ML SUPER RESOLUTION INCREASES CANVAS SIZE

Is something not working like it should? Let us know.
User avatar

2021-06-04 15:46:07

I have made a document 1920x1080 pixels @ 300 dpi

I have imported 20 photos, scaled them up and positioned them accordingly to fit the canvas and added text to each one

I have used ML super resolution on 3 or 4 of them to improve the quality.

I now see that my canvas is 32000x18000 pixels! 16.6 times bigger!

This means that when I export the text/photos and transfer them to After Effects, they are all enormous!

I wondered why my iMac was struggling to cope

What is the solution to this issue?

J
User avatar

2021-06-05 01:35:29

a way round until this bug is fixed is to import the photo as a new tab (done by dragging a photo from a finder window to the Pixelmator app on the dock), applying ML super resolution, then copying and pasting into the currently opened document. if you already have a photo that you want to apply ML super resolution, copy or cut it, go to the view menu > show all tabs (or press โ‡งโŒ˜\) hit the big plus (+) button and choose "clipboard", apply ML super resolution, copy, then paste into your opened document!
User avatar

2021-06-05 02:04:53

Thanks.

In the end I went to the main document and selected IMAGE SIZE in the drop down panel and set it back to 1920x1080 and that seemed to do the job.

The canvas size changed with it

J
User avatar

2021-06-05 02:22:50

Great! lets hope your Mac survives it ๐Ÿ˜‚
User avatar

2021-06-08 10:32:43

by JARROD 2021-06-04 12:46:07 I have made a document 1920x1080 pixels @ 300 dpi

I have imported 20 photos, scaled them up and positioned them accordingly to fit the canvas and added text to each one

I have used ML super resolution on 3 or 4 of them to improve the quality.

I now see that my canvas is 32000x18000 pixels! 16.6 times bigger!
ML Super Resolution is, in essence, a scaling algorithm so this behavior is not a bug but rather a part of the tool's functionality. The ML Super Resolution command from the Image menu, in particular, scales up images by 300%. You probably ran your images through it a few times before the file size grew to 32000 (when applied once, a 1920x1080 px image should end up at 5760x3240 px). I do get that one may want to use ML Super Resolution just to remove noise or compression artifacts but in such cases, the image would, indeed, need to be scaled back down to the prefered size.