kirk23

Or better give as a tool where we could do our own pixels displacing/shifting effects. Procedural filter update for example that could sample RG channels pixels of beneath or whatever second layer for direction/vector and B channel for distance of shifting . That way we could do ripples , distortion, displacement or whatever kind of effects on our own.

They could introduce select invertion in layers pan at least. So we could select a certain tag, invert selection and uncheck other layers. Or simply "isolate" same tagged . Surely it wouldn't take a lot to code.

In layer panel so we could hide layers other than having certain tag

A brush with multiple bitmap dabs like the one in APhoto but keeping connection to vector spline and editable . one click style replaceable etc. Like in Xara

Right now we have 3 parts. Start, end and tilable body. Let it be extra up 10 body parts at least with ability to alternate randomly alone a curve and wider window to setup it. Like it was in Creative House Expression and older Serif soft too.

I would be happy if just their "procedural texture" filter could sample different layers as inputs and re-map them using another layer input to deform UV. Basic image editing operations actually. So we could make our own displace. wrap, smear filters . Like wrap and directional wrap, slope blur in Substance Designer. Without those few very basic features even nodes wouldn't make much difference. If they can't make a useful live filters why not give as a tool to make our own. I'd love we could use Substance Designer export files as live filters here but my guess Adobe would never allow this.

So if I feel I need a different brush I wouldn't have to set all the parameters again? Like in Xara? ps , Solved it it's Alt+click

You can use free open source Inkscape to turn any image into vector silhouette Path>TraceBitmap . And it does it a bit better than Photoshop actually

Ctrl+Alt or Ctrl+Shift when you select a handle by node tool

How?

I had Xright checker once and then misplaced it somewhere. From what I remember it still had been a learning curve. Starting from basics what ICC profiles are. Recently I bought a cheap $15 color checker, no software and it somehow works with LUT layer right away. You make a picture of it , then edit color charts to colors values it should be and use it as a target second image during LUT building. But again I have no idea how precise it actually is. Perhaps less precise but still looks enough to extract albedo diffuse color from something in blue late afternoon shade for example , A reason I use the checker for.. I am not saying an ICC solution is redundant , I just found this workaround and it seems very simple an cost effective

I am not against it actually. It just requires a special soft to profile your camera. hours of manuals reading , hours to figure out how icc profiles works and what they are so you wouldn't set a camera profile as your system one. While this LUT building approach from original picture and another picture with color chart values set by books as a target seems amazingly simple to do and moreover to understand . I just adore a simple solution for a complex problem. But I don;t know really how precise it is vs ICC profile

I truly believe RAW development should be merely exposure range choosing to clip and all . Or not to clip straight to 32 bit file . Never understood why people tweak colors in RAW. It's so much more convenient in the soft itself with adjustment layers. You have LUT adjustment layer where you can make your own camera/ illumination or whatever specific correcting preset . Aphoto have a great advantage over Photoshop with LUT building tool from original and target image pair . Works for me with color charts images as a home made camera "profile".

I'd wish they better be focused on improving procedural filter and layer system. transform links , more live filters. more macro functionality etc rather than burring themself into AI studies. For those who need them there is Topaz. I see zero AI selecting improvement in last Photoshop release either . it's just brush vs lasso, same as before . Working same useless most of a time and selecting only something easily select-able without AI , no problem. Any complicated matter and it's too inaccurate. Surprisingly Photoshop has all its neural filters also totally useless . Super zoom maybe be the only one somewhat helpful except the fact I don't need zoom , I need more resolution and blur killer without damn zoom. I hoped color transfer might be of help. Nope. I did one that makes a seamless collage/ fuse in for any object put over on top of an image on the fly with a simple like 2x2 math matching hue and tonal ranges in Substance Designer . Neighther APhoto nor Photoshop have such simple basic convenience or live filter . All those neural things do whatever except what you need.

The problem is they need to add "sampling from other layer" capabilities to procedural filter. Like some others that can do it "from beneath" or image And ability to remap pixels by other layer/image input like some other filters like Displace . Basically combine features that already have been codded in one place for more flexibility. ps. for moving Layers there is a painful workaround of using symbols brought from Designer, locking symbol containers and then "pixel content" linking to inside of one of symbols from another layer , That way we have something like chain links in Photoshop to sync moving of layers . It's crazy. I am asking to add transform links to Link panel for a year already.

Or make your "equation" filter sample other layer channels. One for distance and another two for direction. Displace is a very basic feature you do with pixels and has so many uses when work properly.

A mask for example where we could write a math formula like LayerA + LayerB * layerC executing the calculation on the fly. That way we could get rid of super complicated nested groups and mask stacks . Gazillion of layers and linked layers you can easily lost among. It would keep the layer stack simple and easy. Let you compartmentalize the whole construct in a simple way without a complex naming and tagging system . A way to keep essential part as layers and technical stuff hidden in expressions. Perhaps as a part of procedural layer interface

if we have our slices remember each own layer compositions I think it could be possible with one click. One more reason to have it

I have just figured out that by recording a switch to 32 bit mode and back in a macro we could get many filters working in a new perfect manner. So I am trying an ancient trick with producing normal map RG (xy) channels by embos filter and B (z) channel by edge detect filter inverted. It all works just fine except gamma . Switching back to 16 or 8 bit makes a gamma shift and I am struggling to figure out the right values. Can't get a perfect 128 (0,5) value for front looking pixels in RG channels. Any suggestions?

That way we could do our custom blend modes . "fill" vs "opasity" depth based calculations and a huge number of our own filters. Some inputs like a = Layer1 Red remap pixels would allow us do do custom image distortions . i.e unlimited number of filters we could share with each other. Could be used to simplify layer stack .Make it elegant and easy to read hiding all the math and blend calculations not essential to represent as layers within the filter

I have version 7 or something. It's very slow and main drawback IMO it can't read other layers input or work "live" . I recall I have it working live in some old Xara version on vector shapes re-rasterizing content every time I tried to edit the shape. Also super slow. Does it still work in same manner? Affinity live filters work so quick . it's pity we can't make our own

Yeah, it's a reason I am trying to invent something from PT one. The displace in current version is totally useless for me . Looks like it works with kind of sin function or something inverting vectors/ displace direction against a center of an image, Equations my guess can move pixels but can't sample other channel or other layer to get values , right?

Basically using Red channel as displace value to GB channels? It does't just fine with noise generators but could it do with an image channel ? Also can it sample a layer beneath like some other live filters?

Here how same imported tga or tiff looks in Photoshop an APhoto with alpha off. Isn't it an import bug basically? I think they should fix it asap before any exotic new formats

Plus an ability to explicitly set bitmap size in pixels , same in Designer