Why pixel processor couldn't be relative to input ?

Yeah, I am doing exactly what you suggested .  Good to know I am at right puth. Thanks again

I am glad this was helpful!

You may also be interested in the $getsize float2 variable, which returns the size of the output image and helps you calculate the exact image proportion of a pixel:

float2(1,1) / $getsize. This lets you be precise and operate on single pixels.

Use a Get float2 node to access this variable.

Best regards.

Thanks a lot Luca

I am just want to make all my pixel processing be  working in actual pixels . Like shifting or scaling  exactly one pixel,   not just some 0.003   for better pixel to pixel accuracy .

So I come  to what Luca suggested  and all my nodes now have green dot to output size in pixels  and input green dot #input size.    It helps to eliminate that suddent  blur  that something happens because of missing primary input or pixel processor .

Parsing of the inputs using the Sample nodes is normalized (0..1) so you usually don't need to know the resolution of the input node. What is it you need to do exactly?

Could you elleborate please.  Sorry, I havn't understood.  For example I need a pixel processor to keep having same  output size in pixels  as its first input.  What should be the function graph  in  "output size"?

Isn't that something you'd sort out within your functions?  Just grab $pos (which is a Float2 from mind) and then apply your offset before the rest of the operations.