An addendum to Supersharp AI
AnsweredFrom the idea Supersharp AI is not bad. It is disappointing that the tool was released unfinished. Skylum wanted to catch up with the competition too much: Topaz, On1 etc.
It would have made sense to concentrate only on one extension and to bring this ready on the market.
Well then, all this has already been written here in the forum several times. We are all surely waiting for an explanation from Skylum for this step, after all Skylum did not hesitate for a second to immediately change the advertising text on their homepage to Supersharp AI after the shitstorm started. But too late, because the original text has already been quoted here several times and can be seen in several YouTube videos.
So pants down, Skylum here you just fell too much in love with your advertising texts instead of delivering!
But I would like to point out something else. What stands out is that Supersharp AI tries to sharpen the entire image. Of course, that's not useful, most of the time anyway. Because a cropped portrait, is supposed to stay cropped with a nice blurred background.
So Skylum doesn't apply a mask to the main object at first. But you can set the mask manually using the Mask AI tool. Only in this combination Supersharp AI makes sense.
One can argue about whether Supersharp AI doesn't just automatically form a mask with the possibility of manual intervention. I think it was too difficult for the Skylum engineers to implement this. Combining it with Mask AI would probably have been less possible then.
Everyone can form an opinion on what would be better. I was initially averse to Skylum's implementation. But I also have to admit by now that it's not bad at all. Topaz doesn't always find the right masking here either, and you have to touch it up manually.
Of the new extensions, Supersharp can certainly be one of the better ones, if it is supplemented by the two missing modes unfocus and universal, as it was previously touted.
-
Hi Member,
Thank you for the detailed feedback!
I forwarded it to the relevant team
Please sign in to leave a comment.
Comments
1 comment