No smooth contour.

[R C-R]

9 minutes ago, JimmyJack said:

Actually there are no variables. Everything has been treated the same.

What resampling method & ICC profile did you use in AD for your export? Are you certain both are exactly the same as what you use in AI for the export?

10 minutes ago, JimmyJack said:

Yes I do think one is better.

I can see differences, but since none have perfectly smooth contours, deciding which of them might be "better" remains a subjective judgement with no definitive answer. For example, to my eye the "P" in one of them is better than in the others but the "a" is better in another & the "O" in all of them is not very good.

[JimmyJack]

18 minutes ago, R C-R said:

What resampling method & ICC profile did you use in AD for your export? Are you certain both are exactly the same as what you use in AI for the export?

There is no resizing being done. Therefore the resampling methods are irrelevant. Kinda the point 😉.
Yes same color sRGB 2.1. Yes.

[Designer1]

12 hours ago, JimmyJack said:

There is no resizing being done. Therefore the resampling methods are irrelevant. Kinda the point 😉.
Yes same color sRGB 2.1. Yes.

@JimmyJackTry with a CMYK document. The smoothing will be a bit worse.

[PaulEC]

On 3/19/2022 at 3:52 PM, Designer1 said:

I will not participate further in this discussion.

A couple of pages later - I have looked at, and tried myself, numerous combinations of factors in exporting files in different ways from several apps etc etc etc, I have come to three conclusions:

1: Different apps produce very slightly different results. None (IMHO) is "better" than another.

2: Any "quality" issues are down to the settings used, not to any intrinsic problem with Affinity apps. (i.e. - there is nothing that needs to be "fixed".)

3: I am wasting my time here and "I will not participate further in this discussion." (At least we agree on something!) 😶

[Designer1]

[Designer1]

First is from CorelDRAW, second from Affinity Designer. Open Sans, 144pt. RGB black.

[GarryP]

Designer1: Have you made sure that the texts in your latest examples are drawn in exactly the same place in each document?
And by exactly the same place I don’t just mean that the layers have the same X and Y coordinates but the furthest-left point of the outer curve of the “O” was at exactly the same place (to the most accurate pixel coordinate available in both applications) in each document and the same for the furthest-top point of that “O” glyph.
If they are not then it’s probably not a fair comparison.

[Designer1]

10 minutes ago, GarryP said:

Designer1: Have you made sure that the texts in your latest examples are drawn in exactly the same place in each document?
And by exactly the same place I don’t just mean that the layers have the same X and Y coordinates but the furthest-left point of the outer curve of the “O” was at exactly the same place (to the most accurate pixel coordinate available in both applications) in each document and the same for the furthest-top point of that “O” glyph.
If they are not then it’s probably not a fair comparison.

The word Open is aligned in the same way in the middle of the A4. The colour is the same and so is the font size. In addition, it must be said that it is CorelDRAW from 2013.

[User_783649]

@Designer1 First one has obviously better quality. In the second sample there are various individual "leftovers" semi-opaque pixels that make it appear more harsh and fuzzy. While in the first sample their opacity appears to be more aligned with their geometrical placement and distance from the main curve. Therefore, an overall quality is better.

I’d suggest you sending these directly to Serif support. Just to not waste your (probably valuable) own time. It will be better to receive qualified response from them or even get some kind of engineer feedback on this issue.

By continue posting here you just trying to convince other regular users. I mean, everyone has different eyes, different screens and different expectations. For some people all these are just... letters, you know. Many people won’t even able to spot the difference between some typefaces, so what to say about subpixels and quality of antialiasing?

[JimmyJack]

Not in the same spot. 
I suggest making one master (vectot) file and opening in various programs rather than trying to recreate something multiple times.

[Designer1]

1 hour ago, Alex M said:

@Designer1 First one has obviously better quality. In the second sample there are various individual "leftovers" semi-opaque pixels that make it appear more harsh and fuzzy. While in the first sample their opacity appears to be more aligned with their geometrical placement and distance from the main curve. Therefore, an overall quality is better.

I’d suggest you sending these directly to Serif support. Just to not waste your (probably valuable) own time. It will be better to receive qualified response from them or even get some kind of engineer feedback on this issue.

By continue posting here you just trying to convince other regular users. I mean, everyone has different eyes, different screens and different expectations. For some people all these are just... letters, you know. Many people won’t even able to spot the difference between some typefaces, so what to say about subpixels and quality of antialiasing?

 

@MEB

I can send Serif support these files. What email does Serif support have? You are also welcome to send to Serif Support.

Support

For questions, reporting bugs and requesting features please post on our forums.

https://affinity.serif.com/en-gb/contact/

[R C-R]

2 hours ago, JimmyJack said:

Not in the same spot. 

As I think is also demonstrated in this difference.afphoto example, where the two PNGs were added as pixel layers to an AP file with the corel one set to the Difference blend mode. If the text was at the same position in both then there should be no differences & the result should be pure black.

[JimmyJack]

3 hours ago, R C-R said:

As I think is also demonstrated in this difference.afphoto example, where the two PNGs were added as pixel layers to an AP file with the corel one set to the Difference blend mode. If the text was at the same position in both then there should be no differences & the result should be pure black.

Mmmmm, close. Yes, if the same the result should be black. But, the difference mode is combining the two so as to give the appearance of a halo around the the whole of each letter so you don't really know which direction one or the other is off, just the total effect.
Keep and blend and give either layer a 50% opacity and you'll see the actual offset. 
Or, just give them normal blends, get in close and toggle the visibility of the top one.

[R C-R]

1 hour ago, JimmyJack said:

Yes, if the same the result should be black.

I was not interested in how they differ, only in demonstrating in another way that they do. That's why I went with the difference blend mode.

[iuli]

Over seven whole pages around the fact that there’s virtually no difference in export quality between Affinity, Adobe, Corel (and others). If anything, I think this topic proves yet again that despite Adobe and Corel being a hundred times more expensive, the quality is (overall) the same.

[JimmyJack]

2 hours ago, R C-R said:

I was not interested in how they differ, only in demonstrating in another way that they do. That's why I went with the difference blend mode.

What I'm saying is that all the pixel could be exactly in the same spots with only a change in gamma setting and you'd still get the halo. Doing the 50% opacity shows they are exactly in the same spot. So the difference blend alone might not be enough.

example file attached, play with the opacity:

halo.afdesign

[R C-R]

3 hours ago, JimmyJack said:

What I'm saying is that all the pixel could be exactly in the same spots with only a change in gamma setting and you'd still get the halo.

I know what you are saying, but all I was trying to show is that the two PNGs are not the same.

[User_783649]

8 hours ago, iuli said:

If anything, I think this topic proves yet again that despite Adobe and Corel being a hundred times more expensive, the quality is (overall) the same.

Fully agree. For the price and licensing method Serif is giving us (it’s almost charity) we should have literally zero complains especially about such minor differences.

I would also like to add that it’s not only Affinity who seems to be showing things slightly worse. Honestly, macOS itself and any software built on top of it’s graphic framework (think about apps like Preview, Safari, Font Book, Pages, Numbers, Keynote, Sketch etc.) — they all show things same or even worse than Affinity. Or just different.

So, no one is perfect! No one.

And probably I’d prefer having slightly less polished output (on macOS difference is not that bad actually, OP has more notable problems on his Windows system) but I’d know that it comes more in line with any other software I use on a daily basis to work and deliver my documents to clients. So overall predictability  —  more important for me.

Honestly, I absolutely don’t mind seeing render quality being improved in some future. Everywhere, not just in Affinity.

 

[GarryP]

16 hours ago, Designer1 said:

The word Open is aligned in the same way in the middle of the A4.

In what way do you mean that?

If you have aligned some kind of text layer, in each application, centrally on the page, then your comparison is relying on the two applications drawing the text in that text layer exactly the same way, which they might not do.

If you have converted the text to curves before aligning the resultant curves, centrally on the page, then your comparison is relying on the two applications converting the glyphs to curves in exactly the same way, which they might not do.

Also, if you are using the application alignment functionality then you are relying on both applications performing that alignment in exactly the same way, which they might not do.

To properly compare the ‘same thing’ in two applications you need to make sure that the documents in both applications contain exactly the same thing and, in this case, it seems like you may not be doing this (as has been shown by the examples above).

[Designer1]

2 hours ago, GarryP said:

In what way do you mean that?

If you have aligned some kind of text layer, in each application, centrally on the page, then your comparison is relying on the two applications drawing the text in that text layer exactly the same way, which they might not do.

If you have converted the text to curves before aligning the resultant curves, centrally on the page, then your comparison is relying on the two applications converting the glyphs to curves in exactly the same way, which they might not do.

Also, if you are using the application alignment functionality then you are relying on both applications performing that alignment in exactly the same way, which they might not do.

To properly compare the ‘same thing’ in two applications you need to make sure that the documents in both applications contain exactly the same thing and, in this case, it seems like you may not be doing this (as has been shown by the examples above).

Affinity Designer does not export PNG satisfactorily. For design in advertising agencies exported quality is not good, for hobby design possibly still goes. You are free to compare themselves.

[Patrick Connor]

17 hours ago, Designer1 said:

I can send Serif support these files. What email does Serif support have?

There is no need, this is already logged for improvement, so we do not need your files thank you, but is unlikely to change any time soon.

[Designer1]

22 minutes ago, Patrick Connor said:

There is no need, this is already logged for improvement, so we do not need your files thank you, but is unlikely to change any time soon.

Thank you for this information. I hope the export quality will be improved. I use Affinity from Windows Store. Otherwise Affinity Designer is a great software.

[Designer1]

On 3/22/2022 at 11:55 AM, Designer1 said:

Thank you for this information. I hope the export quality will be improved. I use Affinity from Windows Store. Otherwise Affinity Designer is a great software.

Here is an example of how free Inkscape can export the PNG. The quality of the export is much better than with Affinity Designer. In this respect, Affinity should really improve this.

[JimmyJack]

I didn't forget!! 

The "answers" to my comparisons.

Two circles:
Left = AD
Right = AI standard png export (better IMO, but very slight)

Text:
1 (top) = AD text at 2.2 gamma
2 = AD at 1.4 gamma, it's default for text
3 = AI png supersample export setting (the best IMO)
4 = AI standard png export (the worst IMO)

Bottom line: I do think there are marked differences. I think this is just one corner of the aliasing issues within Affinity that show up many places. I think, though, that it comes down to company philosophies. Affinity has picked the aliasing protocols it's picked. The results are by design. I personally disagree with it, but hey.... that's what makes horse racing.

Thanks all who chimed in.

On to the next.
So few tests to do, so much time. Strike that, reverse it 😉.

[R C-R]

2 minutes ago, JimmyJack said:

Text:
1 (top) = AD text at 2.2 gamma
2 = AD at 1.4 gamma (it's default for text)
3 = AI png supersample export setting (the best IMO)
4 (bottom) = AI standard png export (the worst IMO)

To my eye, the "O" & "P" in the defaults (2 & 4) look almost the same but the "a" in the AD default looks smoother than the one in 4 & I actually like how it looks in 2 better than in any of the others.

I agree that the AI standard is probably the worst looking overall but only because its "a" is more jagged looking than in any of the others.