Bryan Rieger

I haven't used these machine learning models yet, but I just wanted to say how nice a surprise it was to see that these were optional downloads, and not simply bundled into the existing applications. I really appreciate Serif giving creatives the respect to choose whether or not they want to integrate any machine learning into their workflows. 
Also, kudos for side-stepping the blanket AI hype train and calling these 'Machine Learning Models', which is not only more accurate, but also helps to frame the discussion in a much more informed manner.

Ah, thanks @MikeTO that was it. I didn't realized I had changed them ages ago. this is what my Settings > Publisher > Document > Shortcuts looks like now.
Wouldn't it make sense just to make these the defaults, instead of requiring users go in and set them themselves?

Sure thing @EmT.
In this video the toolbar starts with a black stroke and a width of 5.5, while the colour panel also shows the black stroke, the stroke panel indicates no stroke. Drawing with this results in a curve with no stroke or colour applied. Applying a colour (green) to the strokes makes them visible (0.2 stroke width), yet then drawing another line with a green stroke, 5.5 width in the toolbar results in a no stroke displayed in the stroke panel, and a curve drawn with no colour or width.

CleanShot 2024-10-22 at 09.03.03.mp4    
 

I haven't used these machine learning models yet, but I just wanted to say how nice a surprise it was to see that these were optional downloads, and not simply bundled into the existing applications. I really appreciate Serif giving creatives the respect to choose whether or not they want to integrate any machine learning into their workflows. 
Also, kudos for side-stepping the blanket AI hype train and calling these 'Machine Learning Models', which is not only more accurate, but also helps to frame the discussion in a much more informed manner.

Hi @Iztok, I managed to get an export of your 31.4 mb pdf down to 1.5 mb (see attached) in Publisher 2.5.5 using the default PDF (digital - small size) settings with many of the vector elements still intact. I don't have the font you're using so it's substituted a standard font, but that makes me wonder if possibly the embedded font you are using might be the source of the larger file size. 
I would try changing the settings for embedded fonts to see if that helps, and I'd also suggest ensuring subset fonts is checked in case you're using large, international fonts with hundreds (or possibly thousands) of glyphs that are not actually used in your document.

Geološke poti A 4 V3.pdf
 

I don't think this would happen. I think a lot of people would welcome the fact that one could export design elements or files into Affinity products for further processing and then send it back to Canva. 
Nobody would force anybody to use Canva and those who want it will have a way of doing it.
This is very similar to what Affinity has done with AI. It's an opt in feature.
So those who want to use Canva will have a way of sharing files with Canva and those who don't want to have anything to do with Canva will simply not use it.

This is essentially a copy of the (auto) collapsible iconic panels in the Adobe interfaces, which I use a lot, especially with plug-ins that have UI panels. It just helps to get everything out of the way, and not have to remember where the panels are, or what persona they may be in. Not sure if this impinges on any Adobe patents, but it's a nice quality of life improvement, both on the iPad and desktop IMHO.

I haven't used these machine learning models yet, but I just wanted to say how nice a surprise it was to see that these were optional downloads, and not simply bundled into the existing applications. I really appreciate Serif giving creatives the respect to choose whether or not they want to integrate any machine learning into their workflows. 
Also, kudos for side-stepping the blanket AI hype train and calling these 'Machine Learning Models', which is not only more accurate, but also helps to frame the discussion in a much more informed manner.

I haven't used these machine learning models yet, but I just wanted to say how nice a surprise it was to see that these were optional downloads, and not simply bundled into the existing applications. I really appreciate Serif giving creatives the respect to choose whether or not they want to integrate any machine learning into their workflows. 
Also, kudos for side-stepping the blanket AI hype train and calling these 'Machine Learning Models', which is not only more accurate, but also helps to frame the discussion in a much more informed manner.

So, my 2¢ is that the MASSIVE icons that appear when you interact with the page thumbnails (shown above) are really unintuitive, and I feel like I'll need to keep a desktop reference to decipher what on earth they're supposed to indicate. Why when I drag a page do I get an icon showing me dragging 3 pages? Would it be possible to display what action is being represented in the icon by also displaying a text description in the status bar?
Hiding functionality behind a plethora of keyboard shortcuts is great for power users, but there is ABSOLUTELY NO DISCOVERABILITY for more casual users, or folks coming to the apps fresh. This desperately needs to be made more intuitive, and user friendly.
They're also REALLY BIG, but still really indecipherable.

The slug is used for much more than imposition information.
As a FMCG packaging designer I'm required by my client to use the slug area for production information. (order information, product and packing information and instructions to the printer). Typically for me it might be an area 25-45mm on one side of the page. The printer will mask that area during imposition, but it's still needed in preproduction so that similar packs aren't confused. Confusion can be a big issue when I send finished art pdfs to my client, who then sends the artwork along with product orders to a factory in another country who then sends the packaging artwork on to their printer. If you have multiple suppliers, then errors can easily multiply.

Yes, you can extend the bleed area to include the slug info, but that will also move the automatic crop marks away from the bleed area, which the printer objects to, so the alternative is to draw in manual crop marks. When you're laying out multiple packs (sometimes 30 - 40 over a few days) manual crop marks become tedious and time consuming.

Because the guy doing your print run will forever curse your name, wonder why you didn't just include a slug, and likely charge you double.

"I daresay you haven't had much practice, sometimes I've believed as many as six impossible things before breakfast."

I MUCH prefer the colour picker working this way by default, but I do agree with @loukash in that having the option to use the existing method as an alternate would be useful (when it's available on macOS).

The checkmark (and X) icons are also surprisingly LARGE (make it feel awkward to read/scan), and feel like bad placeholder art (ie: different line widths).
Should I assume these icons are also temporary placeholders?

Hi Bryan,
The page numbers is a tricky one as text frames don't always flow sequentially and might be found on different ranges of pages. That said, it could potentially be of use if it maybe displayed the page that text flow started on? I will pass the feedback along.
With regards to indication of associated tags - that isn't possible. The tags themselves are applied on a paragraph by paragraph basis (as they're part of the paragraph text style, so you might have multiple different tags for each entry in the Reading Order Panel.
I believe this is a bug, as it is actually possible to do that on Windows! I will get that logged.

A couple of things that would be helpful.
provide page number(s) of the item. ie; Title (1), Introduction (1-12), Chapter 2 (13-47), etc as it would give you some indication where the text is and where it runs to. provide some indication of any accessibility tags associated with the text via styles ie; <h1>, <p>, etc. This would provide you with some indication of whether or not the text has an accessible style/tag associated with it, and which one. Otherwise the user is required to manually check all of these to confirm they are indeed what was expected. So essentially each item would contain the following data:
Frame Text/3 {text frame}, 1-4 {page(s)}, <p> {tag}
Also, why do I need to select and item and click the 'go to' button at the bottom left, when it would be far more intuitive to simply click (or double-click) the item and jump to it directly.

Hi,
First of all, congratulation to Serif for this new leap forward. I guess machine learning was unexpected, and apparently will satisfy the requests of many users who need quick selection of people and objects.
I'm happy to see it is a separate module, so this makes me think that only AI features will be available in a Pro option, leaving the other features to the standard/one-time-purchase version.
Multi-page spreads should make Publisher the ideal tool for the small shops specialized in brochures, that I guess are a crowded part of the customer base. Freelancers are likely still left out by the lack of compatibility with the standard format used by the client agency, and I wonder if this isn't the higher number of perspective users.
The new accessibility feature (reading order) is great for those who make publications that have to be compliant with the law (and with the common sense of making publications accessible to everybody). It makes one think that the work on PDF output is going on with conviction.
As for myself, while I admire where Serif is heading, I have to finally admit that I'm probably tied to InDesign for the next five years, or more. The new features are very heavyweight, and will likely not allow the addition of further features until next summer.
What will come next in Publisher? I would believe it will be ePub export, for the self-publishing people. Maybe some improvement in color management, to make Publisher stronger in the small shops. Essential features for a technical communicator, like long tables, object styles, conditional text, will maybe start to be implemented in 2026. Output to the IDML interchange format will come, if it will, not earlier than 2028.
It is a matter of being realistic, and this release is a great help in doing some planning with some sense of the possible development pace.
Ending this post with a sense of deep depression, but I'll be happy to read about how happy are people with different needs than mine.
Paolo
 

I MUCH prefer the colour picker working this way by default, but I do agree with @loukash in that having the option to use the existing method as an alternate would be useful (when it's available on macOS).

I haven't used these machine learning models yet, but I just wanted to say how nice a surprise it was to see that these were optional downloads, and not simply bundled into the existing applications. I really appreciate Serif giving creatives the respect to choose whether or not they want to integrate any machine learning into their workflows. 
Also, kudos for side-stepping the blanket AI hype train and calling these 'Machine Learning Models', which is not only more accurate, but also helps to frame the discussion in a much more informed manner.

I MUCH prefer the colour picker working this way by default, but I do agree with @loukash in that having the option to use the existing method as an alternate would be useful (when it's available on macOS).

I haven't used these machine learning models yet, but I just wanted to say how nice a surprise it was to see that these were optional downloads, and not simply bundled into the existing applications. I really appreciate Serif giving creatives the respect to choose whether or not they want to integrate any machine learning into their workflows. 
Also, kudos for side-stepping the blanket AI hype train and calling these 'Machine Learning Models', which is not only more accurate, but also helps to frame the discussion in a much more informed manner.

I MUCH prefer the colour picker working this way by default, but I do agree with @loukash in that having the option to use the existing method as an alternate would be useful (when it's available on macOS).

A couple of things that would be helpful.
provide page number(s) of the item. ie; Title (1), Introduction (1-12), Chapter 2 (13-47), etc as it would give you some indication where the text is and where it runs to. provide some indication of any accessibility tags associated with the text via styles ie; <h1>, <p>, etc. This would provide you with some indication of whether or not the text has an accessible style/tag associated with it, and which one. Otherwise the user is required to manually check all of these to confirm they are indeed what was expected. So essentially each item would contain the following data:
Frame Text/3 {text frame}, 1-4 {page(s)}, <p> {tag}
Also, why do I need to select and item and click the 'go to' button at the bottom left, when it would be far more intuitive to simply click (or double-click) the item and jump to it directly.

I haven't used these machine learning models yet, but I just wanted to say how nice a surprise it was to see that these were optional downloads, and not simply bundled into the existing applications. I really appreciate Serif giving creatives the respect to choose whether or not they want to integrate any machine learning into their workflows. 
Also, kudos for side-stepping the blanket AI hype train and calling these 'Machine Learning Models', which is not only more accurate, but also helps to frame the discussion in a much more informed manner.