-3. It's also really powerful but the latency is high and API doesn't support inpainting. So, we rather host and refine the model than be a wrapper on an API. Also, again Dalle doesn't support inpainting, so you cannot in midjourney and dall3 put your product in the AI image.

We will continue exploring different models as the space evolves and bring the best options for our clients.