User_783649

Hi guys,
I'm wondering what would be the optimal workflow whith regards to sharpening when you try to export (Exporting through the Photo Persona) a JPG from an existing picture? When I work on high res photos of course I don't want to resize the original before output to be able to properly adjust the sharpening settings for the required output resolution. On the other side a proper judgement of sharpening can only be done in the final resolution. The export dialog where you enter the output resolution does not allow setting the sharpening parameters for the entered resolution. I don't see a proper way in the export Persona either.

For proper adjustment I would need all sharpening parameters and a 1:1 pixel precise view in export resolution. To me this seems to be a quite commen process. How do you realise this in Affinity Photo? What is the recommended workflow.

Cheers, Timo

Thinking a bit more on it, I'd guess that this was probably done intentionally. Using bilinear always would make it look blurry when you zoom far enough where you would actually want to see individual pixels.
Still, every other application I've used has handled this better. I've never encountered this problem in Photoshop, GIMP, Paint.NET, Krita, etc. None of them have the jagged edges caused by NN while zooming in, but show discrete pixels properly when you zoom in far enough. Obviously I don't know how any of these actually work, though an obvious answer that'd better than what AP does now is to just use bilinear until a fixed cutoff. Something like a cutoff for %zoom, or actual pixel width used to display each pixel in the image (e.g. # of pixels on your monitor for each pixel).

I can only echo Aongus's words. Affinity suite has been a god sent for me for my book projects and it delivers outstanding results and value for money. I don't what I would do without the trio. (Although I use Publisher & Designer the most). Good on you Serif. Keep it up!!! 

I can't understand the negativity in some of the posts on this thread.
The Affinity suite is very much under active development, with six updates in 2021 alone, and a patch in March this year. The focus seems to be on refining the details for a professional workflow rather than on feature parity with other platforms. Serif has made it clear elsewhere that a major upgrade is in the works for 2022, and it will be a paid upgrade.
Affinity's value is outstanding, for software which (for me) delivers fully professional results in both newspaper and magazine illustration and greeting cards. 
For comparison, the Creative Cloud multi-app plan that most professionals use costs $600 or €737 annually -- more than 4 times the once-off cost of licensing all 3 Affinity apps. 
Even plugins for Adobe Illustrator are now adopting the subscription model. Phantasm now costs $140 a year plus tax (23% where I live). Again more than all three Affinity apps combined. Another plugin, EXDF-Pro (for importing DXF files) is $99 a year plus tax.
So I'll be happy to support a paid once-off upgrade given the alternative!
No inside information, but I wouldn't be surprised if Serif made some announcement next month to coincide with the Apple developer conference.

I understand that, but I'm not looking to work with vector graphics. My expectation is that when I select "Bilinear" under "View Quality", the viewport will be always be rendered with bilinear sampling, and not nearest neighbor. If you zoom out further than 100% and toggle between bilinear and nearest neighbor, it's clear that it does respect the setting. But as soon as you zoom in to greater than 100%, it uses nearest neighbor regardless of which option you select.

With zoom levels > 100%, the viewport is always rendered with nearest neighbor, even if the bilinear View Quality option is selected.
The attached images demonstrate this. This was done using the latest version 1.10.5.1342
the original image:

screenshot of the viewport zoomed to 125% (with the View Quality preference set to bilinear):

the image resized 125% (625x625) using nearest neighbor:

the image resized 125% (625x625) using bilinear:

 
The screenshot of the viewport at 125% is clearly using nearest neighbor when comparing to the other two resized images.

Thank you for these words @Sotalo

That is exactly my experience with Affinity suite as well.
After more than a decade spent in Adobe suite for me it’s like entirely new wonderful world full of new possibilities.
Lots of things I get used to previously and thought that it’s normal now feel stupid and simply weird.

Seamless transition between personas and apps, whole non-interruptive workflow is one of the reasons I switched comlpetely.
Everything seems better for me in Affinity suite overall. Not completely bug-free, but overall experience much, much better. I'll never come back.

I’m also with you in regards of subscriptions. For last years Adobe products have been stagnating. Whole forums full of bugs, rants, hopeless requests. Lots of unsolved core problems, including legacy code, performance issues and more. Lots of small bugs being introduced and reintroduced with every minor update. Again, again and again. Sometimes I feel like there’s no QA at Adobe at all. I could literally feel the pain that all these people were experiencing because company was totally ignoring them for a long period of time.
I remember how people were asking for artboards in Photoshop. For five years or more. And then Sketch appeared on the stage. Only after this they started doing something in this direction. They made XD. But it is a joke app. There are important feature requests that are four or five years old as well. No one cares at Adobe. But they made 3d transforms, voice control and other fancy stuff. Because it's cool, right?
There are lots of similar examples. I’m just tired of all this nonsense.
I just don’t understand why I should continue paying monthly for software that progressively becomes worse and worse, slower and slower. Just let me use the version I'm okay with forever and stop charging me at all once you fix all your issues. No, you can't stop paying us. You loose all access.
This whole last year of Photoshop updates, for example. It’s a complete disaster. They finally started to change their GPU code because suddenly they realised that OpenGL is being deprecated. Apple warned about this many years ago. Guys, there’s Metal you know. Hurry up.

Adobe definitely had time and resources to carefully plan and implement all the changes in separate internal beta branch without breaking core releases every month. Lots of people were affected by this and many other issues.

And now they’re removing 3d completely because it became too hard for them to support it in modern operating systems as they say. I believe they just want all you to purchase a couple of another subscriptions to Substance. More subscriptions, more money!

There were times Adobe was focused on professionals. But they are gone now. Now they turned to younger audience who’s just starting out. With all these fancy features being added all around like replacement skies and neural filters. One-button quick solutions. Don't think, it's already here. Just press this and that. Cool, right?
It’s not surprising to see there’s no more focus on performance or other important things.

And I can understand this. There are way more beginners than professionals. Beginners are easier to handle. They’re not demanding anything serious like pros, they’re easier to manipulate. It’s easier to explain them, that something is done “by design”. They're less sceptical and easier to impress.

And maybe that’s fine and good for Adobe we know today. As we all see they doing incredibly great in terms of revenue.
But seems like they just don’t care about pros anymore.
That's why some of us switching and looking for alternatives. And I'm very happy that companies like Serif exist.

I'm not sure it's a good idea... like the others said, it would be a return to the past to have apps that hung the whole computer and needing to be used as in old Apple II mode: 1 app at a time, stop everything else.
Or I misunderstood.
But today, we need different apps running, checking and answering emails... work is multitasks.

@Dan C @Patrick Connor

For comparison, the same design with 120 Tiff images in InDesign. Unfortunately, one can hardly work professionally in Publisher due to the system overload. Making now in InDesign.

Hi lloerau,
Welcome to Affinity Forums :)
This is a known issue that's already logged/aasigned to be looked at. It happens when you use arbitrary zoom levels. If you set the zoom to 200% for example it shouldn't render like that. You can press ⌘ (cmd) +1, ⌘ (cmd) +2 and ⌘ (cmd) +3 to quickly set the zoom to 100%, 200% and 400% respectively.

this issue still seems present in 1.8.3 is that correct? I tried all settings, bilinear, metal open gl, rendering still is jagged at certain zoom levels.
Also still issues with unsmooth lines (see second screen shot). I'm a sketch artist and don't want to use the stabiliser. I was hoping I could match the sketchbook pro feeling. 

Third issue is the pressure sensitivity, that is also not always smooth:
 
 

I love AF and it has replaced PS for me which says a lot as I have been using PS professionally for 20 years. But please fix the brush strokes and rendering and it will be perfect. 

Hi Sima,
 
No, it doesn't:( "Bilinear (Best Quality)" is set by default. Changing other settings in the Performance tab has no affect on jagged lines. 

Hello! I'm trying to use AP as a digital painting program and here is a weird thing which bothers me. When I'm in Photo Persona and zoom more than 100% I can see jagged lines. It doesn't affect final lines quality but is a bit annoying anyway. Other programs (like Krita or Sketchbook Pro) smooth lines when zooming. Moreover, if I switch to Export Persona in AP, then my lines look smoothed as in the mentioned programs. I've attached two pics to illustrate what I'm talking about - one with jagged edge (I took screenshot and cropped it) and another is an exported via Export Persona slice. The scale is different, I know, but I hope you get the point. Ok, in short, is there any option to switch between jagged lines (which may be useful for some reasons, f.e. pixel art) and the smoothed ones which are better for sketching/digital drawing?

Currently, Affinity Photo does not provide a absolute trustworthy preview of the current document.
For any zoom levels not equal to 100%, you get misleading rendering of noise and unsharp mask filters With View Mode "Bilinear", you will always get false rendering on pixel level at hard edges Even with zoom level 100%, you could get get false rendering when dealing with pixel art For zoom level below 100%, you could get completely wrong colors caused by resampling (same bug report above) Many live filters can introduce rendering artifacts at any zoom level, including 100%, leading to severe deviations from final result, especially when combined with blend mode "difference" The false color by resample issue could bite you at almost any zoom level except 400% and integer multiples of 400%  
The only way to get a reliable preview of your document is to either
Export - leaving the application Merge visible - introducing superfluous temporary layers, must be created on top of layer stack and deleted after use.  
Even the export preview 100% function is currently broken on iPad.
To overcome all these issues, it would be helpful if Photo would offer an "absolute trustworthy preview" move, similar to Designers View Modes.
Requirements:
No color artifacts (false colors) by resample in case of zoom <100% (use color-correct average, not false sub-sample) 100% accurate rendition of result ability to zoom to any level (at least -8x to + 8x) Ultimate priority on accuracy, no priority on performance. Selectable options to deal with non-opaque pixels (color of matting background) Regular keyboard shortcuts and mouse actions to zoom / pan  

This looks similar to the problem here: 
 
 

I've not looked at the files, but from the screenshot it looks like an sRGB document. Possibly the original is printing as K only, whereas the new one is composite black?

Just wanted to let you know that i've logged this suggestion with our developers 👍

It has been pointed out multiple times here by some of Affinity developers, that Bilinear method was chosen by design as a quality/performance compromise. It provides fairly acceptable image quality at a lower performance cost. As we know, Bicubic uses 16 pixels (4x4) and Bilinear uses just 4 (2x2) so it definitely allows to process and draw things faster. But the quality...
While I may accept that this might be totally okay for most people, I personally find it strange to not being able to see what exactly is going on the canvas I work with. I need more precision and having that level of precision even when "it's only for viewing in the program" — crucial for any kind of professional workflow. At least for me.
Even for downscaling, which I agree we use most of the time (because who the hell upscales images? it's a crime!), Bilinear is not the best option, be it Affinity or Adobe. Everyone can take a screenshot of this forum post or some other piece of text and place it as an image layer on their canvas.
When we downscale this layer we can see how bad everything became. Some letters loose their proportions, some became very blurry and other became too blocky and crisp. While with Bicubic everything looks way better and more uniform, obviously. Tonal gradations are smoother and whole image has fewer interpolation artifacts.
Hopefully Affinity introduces Bicubic as an option in View Quality settings. After all, it's not a battle of which algorithm is ultimately better. I should agree that in some cases some algorithms may work better than others. But having more options for canvas view quality would be beneficial.
 
 

This corresponds to Alex's post - thumbnail on the Layer panel is identical, because here it is rendered using the OS without the ICC profile application, etc., while the canvas is rendered with the application of all the above settings, and therefore the colors are different.

Thanks for all the input.
I found out what causes the difference: Relative Colorimetric vs. Absolute Colorimetric.
I wrote "the color presets were identical". This was wrong. They were different regarding the rendering intent.

They look different because they are different.
in Publisher the colour is...

In Photo the colour is...

Publisher and Photo 1.10.5 on Windows.
Same color looks different.
In both apps a doc in mode CMYK ISO Coated was created. Color preferences, document color mode and set color in color panel are the same. Both apps are displayed on the same monitor. In both apps the same hex color is used.
The colors should look the same in that case, but they don't. User error of bug ?

cmyk.afphoto cmyk.afpub

It has been pointed out multiple times here by some of Affinity developers, that Bilinear method was chosen by design as a quality/performance compromise. It provides fairly acceptable image quality at a lower performance cost. As we know, Bicubic uses 16 pixels (4x4) and Bilinear uses just 4 (2x2) so it definitely allows to process and draw things faster. But the quality...
While I may accept that this might be totally okay for most people, I personally find it strange to not being able to see what exactly is going on the canvas I work with. I need more precision and having that level of precision even when "it's only for viewing in the program" — crucial for any kind of professional workflow. At least for me.
Even for downscaling, which I agree we use most of the time (because who the hell upscales images? it's a crime!), Bilinear is not the best option, be it Affinity or Adobe. Everyone can take a screenshot of this forum post or some other piece of text and place it as an image layer on their canvas.
When we downscale this layer we can see how bad everything became. Some letters loose their proportions, some became very blurry and other became too blocky and crisp. While with Bicubic everything looks way better and more uniform, obviously. Tonal gradations are smoother and whole image has fewer interpolation artifacts.
Hopefully Affinity introduces Bicubic as an option in View Quality settings. After all, it's not a battle of which algorithm is ultimately better. I should agree that in some cases some algorithms may work better than others. But having more options for canvas view quality would be beneficial.
 
 

Can we get an update on this, please? If there's a chance we'll see a proper canvas view quality option (not bilinear, sorry) in all Affinity apps?
Be it an overall zoom level different from 100% or any kind of image layer being scaled down — observable image quality is mediocre at its best.
Rough and fast approximation by using computationally cheap and visually imprecise algorithm is not a suitable way for professional apps which Affinity certainly claims to be.
I'm personally willing to sacrifice some performance for better canvas fidelity. I have a good and powerful system and just want better image quality.
I personally already accepted the way you (re)render the canvas with those blocky tiles all around. Sad you can't do better, but I think I'm already used to it.
But I can't accept the way you resample everything on the canvas with bilinear.
At the very least, make Bicubic a third option in settings. Please.

Bicubic resampling is a good, universal resampling method and definitely should be made a default choice (or at least an option) in Affinity Photo.
While we may found it in export window, next to more advanced Lanczos options, it seems logical to include it on a canvas level, in View Quality settings.
Also it would be great to see it being added as an option in Designer and Publisher as well for the sake of consistency. As of now, in all three apps, we're limited to use Bilinear which doesn't provide the same level of visual quality and details precision especially while zooming in and out, resizing objects.