I have a number of PNG files with transparency, but for technical reasons I need the expand the total canvas size of the image. I tried using XnConvert but I can not seem to work out how to do the canvas resize with out the new space being created being filed with a solid colour.
SOLUTION
When making some screenshots to phrase my question properly I discovered the solution to this. Sorry to waste your time, but I thought I would still make the thread in case others come here who are equally as much a silly bunny.
This image shows a simple resize image and then resize canvas function... in the resize canvas function I have circled the value that controls transparency for the new space created when you increase the size of a canvas.
Just a quick follow up.... if you are trying to get transparent backgrounds from a Canvas Resize and the source is like a jpg or something you need to add a action to add the alpha channel
This is the same simple reze action as before but before the canvas I have added the "Change Colour Depth" filter and simply ticked "32bit" and set the Alpha to 255.
Finally in your output make sure you have selected a format that supports alpha channels.. like say PNG....
So do I understand it correctly, that it's possible to set the alpha in the Canvas resize action of the Batch convert, but it's not possible in the Canvas resize accessed directly from the main menu? What's the reason for that? It's complicated and unintuitive.
Sorry, I just noticed you're talking about the NConvert, but I'm concerned in XnViewMP.
But your question helped me to solve my issue anyway (the XnViewMP's Batch convert is identical to the NConvert, I think). I just don't understand why the Canvas resize action in the Batch convert is different from the same action accessible directly from the XnViewMP's main menu.
The above posts with screenshots refer to XnConvert [in 2016], NConvert is the equivalent command line software.
Any differences between the current versions of XnConvert and XnView MP Batch convert could in principle be due to different update states, or equally perhaps to a random factor...
