Don't store conversions that make no difference #2200
-
Given a conversion like this:
if an image is below this size limit, nothing needs doing, however, ML saves a new version anyway. A nice optimisation would be to somehow detect null operations like this, and serve the original image instead, saving both storage space and processing time. The hard part is detecting this! I guess ideally feedback from the manipulation would say whether it did anything or not. An alternative might be to process it, then take a hash of the result and see if it's the same as the original. That would still incur processing time, but would still save io bandwidth and storage space. |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment
-
I think that total processing power needed for detecting this will far outweigh the benefits for that one upload that might be identical. I also think storage space is cheap. So most users won't need an optimisation like this and I prefer to keep to code very simple. I might change my stance on this if this would be a problem for more users. Thanks for suggestion this though. |
Beta Was this translation helpful? Give feedback.
I think that total processing power needed for detecting this will far outweigh the benefits for that one upload that might be identical. I also think storage space is cheap.
So most users won't need an optimisation like this and I prefer to keep to code very simple.
I might change my stance on this if this would be a problem for more users.
Thanks for suggestion this though.