alpha channel lost when resizing 8bit png

[adamhj]

when resizing a 8 bit png image file with alpha channel with most arithmetic (except nearest neighbour and bilinear), the image is converted to an 24 bit true colour picture and the alpha channel is lost

btw, is there any way to convert a 8 bit png image to 32 bit(keeping alpha channel) in batch processing?

[xnview]

Could you send me a sample?

[MaxSt]

Yes, the first two: nearest and bilinear work fine.
But more advanced ones, like Lancsos, lose alpha channel.

Here's some examples:

http://www.google.com/logos/2011/verne-hp.png
http://www.google.com/logos/2011/bunsen11-hp-sprite.png

[xnview]

Oh ok, for such picture, you must convert them in 32bits before. Other method than nearest & bilinear build always a 24bits image

[adamhj]

so is there anyway to convert them in 32bits in batch processing tool? i can only find an "add alpha channel" method but it only replace alpha channel

[xnview]

so is there anyway to convert them in 32bits in batch processing tool? i can only find an "add alpha channel" method but it only replace alpha channel

I'll add it in XnView. You have it in XnConvert