Potential roundig bug with procedural texture filter

[NotMyFault]

Hi,

in RGB color mode, document size 256x256, procedural texture filter gets wrong value for x=0:

The first column (x=0) gets value 0.5 / 127 (measured) instead of 0 (expected).

1.10.1.1142 (retail)

Update 1:

Seems to affect all color modes (GREY, LAB, CMYK) and all bit length /8 /16 /32

 

Update 2:

This might be by design that it is required to shift x by 1/2. If yes, this should be documented as it potentially breaks procedural filters potentially by rounding issues.

Question

So can you please confirm or disprove the following statement:

• In procedural texture filters, the pixel within the document of size w are counted
• x-axis: x from 0.5 to w-0.5 in increments of 1.0
• y-axis: y from 0.5 to h-0.5 in increments of 1.0

Example: size 1x1

1 pixel x 0.5, y 0.5

Example: size 10x10

100 pixels, x from 0.5 to 9.5, y from 0.5 to 9.5

 

 

rgb16 PT filter rounding issue for x equals zero.afphoto

Edited September 5, 2021 by NotMyFault
Added updates (marked as such)

[Ezbaze]

This still happens in v2.1.1 🤔

[NotMyFault]

Assume this is not a bug but by design. Unfortunately Affinity (moderators) never answered the question.

[NotMyFault]

And if affinity would change the function, it could break many existing documents.

[Ezbaze]

1 minute ago, NotMyFault said:

And if affinity would change the function, it could break many existing documents.

That is an interesting problem, I guess there could be a conditional case based on what version the document was created in, but that would just result in messy code in the long run ._.

[Leigh]

On 7/8/2023 at 5:54 PM, NotMyFault said:

Assume this is not a bug but by design. Unfortunately Affinity (moderators) never answered the question.

Sorry, not sure why you initial post wasn't answered. I've spoke to our developers and they've confirmed it's by design. I will see if we can get something added to the help to briefly mention this to avoid further confusion.