You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Your model notices that NGC7714 is a spiral galaxy. That's cool but actually, it is also a merging galaxy. Your generated images of the galaxy look artistic instead of scientific. Can your model generate real .fit files for science or do you hope to generate artistic images only for the public?
The text was updated successfully, but these errors were encountered:
Your model notices that NGC7714 is a spiral galaxy. That's cool but actually, it is also a merging galaxy. Your generated images of the galaxy look artistic instead of scientific. Can your model generate real .fit files for science or do you hope to generate artistic images only for the public?
Thank you for your insightful question. It indeed aligns with one of our future optimization directions.
For the generation of professional observational images (especially in the aspect of galaxy morphology), we need to further fine-tune our diffusion model, similar to previous work where the diffusion model was used to construct gravitational lenses.
However, as our main focus lies in the variable star field, we are more inclined towards enabling the model to understand image features and then generate code for drawing (like using Python to plot light curves) or through tool learning to use existing astronomical tools (for instance, there are deep learning methods that can generate spectra based on input parameters). In these cases, we believe it can generate more precise and scientific images.
Your model notices that NGC7714 is a spiral galaxy. That's cool but actually, it is also a merging galaxy. Your generated images of the galaxy look artistic instead of scientific. Can your model generate real .fit files for science or do you hope to generate artistic images only for the public?
The text was updated successfully, but these errors were encountered: