Creafora seems to have the best of the participants from the old Serif forums.
When I do a global search for Affinity info, I tend to get hits on the old (now read-only) forums, and Reddit. Never Discord.
Edward
Creafora seems to have the best of the participants from the old Serif forums.
When I do a global search for Affinity info, I tend to get hits on the old (now read-only) forums, and Reddit. Never Discord.
Edward
Thanks for the workaround ... I think. 😎 I'll probably stick with the Old Ways for now.
Elaine Giles has an excellect long tutorial, entiled 3.1, a Deep Dive. She has a whole section on "Captions"
A flaw which she actually shows but does not discuss relates to captions on the side.
I updated to 3,2 to start a new document that needs a couiple of captions. I wanted to put the captions on the left, but with the caption text in normal orientation, just as it would be if I created a normal text box. When I set alignment to "left of object", however,, the text is rotated 90 degrees, indicating that the entire caption box has been rotated as well as displaced. No rotation symbols appear about the caption box, and the R box in the Transform panel is grayed out.
I sent feedback. I do have hope for the new caption feature, but it was obviously released with inadequate evaluation. The help even mentions rotating captions, yet that appears to be NYI.
BTW. on the Affinity server at Discord around 1.600 memebers are online. How many users does this forum have? While I'm typing this 2(!) members are online.
Of course, there's the question of how "online users" are counted. Discord has motive to exaggerate the numbers, Creofora does not. I haven't looked at the Discord Affinity for months, but I have not rebooted my computer nor signed out of Discord, so quite possibly they are still counting me as online.
(This is not to disagree with anything said about the unlikelihood of anyone representing Canva/Affinity reading this forum.)
I guess that is an idea; a line that is obvious enough and separate enough to be removed. That is a workaround.
A line with no fill and no stroke is invisible. Not obvious -- you might have to select it using the Layers panel. And depending on your process, probably no need to remove it.
I print several hundred such cards each December, and converted to Affinity a couple of years ago. (I also design and print the envelopes in Affinity.) I use a 220gsm alpha-cellulose stock from Red River, printing on an Epson ET-7750.
I set it up as two 7x10" pages, one for the inside and one for the outside. For setup purposes, I forget that it's going to be 7x5" after folding. If you are setting it up as 7x5" pages, that brings in all the issues of imposition, which I skip. Since I'm printing a large number, I print about a hundred of one side, then the other side.
I have no problems with the setup, once I've defined the custom size in Windows, so I'm not sure what more to suggest. For me, this is a totally straightforward setup.
Surely it's been asked, but I haven't found the answer ... do any problems result from installing Affinity v2 and Affinity Studio simultaneously? I would expect it to be OK but I'd prefer not to dive in without a depth measure.
Following up ... I made a couple of modifications to the technique. Mainly, the method detailed at the above link results in "points" emanating from the corners, which you can see in the images in that post. Here's my result.
1. Open the image and size it, OR create a picture frame and fill it.
2. Cover it with a ROUNDED rectangle. Set the corner as appropriate. I found that I needed larger corner values for smaller images -- I used 5% for an image just under 5x7", but 7% for an image half that size. In any case, I just dragged the slider until it looked good.
3. Add an outline layer effect, alignment: inside, fill style: contour, gradient: 0% opacity at the left end. (I think this last, adjusting one end of the gradient, threw some people who posted in the old thread.)
4. Still in the Layer Effecs dialog, set the Radius for the effect to the desired border width. (I used 66px for a 5x7 image, 48px for a smaller image, this being at 300dpi.) As noted in the original post, if you need a value greater than 100, you can type it in.
5. Close the Layer Effects dialog.
6. In the Layers panel, drag the rounded rectangle thumbnail and drop it onto the image thumbnail or picture frame thumbnail, thus converting it to a mask.
How do I reverse in Affinity? I didn't find a way to do it?
I didn't say in Affinity, I said software. I googled <<software to reverse a string>> and got many hits. The first, https://onlinestringtools.com/reverse-string, works (in English). I think that one is doing it locally (using Javascript) but if I were concerned about privacy, I was use something that's clearly local. I could even program it in Excel, probably requiring VBA.
By "simple", I meant just reversing the order of letters without taking other factors into account. Alfred has pointed out situations where simplicity is not adequate.
I don't follow the discord, do you have a link or a name for it?
(Sorry about the image, I haven't found how to stop this forum from converting a link to an image.)
Single words or short passages in Hebrew or Arabic can be entered, and the glyph shapes will display correctly when a suitable font is used.
There are various ways to reverse a string, so that one could type normally. I don't have a "best" recommendation, and of course it's still far less convenient than RTL support in Affinity, but I think it would be better than typing backward.
A solution to this problem was discussed on the Serif Affinity Forums ‘a few’ years ago
This seems to work well for me, once I got awl thee detales write.
I'll keep the other suggestions in reserve. This method (like some others mentioned) has the advantage that I can apply it to a picture frame, giving me the fade on the frame even if I use the picture frame tools to adjust the image.
Thanks all!
Edward
Sounds simple, but the only way I've found to do it seems excessively heavy. I want an image's edges to fade out over a small portion of the width and height. (The reason is that the image will fill a card to the limits of my Epson ET-7750 printer -- which is capable of full bleed but not on the size cards I'm using -- and the registration is not precise enough to prevent the image from appearing non-centered. Fuzz to the rescue.)
I've done it as shown in the attached. I created four white rectangles covering the image, and for each, I made an opacity gradient on one edge, a different edge for each rectangle. The result is satisfactory, but is there a simpler method? Seems like something for Layer Effects, but I've looked through that panel and haven't found anything for my purpose.
![]()
I'm still sidetracked but I WILL get back to this issue. ![]()
Thanks all, I'll report in a day or two when I have time to read and experiment.
From my experience, these are display artifacts that will not print on paper.
Unfortunately, in this case the artifacts are printing.
Thanks for the suggestions ... I'll have to study the search results more. First try, entering in pixels makes no difference. Since my page settings are 300dpi, 10x7 inches, when I enter the equivalent pixel values, AfPub just changes them to inches. Same for margin, 36px gets changed to .12 inch. And the artifact remains. More to study ...
I posted this over on Alfred's forum (at least we know who's running that one) but no replies in a week, so trying here.
See image below and attached AfPub file. Basically there's a transparent topmost rectangle, which contains an opaque rectangle whose lower left corner coincides with that of the container, and also contains an image below the inner rectangle. This is AfPub 2.6.4 on Windows 10.
![]()
The positions of the lower left corners have been entered using the Transform panel.
Preview mode is enabled.
The image layer has been enlarged so that extends outside the container on the bottom and left.
Thus the lower left corner in the image should show nothing -- the image is cropped by the container, and the entire lower left corner is obscured by the inner rectangle.
However, as seen in the image, a bit of the image is leaking. I know it's the image, not a border, because if I reposition the image so that lighter colors are beneath the cropping rectangle, the color of the leakage changes.
Of course, the intermediate rectangle is important here. Without it, a tiny expansion of the image would not be noticed.
The phenomenon is sometimes fugitive. In this case, changing the view zoom level may cause the leakage to appear both left and below, or just one or the other, or not at all. The same applies to an exported PDF being viewed in Foxit PDF Reader. (I haven't tried other PDF readers.) If it shows on screen, it's pretty certain that it will show in print, perhaps because the lower resolution on screen allows it to hide between pixels.
Can anyone explain to me what's going on? Is it a bug? I'm well aware of the issues with imprecision in "real" number representation -- I learned that in 1966, in my first programming course. I don't know in what units Affinity stores the X and Y positions internally. Still, if that's the issue, I would think that programmers would have been dealing with it since the earliest days of graphics software.
Or if not an explanation, a better structure? I know from experience that it works to move the intermediate rectangle so that its corner is outside the container, but that's a kludge.
Background: I've been using the picture frame + caption described by Jeremy R at https://forum.affinity.serif.com/index. … be-blind/. (I have made sure that thread is archived in the Wayback Machine.) I used it for about 30 images in a newsletter last February, and had to look carefully at the proofs and make adjustments to eliminate these lines that were appearing in many cases. I have an expanded write-up of Jeremy's instructions, which I want to post to help people (like me) who need significant time to understand the original. I'm just about ready -- I even have an assets file to go with it -- but I'd really like to modify the technique to address this issue. I've presented it here using plain rectangles rather than picture frames just to rule out picture frame features as the cause.
Edward