Hello, I’ve noticed a difference in output when editing images with my application compared to Google Studio. In my app for personal use, I edited a picture of a boat and requested the removal of specific elements. While the elements were removed, the boat’s appearance also changed. However, in Google AI Studio, the edits are precise and the boat remains exactly as I wanted. I don’t use any prompts or system instructions internally in my app, I give the instructions directly through the UI interface, do I need to set system instructions for the gemini-2.0-flash-exp-image-generation model?
1 Like
Hi @Bassam_El-Ashkar , Welcome to the forum.
Ideally, the models performance should be consistent whether accessed via the API or AI Studio.
Ensure that the same hyperparameters are configured in both the API and AI Studio.
Thank you.
1 Like
same hyperparameters? What’s that? can you share a screenshot where to find this ‘hyperparameters’ in Google AI studio?
1 Like
