In the physical version, if the worm/pigeon and the grid are exactly the same color, we still consider them separate entities.
In the digital version, if a group of pixels appears to be part of some object, that's as good a reason as any to argue it is part of that object (though strictly speaking, it's a meaningless question since it's just a bunch of pixels).
One is an illusion, or the set of all illusions, and the other is the set of all images generated by pixels.
Not all images generated by pixels are an illusion. Otherwise you would say that all images generated by, say, acrylic paint, are an illusion because they're made of molecules.
That we perceive paintings of objects as objects is an illusion, clearly. But that is not the same illusion as this "pigeon neck" thing.
An illusion is not an illusion if what you perceive is what is actually happening.
The whole point of the "pigeon neck" illusion is that it tricks the eye into perceiving areas as growing that we know are incapable of growing. If the areas are inelastic matter, that's a cool effect: we know that a printed image on a transparency sheet cannot truly grow just by sliding it around a table.
If the areas are pixels... it's less significant. Nothing is actually moving to start with. It's all neighboring lights flashing on and off. If a group of those lights is part of an object, it's only to the extent that the viewer perceives it as such.
Note that this is not the case with other digital illusions. This one in particular is ruined because it relies on a strip of pixels 'belonging' to one of the two (perceived) objects that sandwiches it.
Images on a screen, or a painted painting, etc, I disagree.
There is, objectively, data being presented, which we perceive.
You're glossing over the importance of a particular scale, the scale at which humans operate in on a daily basis, and making the assumption that reality is what's happening at a smaller scale.
This is incomplete.
If we want to talk about the image, the underlying components become progressively less important as the scale decreases. When we view a painting, or an image on a screen, when are less interested in up quarks, down quarks, and electrons, and more interested in the image and the medium.
In the digital version, if a group of pixels appears to be part of some object, that's as good a reason as any to argue it is part of that object (though strictly speaking, it's a meaningless question since it's just a bunch of pixels).