Code: Select all
convert.exe input.tga -resize 128x128! output.tga
Code: Select all
convert.exe input.tga -alpha off -resize 128x128! -alpha on output.tga
Code: Select all
convert.exe input.tga -resize 128x128! output.tga
Code: Select all
convert.exe input.tga -alpha off -resize 128x128! -alpha on output.tga
Code: Select all
%IM%convert ^
-size 1000x600 -background rgba(100%%,100%%,100%%,0) ^
label:snibgo ^
-trim +repage ^
s.png
%IM%convert ^
s.png ^
-resize 25%% ^
s1.png
%IM%convert ^
s.png ^
-alpha off ^
-resize 25%% ^
-alpha on ^
s2.png
%IM%compare -metric RMSE s1.png s2.png NULL:
Correct.Oogst wrote:I thought "-alpha off" was a parameter, but it is a command, isn't? So do I understand it correctly that if it says "-alpha off -resize 25%% -alpha on", this actually means three commands in a row: it turns the alpha off, then it resizes, and then it turns alpha back on. Do I understand that correctly?
I think you mean that SplashScreen_Original.tga has an alpha channel, but all values are 255 (ie the maximum, so full opacity).Oogst wrote:As for the sharpness: my testing image actually has an all-white alpha.
Yes. The image has no transparency, so ignoring it or not should make no difference to the result. But there is a difference.Oogst wrote:Shouldn't that give the exact same result in both cases?
There is a difference. Overall the difference is RMSE 0.3%, which is very small, but concentrated in certain areas. Where lightish pixels are surrounded by dark pixels, we can see they become lighter in the "-alpha off" version.Oogst wrote:The difference is difficult to see, so I have put the original image and the two resized versions online for download.
Code: Select all
convert ^
SplashScreen_WithAlphaOnOff.tga ^
SplashScreen_WithoutAlphaOnOff.tga ^
-compose Difference -composite -auto-level ^
ssDiff.png
Code: Select all
convert.exe input.tga -resize 1024x512! output.tga
Code: Select all
convert.exe input.tga -alpha off -resize 1024x512! -alpha on output.tga