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
when you say 1.2-1.4x speed up does that mean it's 120% or 140% faster or 20% and 40% faster?
Also is it possible to include this in the original repo stylising per frame so it doesn't require a batch but stores the style features and uses them unless you execute a new command with a different style input?
The text was updated successfully, but these errors were encountered:
@Sugarbank I meant 20% and 40% faster but it depends. On a second thought it really sound misleading. I will change the description when I push audio support. Thanks!!
Basically the speedup depends on:
Size of style image vs content image
Number of content images (Total frames of a video).
If style image is smaller you get less speedup whereas when it is bigger you get a higher speedup. The operation of computing artistic images would be nearly 100% faster when style image and content image are of the same size (assuming number of content images is greater than 10).
This is already included in the original repo as well. I just haven't mentioned it. You just need to call testVid.lua instead of test.lua for this to work. I mean:
I just did not want to change the original test.lua because Xun seemed to be busy. So, maintaining a separate file with changes would have made the pull request easier to merge.
when you say 1.2-1.4x speed up does that mean it's 120% or 140% faster or 20% and 40% faster?
Also is it possible to include this in the original repo stylising per frame so it doesn't require a batch but stores the style features and uses them unless you execute a new command with a different style input?
The text was updated successfully, but these errors were encountered: