NotAffine
-
Posts
12 -
Joined
-
Last visited
Posts posted by NotAffine
-
-
Hey, so I made a minimal example of this in the attachments of this post.
Gist of it
If you have two perfectly aligned, angled shapes and you want anti-aliasing as to not have jagged lines everywhere, anti-aliasing will remove enough of both shapes (instead of adding to at least one of them) so the background shines through.
In this example I made two irregular shapes with the pen tool. The orange (upper) shape has its upper nodes snapped to the upper corners of the artboard. The lower ones were placed along the sides at different heights. The purple (lower) shape has its lower nodes snapped to the lower corners of the artboard and the upper nodes are snapped to the lower nodes of the orange (upper) shape. Theoretically this would mean the resulting image should have no gap between them. Practically it does (except if you disable AA for this artboard - which makes everything look jagged instead).
The background is just an eyesore of a green color, which shines through in the exported PNG image at all resolutions (see my export settings).
Yes, I know
you can "just" overlap the shapes. However:
1. This is a really annoying process if you have a large project. Not only if you've already finished and you now have to fix it but the creation of this kind of artwork is still annoying cause you can't just rely on snapping, you need to zoom in and move everything just slightly inside each other.
2. It's pure testing of what actually works. If you have more complex shapes, you might need multiple tries of shoving shapes under shapes cause sometimes the first try is not enough.
Somewhat important note
I've tested this behavior in Affinity Photo, Adobe Photoshop and Adobe Illustrator and it's just as broken there, too.
This is how the edge looks (zoomed in):
Solution?
See my purple text above with (what I hope is) a hint towards fixing this issue. I'm not good with graphics programming myself but maybe it's possible to influence the anti-aliasing by looking at what shape is above the other in the layer hierarchy? In this example one passable solution would be to either fill the created gaps with the shape that is "above" in the hierarchy (which is purple) OR to tell the renderer to remove (sub)pixels from the lower shape (orange, lower in hierarchy) while adding to the upper one (making the purple shape effectively a bit bigger).
-
I just created a new project (newest Designer version), created a pixel layer and used the fill tool. Then the following happened:
-
6 minutes ago, haakoo said:
I could've sworn I tried the same thing but yeah, it works now. But only if I select the nodes (with the node tool selected) and not the whole curve (with the transform tool selected). Right?
-
1 minute ago, Lagarto said:
But I am condient that Affinity will get these wrinkles smoothed out eventually.
The one who helped me combine the shapes originally said Affinity gets said to fix that since around 2015 and they still haven't done that. Of course I can't just believe such a loose statement but if that's remotely true it's pretty sad.
This makes working with symmetrical shapes (which are quite common) soo much worse than in Illustrator. If you have a bigger project with symmetrical shapes you'll loose hours on these small details.
-
5 minutes ago, haakoo said:
zoom in very closely bring them both to the document snapline,then its no effort to add both the green shapes together
I think you didn't get the problem.
1. The snapping itself should snap them together so you don't need to do that in the first place.
2. If you do what you said and then zoom in even further, you'll see either that the shapes still don't meet or that the shapes already overlap. You can't possibly put them together just like that, there'll be always a zoom level where it doesn't fit any more.
-
4 minutes ago, Lagarto said:
The nodes can be merged but only with considerable effort
Exactly. I managed to do it with a different method but it was a comparable effort to do so. This could be improved greatly by "just" fixing the snapping, I think. Or, if the actual problem arises because of some problems with the shapes themselves, there should be an option to see/fix those issues!
Do you agree?
-
3 hours ago, Dazzler said:
To solve this, click on each of the green layers and click the cog icon at the top of the layers panel. Then in the dialogue that pops up, select the coverage map and drag the left hand point up to make a straight line across the top and it'll solve your issue.
This only solves the issue visually. They're still two different shapes. You can see that if you zoom in quite a bit. Or if you let Designer show you the nodes.
-
Ah, damn, I uploaded the wrong file. Uploaded the final result but I forgot to save the history there. Uploading it here now. I'm sure you'll spot the seam in the middle of the green two curve layers.
They are snapped to each other but they still have a seam that only gets bigger the further you zoom in.
@Callum
-
So I tried making a shuriken. For that, I wanted (at first) to make a triangle and cut out a bit of that in the middle. I made a quite simple shape, wanted to duplicate that, flip it horizontally and then combine those. But there remains a seam, no matter how I snap the two together.
Someone on an unofficial Affinity Discord Server (credits to gromofdoom) helped me out by saying the snapping was never really working in this software and I need to overlap the two halves, combine them and then delete the unneeded nodes.
Why does Affinity make such a simple task this painful? That's a pretty difficult workflow for something that wants to be an Adobe alternative.
Or am I blind and there's actually a better way to do that? I attached my project file (with extended history in it!) so you can see what I did (there's much stuff I did which didn't make any sense, lol).
Antialiasing destroys angled shapes in Designer [+ Photo]
in Pre-V2 Archive of Affinity on Desktop Questions (macOS and Windows)
Posted
Yeah, I thought of that I I also thought that might be fixable? I mean, it's probably possible to somehow create an AA algorithm that takes in that knowledge. But as you later pointed out, it's probably not going to happen.
Sadly, that doesn't help me cause then I still get a partial shine-through background.
Like I said, I'm not very familiar with rendering software (although I've took a course on graphics rendering at uni) so I wasn't sure what exactly the counter points would be.
I'm sure it gets better with time but the thing is I want to make poster designs for my walls and printers can very well just print the jaggedness of my image.
So I have no idea how to actually realize my ideas except by shoving layers under other layers.
P.S. No, you're not correct. I've zoomed in on text very often and I know how (generally) displays work. ^^