Frank Jonen

Here's the thing. Affinity changed the way I work. I'm not shy of using raster based effects anymore as they're handled well across the apps. It creates a problem with PDF output though. Text outlines inevitably blur when they are anti-aliased against their background. To a point that can be addressed by exporting with different resampling algorithms. But some text just has to stay vector based. Right now, both apps just crash when I export my layouts to PDF. Stacked LUTs, live filters, item blurs, mixed media items parented to text objects… it's a problem factory. Ideally we could have a way of tagging a shape or text layer and say: I need this to stay intact. The app then "does stuff" to rasterise unpremult, duplicate the shape and whatnot to make sure the tagged objects come out as vector shapes in the PDF. It can all be flattened with the text outline slightly eroded inwards behind a vector shape even. It's similar to what I'm doing right now. Export my document to several layers and put them back together in a simplified way, re-nesting flattened elements. Another option would be "output layers" for lack of a better term. Like groups you'd put items into them in the way you want the output to be organised. Everything in these would then get flattened, retaining the original appearance by baking in blend more results into the objects so they all just have an alpha and nothing more. In return this makes it easier to retain vector shapes. Either via dilate, bake and nest or by pixel tracing how the vectors would be affected by raster effects. In some cases it could even be sufficient to render these effects as a mesh gradient fill or a set of vector shapes blended together.

You can use the fill tool and fill with 'history'. Then mask it. I've never used it, but it's there.

Masking bug Selections made in masking mode are persistent, invisible, can't be deselected and are saved to the document. Have a pixel layer with any sort of stuff on it Move to masking mode (the one with the red overlay) Draw gradient to mask portions of it Select > Deselect Save Open again in either Designer or Photo Be frustrated that nothing you add to the document is visible unless in the non-masked area Deselect still not possible Only way to deselect is to create a new selection via > select all and deselect that one

Inpainting brush bug Inpainting just stops working and you have to restart the app to get it back again. Here's how to recreate it. Bring in a pixel layer to a comp, so it has an alpha on its own layer (RGBA) Nest another pixel element to it Use inpainting brush on the parent layer - it works Create a new pixel layer on top - set inpainting to (Current Layer & Below - it keeps forgetting its settings) Paint on the new layer and it creates a direct clone of what is below, no inpainting This behaviour is now default until you restart the app. System: Yosemite 10.10.5

Great. I'm still on Yosemite here (have to, Nuke isn't certified to run above Yosemite).

Here's something that's bothersome for me with the output to PDF. I have a vector shape, within that shape I have bitmaps and other vector shapes nested. A couple of LUT layers and bitmap comps later and I output to PDF. The bitmap sections itself are cut up pretty well by features, so that's cool. There was pretty much zero waste there. However I lose the vector outline of the base shape that the other stuff was nested in. This increases the file size quite a bit since it introduces an extra alpha to the file. The PDF output should retain the outline and just have PNG portions overlaid on top for transparency. It could be that it's too hard for the app to figure out which outline to retain in a mixed media comp. I'm way out of Illustrator territory here, so I don't really have a comparison. If that's an issue, maybe we could help it along by tagging a shape with 'vector cut' for example. Which then would also carry through to any bitmap format that can carry outlines (to flow text around and make vector knock outs). We can already exclude shapes from snapping, maybe that could be added to the same menu section and be displayed similarly in the layer stack. (Probably something worth having anyway). Edit: Nesting the layer group in another shape retains the outline as vector in the output. But then I have both, the PNG alpha in addition to the new shape which bear the potential of a moiré effect.

Not right now, but if you want so see the bug in action. The easiest way to make it happen is to click on another app in your dock or to TAB out to another one. Once you come back to Affinity Photo, the dialogue pops away and from there on out you can't get it back.

Load a LUT, bring up the file browser, then click on something else, so the file browser loses priority and becomes invisible. Affinity Photo cannot recover from this state and requires a restart. Otherwise every further LUT layer in this session will not be able to bring up a file browser.

Select a path, click on "magnetise" and as long as "magnetise" is enabled, the set origin point (Transform Panel) will stick to the path. Example: You want to align a lot of elements in a perfect circle (or other shape). You create a shape, magnetise it. Then select the group of elements you want to align. Set their origin point where you want them to snap and drag them across the circle. They snap into place. Now distribute vertically and horizontally and they distribute while sliding along the shape's path. A "distribute on shape" would be great too, it poses a problem with complex shapes though. Maybe a "user do you know what you're doing?" warning would be sufficient there.

Also an interesting way of keying. Right on the vector scope with bezier shapes.

I saw this today and it would seem a great idea for Affinity Photo. Possibly in the develop persona where it's more about interpretation of greyscale data than RGB data.

There's hope. FontLab is getting a better interface in version 6. More and more people move to Glyphs though. I haven't given it much of a look myself but from what I've been hearing it's fast, makes common tasks easy and is less tech focused and more creative supporting than FontLab Studio (which is very tech focused). There are more options now to get started.

In general I keep what I like. When I use a font from a dead foundry, I tend to fix it myself. When I like a font but it's crap, I see if it's worth fixing it alltogether ot fix the parts I use, when I use them. I have a lot of good TrueType OpenType fonts and some bad PostScript OpenType fonts. You can't generalise by extension what quality or versatility a font has. While TTFs type of hinting is more screen orientated, Post Script works better on film rasterisers for print. These font sync apps don't take up much CPU, but install Little Snitch (or otherwise monitor your outgoing traffic) and you'll rather quickly uninstall some of them

BTW, I'm editing a piece of lettering with a crap font. Someone thought it's awesome to place 7 nodes seemingly on top of each other. It'd be nice to have a counter for nodes as well. Not just items. With Google Fonts for example the availability of really terribly made fonts has increased. Tools that help coping with that decline in outline quality would certainly be welcome.

Not sure if a bug. Bezier tool: The right control point isn't snapping to anything. Left control point snaps to guides. Feels like a bug since both should be snapping. Smooth or smart options. Happens with both. All snaps are on. Full Schnapps ahead!

Possibly the most affordable variants out there for pro panels. And they have an API. http://www.tangentwave.co.uk

That had happened already. Adobe famously told Pixar engineers to fk off, that they don't know the format they invented (EXR) to begin with. Now we have a similar issue with ACES. Adobe doesn't care. They have Adobe RGB and Kodak's ProPhoto RGB and that is enough for them. Which is understandable. Adobe created this island where it's hard to get in and out of by purchasing post production tools. So a lot of people just stay in there and accept it. That's why I'm so glad we have Affinity now. Which had the downside that I have to fix textures and mattes that don't hold up in 32-Bit space exposure changes because they were done in Photoshop. Not fun, but as more people move over that is going down as well. As for the compositing workflows. As long as it's documented how channels go together I'm fine with it. It leaves room for continued innovation in how renderers handle data. Chaos Group just made it super easy with how they're treating the output to go together all additively. There's also a book by Ron Brinkmann on digital compositing that goes into why things go together the way they do.

If you haven't already, what you want to do is flatten before converting to output formats. Blending reacts vastly different in floating point spaces than in integer ones. With online photo services I'd stick to 8-Bit JPEGs with embedded sRGB colour profiles. With sRGB you technically don't need a profile, but I've seen things go so utterly wrong with untagged files that I'm never doing this again. Especially with wide gamut displays that are actually set to wide gamut. For these you still need to convert to and embed Adobe RGB or the reds will overshoot. When you see these overshoots, it's not a majority. No need to panic. If it looks good on an iPad, that's what counts. PNG has issues with some browsers still with 16-Bit display getting the gamma wrong. Especially on mobile (Android). If your 32-Bit files look right in ICC sRGB (Linear) - Affinity's default space, all you need to do is flatten the document, THEN convert to 8-Bit for portfolio use. This converts nicely to sRGB.

And Levels. That's the big one for me. I only use it for the Gamma slider (that stops at 2.0), for everything else levels related I bring in a LUT.

I've done the thing today where you duplicate a line of text (one word) expand the outline of text and merge it. Goal is to have a second vector outline that can be shifted. Try it. Set the outline to around 10pt @ 300dpi with like a 100pt text. Set the outline to outside and do as I did. Maybe that example can help with diagnosing the issue. It requires a lot of cleanup and the outsides intersect. Geometric Sans 'T' shapes often bulge out on the sides and top instead of being straight. I vaguely remember similar shape issues from the first version of Illustrator that had a variant of Streamline (yeah, not really) implemented. It was bulging out as well on angles close to or at 90° and close to or at 45°. I always had to redraw these sections back then. Maybe this is similar, maybe not, but there it is.

Just tried it. Does the same here. Another thing. When I start a document in 16-bit mode and go up to 32-bit. The Artboard backdrop gamma isn't converted but assigned it seems. Brightness comes way up when I do that.

I used to have a font like that. Numbers were spaces that matched leadings from my Weiss Antiqua drawer. Top row had various dot shapes with various spacings (lowercase regular, uppercase 2x). Home row I think were underline options with various cuts for descenders. I don't know where that went, but maybe it's time to make one like that again.

I've reset Designer and now it works. Something got screwed up. BTW is there a way to make the text frame being persistent instead of going away constantly once I select another object? More questions: How do I set the character to be repeated, the distance from fill characters to word? The part that the reset didn't fix: I'm still flying blind when setting up tabs, no visual way to see the tabs in context. No invisible characters, no tab editor. Just the little list. What I found in the list: When you confirm edits in the list with 'enter', the entry gets deleted, you have to use tab to go out of editing the value of the tab stop (when you edit it within the list).

Quite likely Vray just comes in as scene linear. Going from Chaosgroup's support pages: If you haven't set up OCIO in Vray, there is nothing there to manage colours. When you output from Affinity you can uncheck "Perform OCIO conversions based on filename" in the preferences if you don't use OCIO (but have an OCIO config installed) to avoid confusions with filenames. Good luck with the renders

Your file works for me, I just can't create these here. I tried setting different languages on yours, no influence. They kept working. Hmm. Weird. Thanks for posting it. For now I'll just keep editing a copy of that whenever I need tabs. Not ideal, but whatever works.