Frank Jonen

I'm using 32-bit almost every day in Affinity Photo. If you only know 32-bit from Photoshop you'll find a few things different here. Besides ICC sRGB Linear, you have the option of viewing and translating your image through OCIO transforms and exporting with the OCIO "profiles" attached to your EXRs (add the name of the OCIO transform to your file when exporting - daft, but works). The most popular OCIO configuration right now is nuke_default, it's most compatible with regular sRGB / Adobe RGB workflows. You can download the config here: http://opencolorio.org/configurations/index.html If you work in an ACES workflow with other facilities, ask for the version or have them send you their config. You can just use these in Affinity without any extra work. That goes for Designer and Photo. So you can do complex vector work as well without leaving your project's colour space. Particularly great for making displacement maps with vector shapes, design and export in 32-bit float and no worries about stair-stepping (unless you use ZBrush, 16-bit integer limit there for alphas, maybe R8 has 32-bit now, but R7 still hat the 16-bit limit) I'm not familiar with Vray outputs as I haven't used it yet. But it looks very straight forward from their website. Set it all to Add and hope for the best. So far I've dealt only with Modo and Mantra renders in Affinity. No reason why Vray shouldn't work, especially since it's all additive. Select all layers and set it to Add. For the advanced mode you could experiment with setting a group to Add and blend layers within that group. The output of that group will then be additively blended. If your image looks too dark, the sRGB gamma is baked in. Set the viewer to Gamma 2.2. Looks right? > Add a Levels layer on top of that stack and set the gamma to 2 and another Levels layer set to 1.2. Levels top out at a value of 2. If the image is too bright, same dance, different shoes. Set all to Gamma 0.454545 (it'll round to 0.455) and you only need one Levels layer for that. In OCIO you set sRGB to Linear or Linear to sRGB for the same thing then. With one exception. ACES doesn't yet have a path back to true ICC sRGB, I'm still working on that. Blur is just a blur, it's not a defocus node, so it only blurs things. Affinity Photo has a defocus node, but it has no inputs. All channels come in as regular layers. You need to make your own masks / alphas from them. Except your alpha, which you can set to: Delete the masked area (like Photoshop) - Perturb zero EXR alpha. (perturbed describes how you feel after using it) Mask the mask area (fill alpha to white to get it back) - Associate OpenEXR alpha channels Leave your file the fk alone. - "Uncheck Everything!"™ - That way the alpha is just a regular layer. I call it "serenity mode". In some cases the un-premult mode can be useful, you get that with "Post divide EXR colours by alpha" I use the Depth channel this way: Duplicate it make sure I got the right 0 - 1 ratio dialled in Layer > Rasterise to Mask hide Whenever I need it, I make a duplicate of the mask and add it to a group (as group mask), then do the atmospherics for example within that mask. Now IF we had symbols available in Affinity Photo without having to go through Designer, we could have ONE mask and instance it. Hi there @Andy Somerfield There are probably reasons why Symbols aren't in Photo, as I manage to crash Designer on a regular basis by keeping on editing them. Now imagine people putting a whole filter stack into a symbol and re-using that on every other layer group. I hope that helped a bit and didn't make it more confusing. Affinity just has more options in that department than Photoshop.

Yes, with a tailored test to just one tabstop, it sort of works. Now try with a real world scenario. Word . . . . . . . . . Centred Part . . . . . . 1234.56 _ _ _ _ Right Aligned Word should be Left aligned, Centred should be aligned centre, numbers should be aligned by dot, right aligned should be aligned right. Instead it does absolutely nothing.

That's pretty much what I suspected. I think it's something that'll need to be addressed somehow, otherwise new users will constantly ask about it in the future, creating support overhead / frustration. The way I ended up with the inflation was when I tidied up the project and aligned all artboards evenly by distributing horizontal centres. So a way you could deal with this would be to tell the align tools to only use full pixels when dealing with artboards. Another option would be to only allow artboard edges at full pixels. That way it doesn't matter what any tool does to them. I haven't used artboards in the Adobe apps much before because of how annoying the apps started to behave with them in use. So my expectation set is: Artboard = document, artboards = documents within a document. Each with their own 0,0 point. So given an own 0,0 point, the path from 0 to 1024 should end at 1024, no matter where the artboard is placed. Placement of the artboard on the infinite canvas should not be able to interfere with how it is converted to a slice. That just makes no sense from a user standpoint as the "sub document" itself is pixel accurate. The only thing that does is create frustration.

Bug: Typography Engine, General Tab stops. There's no visual editor for placing them. It's guesswork. (AKA: Not good) The fill characters don't fill. Just one instance is placed. Edit: Alignments in tabs do nothing. Glyph browser distorts all glyphs to a square preview box, already reported by someone else in last version IIRC.

Bug: Export Persona When you set artboard positions to non integer values, the slice tool will get the values wrong. You'll get 1025 x 1024 on a 1024 square for example. Set it to an integer position and it gets it right.

Yes, I was expecting it to commit, but… your view brings up an interesting option. Why not have all scale, rotate, shear, etc. operations as modifiers (maybe with a toggle for cases where modifiers would be overkill)? This could also help with getting the vector warps done quicker, I reckon. Also helps with iteration. If you have modifiers exposed to the user in the layer stack. Similar to the fx or adjustment layers. That'd also make it possible to quickly apply complex transforms to a lot of items at once, maybe even store a modifier stack in a symbol-like fashion. Once you're good with how something looks and want to refine manually, click 'bake to object' and it's applied to your object. Or hold alt + click and create a new object with the modifiers applied while hiding and thus preserving the original.

Is the "Reset Selection Box" supposed to bounce right back when the object is de-selected?

Indeed there are. If you're doing stroke gradients it's absolutely maddening to constantly bring up that tiny widget at the top of the screen. After a while I just gave up, used an online gradient generator and brought it in with an SVG. The "creative" solution was for the birds, I can't actually see the gradient behind the blue overlay. Even when I make the stroke the wrong size just so I can see the colours during editing, the overlay distorts colour perception, so does changing the line width. I found the on-document solution to be quite terrible for a lot of use cases. You can't type in percentages of stops, you can't just click through and have the next stop placed at the 50% point between the stop you just placed and the next. It's a frustration bath for so many things. Everything has to be made in tedious manual steps and with guesswork. The time that takes just isn't available for creative tasks anymore then. This caused me to look for other options, like the various online tools for gradient generation. Also a thing that'd be useful with gradients in general but nobody bothers with is gamma lock or in HSL/HSV a saturation and luma / value lock. That way a gradient would always have the same saturation and lightness while blending across hues only. Right now it's a matter of checking and re-typing on each stop. We can edit all the values, why not put a little lock icon next to them? Linking would be useful as well. An easy way to bring up the values of an HSV gradient all at once, or decrease the saturation. This would save an additional adjustment layer. Adjustment layers can become quite expensive when part of a symbol group. This would be a simpler way.

Lines can't be turned off for active objects. Problems this causes: Accurate colour decisions can't be made while an object is selected. Myriads of unnecessary clicks to constantly deselect, select, call up the tool again, adjust the value, da capo. Accurate placements to non snapping objects because it's impossible to see past the UI edge overlay. Given the problems this causes, I think it qualifies as a bug. Also arrow key movements with the hand tool selected were once possible (so the outline overlay isn't in the way), but then it got broken.

Also can we have the Gradient panel detachable? It's extremely annoying to always have to mess about in the top left corner of the screen in a tiny area instead of where it's comfortable, like right next to the object you're working on for example. Also for complex gradients it should be able to scale that window/panel up ridiculously large. Especially when you have to simulate physical materials in a texture that have hard stops.

I'd much rather see implementations like Resolve or Baselight.

It just lets you move a point, not extend from that point on to a new node. Two different things. One is a move operation, the other generates a new bezier node / line point at a specified distance and angle.

One tool that's indispensable in CAD is line extend. It's something that's sorely missing from illustration tools. You select the tool Click on the line Drag in the direction you want to extend OR enter the value you want it to extend The tool knows what angle a line is at and just extends it. It'd also be great if we could enter an angle. That way we could extend a line by say 33° to the line at the chosen anchor point.

Splitting a closed vector shape by defined value (standard math applies like in the other value fields) or percentage slider (with and without symmetry) along its base shape. Complex shapes like a 'Q' could have multiple user-selectable paths which can be exposed and manipulated by the user for fine adjustment. Or simply roughly drawn in, the next perpendicular lines a tracing 'ray' fired from that line would create the 'stem', so the tool itself can apply centre lines. I made a rough glyph shape in Modo to illustrate the basic functionality. The tool should have the option to either create a new path (for overlays) or just cut the shape in question.

"Lock children" gets forgotten upon save. Just accidentally distorted an image because I set the container to 'lock children'. You open the file again later and all checkboxes have reset.

Do you think it's possible to add a Slope, Offset, Power, Saturation control layer (CDL) that also allows .cc XML file input? Attached you an example for an exchange file here for reference. match transfer.cc.zip

A bit of a list for the current beta (3). Bug 32-bit features When stitching 32-bit images the colour / gamma information gets messed up. Also there is no OCIO support in the stitcher, so it'd be hard to see what you're looking at once the colour flattening bug is solved. Wishlist 16-bit half-float It's one of the cornerstones of the ACES spec that some facilities require. So an input / output would be great. For input it should be recognised as 16-bit half-float and be treated as such, not as 32-bit float. It would still be in a 32-bit float space inside of Affinity but properly imported and exported. I'm coming across this now more often with ACES work. Some game texture artists use this format in their pipeline as well. So it's a broader group that has a need for it. OCIO implementation An option to read the colour space from the EXR would be great and more reliable than do it by filename. Also have an option to set it during file export. This could be simplified with the OCIO settings file I suggested before. Setup once, fire and forget. It would also solve the current issue with EXR export crashes when a space isn't specified in the filename.

After having used the 32-bit modes almost exclusively now since we had them available I noticed some issues with OCIO configurations and maintaining them. What I'm suggesting is a more global config file. I've attached a YAML example. YAML because it's easy to read, duplicate and modify. It can be easily exchanged when working in groups. Download project config from Shotgun, merge with own config and use it right away. Affinity currently has a config switcher that doesn't switch configs. This way it would work and it would be easy for the user to manage. The ICC part is for Affinity to determine which profile is used for converting OCIO data to ICC data for delivery use to devices and services that don't use OCIO (print shops for example). ocio affinity.yaml

Library and Macros are in dire need of fixing. Either they render the app unusable or upon export can't be imported.

You can use live booleans instead of the shape builder. Affinity calls live booleans 'compound'. You can find them high up in the Layers menu.

Is the Symbol bug already being addressed? I mean the one where Symbols that worked fine in Designer and Photo (App Store versions) go completely out of whack. PS can we have plaintext variables in 1.6? It would really help with label making automation. Example: Part of a symbol's text is a variable that's styled differently across a couple of different symbols, but the text is only changed once. Currently that's not possible.

Can you look at the clone stamp functionality again? It would be much easier to use if the crosshairs in source mode would work in an absolute way rather than this weird sort of relative way. Also an option to toggle a straight line between the two would be a great help with aligning the source to the target.

This screenshot from the Autodesk Sketchbook iOS page outlines nicely what I'm talking about. Symmetry has already been requested numerous times. This is more about the foundation before it becomes symmetry. Have an option allowing us to define out own symmetry tools (click line, define what it should do) Tie it on top of the coming replicator system (blends, particles)This would allow us to clone and array (radial, grid, etc) the centres of symmetry as an object to create clock faces like the example with two stokes. Car tires with a few more strokes, and so on.

Talking about 32-Bit: A lot of things just break in the beta when you work in 32-Bit.

Ah right there's that. I rarely look that far up the screen when working so I haven't noticed. I guess I just figured devs don't put important stuff away from focus areas anyway. Especially when it's so blurry you can hardly read it. The "used" category seems to be what I'm after. No idea why it isn't in the characters panel where it belongs. Makes no sense to just have it in that blurry mess tucked away up top.