Closed mabushey closed 7 months ago
This is not an intended use case of the metric, and I can't think of a simple trick to achieve what you want with the algorithm. Sorry.
Note that this project is more or less a port of the C implementation of the SSIMULACRA2 metric. If you have questions related to the metric or the algorithm itself, it might make more sense to ask a question over there.
Thank you for answering my question!
To clarify a bit more, I don't think there's a good way to achieve this with any metric, without having some "original" to compare to. Video metrics are typically used to compute loss when compressing a source. In the case of having two different sources without an original that they were encoded from, the only real way to determine the better one is with your eyes.
This is awesome, thank you. One of my applications is slightly different, I would like to compare two different sources and determine which is better.
Is there a way to determine which image is the better one? Right now it just assumes the first is. This may be out of scope for what this algo does. Thanks for the help!