dmstraker

My experience as a user and a one-time coder is indeed users often know what they want and also 'it's complicated'. The worst development approach is probably to focus first on the loudest users. Better to listen and think before saying yes. It can be good to have a stream of work delivering easy solutions, yet this should be constrained lest all you deliver is a can of little fixes without some bigger rocks to wow the crowds. In other words, listen widely and prioritise based on customer priorities and effort required. Also pursue competitor-busting innovations that customers haven't thought of yet and which will drive stepwise sales. Check out also Kano needs: https://en.wikipedia.org/wiki/Kano_model

A few: Separate colour selection tool based on the system used in HSL Live Apply Image (so I can go back and fix formulae) Better use of multi-core processors, especially in image transformation (I've an I9 and things like Apply Image are still slow) Click the 'x' in Develop Persona to kill image rather than having to use 'Cancel'. This confuses noobs and irritates the forgetful experienced. Scripting to compete with Photoshop Actions Plug-in 3rd party personas ability (or option to show Plugins this way) Smaller .afphoto files (better compression?) HSL as native colour model alternative to RGB, CMYK, etc.

When working with macros, especially on a touchscreen it's really to accidentally set off a macro when you only intended to move the cursor or scroll the macros. This is a real pain, especially when the macros do a whole bunch of processing that takes a while. Not wanting to be a party pooper for those who like one-touch, so would be really nice if you could set a preference to double-check macro execution with an 'are you sure?' dialogue. This would also catch the double-jeopardy of accidentally touching more than one macro so setting off a chain of them! Thanks!

I know it's picky but it would seem to make sense for PC/Mac/iPad to harmonise where it is feasible (understanding the different interfaces, of course). I'm an old coder and quality engineer and sometimes unnecessary little things just get to me. I have PC and iPad versions and here are just a few of the niggles in the shape menu: Sequence is different, top to bottom. Sequence is illogical on iPad (more sensible on PC), eg. with the infrequently used square star near the top. Double star is missing from iPad. PC spelling is Donut but iPad is Doughnut (I'm in the UK, btw). Hopefully not too difficult to sort out in a future release. It would make an old moaner very happy.

The Tab to see the image full-screen is jolly handy. But to see it fit the screen you still have to hit Ctrl-0. What would be really nice would be to be able to do a Tab and see the full image fitted to the screen. Easy enough to do with something like Shift-Tab. As an extension, also, you could also resize and relocate back to the previous position (eg also with Shift-Tab). Diolch yn fawr!

Yup. I was a programmer too. Assembler, C, ... And good scripting would be fab. Maybe in 2.0? Meanwhile the round/clamp algorithm seems a feasible option (unless anyone cares to shoot in down in flames).

I've been tinkering with conditional expressions in Apply Image. Ideally you'd do something like (conditional-expression?A:B) so if conditional-expression (eg x>0.5) evaluates as true then the result is A otherwise B. But they are not natural, so my current thinking is: roundup(clamp(expression,0,1)) If expression evaluates as negative, then the result is 0 otherwise 1 (including if it is >1), so you then multiply the whole thing by what you want to actually end up with. The 'else' alternative would be found by adding a similar term with the expression reversed. eg. DR=roundup(clamp(SR-0.5,0,1))*SG+roundup(clamp(0.5-SR,0,1))*SB This should set red pixels to equal to green pixel if the red pixel < 0.5, otherwise it should be blue. This, for example, lets you play with DIY blend modes which flip formula either side of 0.5. Untidy, I know. Does it make sense? Any improvements?

Most excellent and thoughtful conversation and ideas. Some open questions (no judgement, just curiosity), numbered to facilitate replies: Grey layer needs contrast mode blend, such as Overlay. What is the difference if you just use a transparent layer and no blend? Or just a transparent layer and Overlay? It would seem you can still revise painted effect with lower opacity brush or simply erase brush (easier than grey brush). When using curves, eg, darkening pull-down, there will be a greater effect in mid-tones. Helpful if this is what you want, though decreasing effect towards shadows and highlights. What if you set the curves first (any shape) then moderated these in masking either by selective black erasing or invert and selective white paint-back? What is the difference between using Luminosity blend (which could impact Opacity usage) and setting the colour space of the curve as CMYK Black or LAB L? What is the balance between manual, painted dodge and burn and using various forms of generated masking, such as Blend Ranges luminosity constraints, Channel-based masks, vector shapes, general selected area, and feathering/blurring to soften selection edges?

When saving a macro into a category, the macro name is scrunched up on the left -- should use the real-estate available for easier reading. Cheers!

Hue multiplies with Saturation to cancel out the min-max numerator and denominator. Just channel maths, not creative blending as such. H=(Sx-min)/(max-min) S=max-min so H*S=Sx-min then L=min so original value (in combined layers) = H*S+L ... And yes, probably some trunc/round error? Maybe accuracy of float as values are 0-1. Perhaps it does the sums in 8-bit?

Macro set now done, including new(?) RGBW:

Attached is a macro set for splitting RGB image into separate layers for a number of colour models. RGB: Basic separation RGBW: New model! Like CMYK but Red, Green, Blue and White RGB-W: Like RGBW but only two layers for separate colour/tone editing CMY: Effective inverse of RGB CMYK: Standard internal model CMY-K: Like CMYK, but with only two layers for separate colour/tone editing HSL: New model! Hue, Saturation and Luminosity on separate layers! HSL-RGB: Same as HSL, but with Hue broken down into separate RGB layers Principles used in doing these: Start with 'Merge Visible' to create flat layer for base (saves tripping over adjustments) End with single layer group (so you can delete it in one go if you don't like it Named layers, so you know which is which Keep it simple (just the layers, no 'edit ready' adjustments such as curves, sharpening) Fast as possible (so RGB done with channels, CMYK with curves, etc), though new ones need Apply Image which can be very slow! Any bugs, etc. please let me know. Do be patient with Apply Image adjustments and don't try to 'Esc' out as this can leave you in limbo. Enjoy! Dave's Colour Models V1.afmacros

You can create groups, though it's non-obvious. Select bottom layer of group. Arrange/Group Select next layer up. Arrange/Move inside And so on. Currently working on macro set around this and more. May release today if I finish it.

Thanks, John. Post fixed. Yes, I wondered about the divide-by-zero problem. But APh seems to cope ok. I tried it on a monochrome image and the Hue was same mono, Saturation black and Luminosity mono. Difference'd with original showed no abnormalities.

Here's the result of a whole bunch of experimenting over Christmas and the New Year that led to a way to not only create separate Hue, Saturation and Luminosity layers, but also how to recombine them back to the original image. As a result you can now separately edit each of H, S and L, applying the full power of APh to each. In summary: Duplicate twice (three times if you want to keep a copy of the original) Rename Hue, Saturation and Luminosity from the top in Hue layer, Filters/Apply Image, Use Current Layer as Source, Equations (check), then set: DR=(SR-min(SR,SG,SB))/(max(SR,SG,SB)-min(SR,SG,SB)) DG=(SG-min(SR,SG,SB))/(max(SR,SG,SB)-min(SR,SG,SB)) DB=(SB-min(SR,SG,SB))/(max(SR,SG,SB)-min(SR,SG,SB)) Apply, then set Hue layer Blend Mode to Multiply In Saturation layer, go to Apply Image as above and set each of DR, DG and DB to max(SR,SG,SB)-min(SR,SG,SB) Apply, then create a group for Hue and Saturation layers, and set the Blend Mode to Add In Luminosity layer, go to Apply Image as above and set each of DR, DG and DB to min(SR,SG,SB) Now you should see the original image. Edit the HSL layers as you will, noting: In editing the Hue layer, you change basic hue but not saturation or luminance. This may be a bit confusing and seem like nothing is happening if you're not careful. In editing the Saturation layer, this is not a luminance layer, so don't try sharpening it. Rather, make areas darker or lighter to change colours to be greyer or more vivid. In editing the Luminance layer, you can apply noise reduction and sharpening here, rather than on other layers (unless you're getting really creative!). Further note: if you want separate RGB hue layers, you can make, for example in the Red layer, DG and DB zero and then Add and group. Note for the purists: Yes, it may not use the 'proper' algorithm, but it is a working, practical solution. And here's a video of it all, from my InAffinity YouTube channel:

When creating a gradient, it appears to be dithered. Fair enough for photo work, but sometimes you want a cleaner gradient. For example, if you create a b/w gradient then posterise it (to create a step wedge) the edges between bands are not clean. This happens in PS if you've got the 'Dither' checkbox checked when creating the gradient. Could we please have a Dither checkbox when creating a gradient, so we can turn this off if we want a cleaner gradient? Thanks!

Regarding my InAffinity YouTube Channel, I've just had the below in email from YouTube. There's now around 350 total videos uploaded on Affinity Photo... https://www.youtube.com/channel/UCOnLUmyPHr2rayOHVHWsHVw See update video for update, plans and subscriber comments. https://www.youtube.com/watch?v=apgypHgxcmA Dave ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ InAffinity for Affinity Photo Your 2019 Year in Review You uploaded 218 videos and gained 2371 subscribers, which goes without saying, you rule. Thanks for making this year one for the books. 13,349 Total Watch Hours Fans can’t take their eyes off you! In grams, that would be like eating the recommended serving size of 16 bags of Flamin' Hot Cheetos. 10,703 Total Likes 203 Liked Selecting the Sky Using Channels in Affinity Ph... 236,681 Views

Add white shape above image, Screen, with extreme gaussian blur (r=1000). Gives discoloured lines as below. It's not rendering (export to jpg shows it too). Experimenting shows lines appear when blur radius gets to about 500px. .afphoto file is 484mb. I can upload to somewhere if you like.

HSL. Select a colour, then the picker. Dismiss HSL box via top-left 'x'. Cursor is still in picker mode and appears still to be operational. (should return to normal cursor after HSL box dismissal)

I've done a few YouTube videos on Custom Blur: https://www.youtube.com/results?search_query=inaffinity+custom+blur

Try the attached. On the face of it, it looks like a simple image with Curves adjustment. But when you play with the Curves it doesn't behave as it should (I've tried it with beta and 1.7 release). For example pulling Curves up to the top left does not result in a whiteout (as happens when the original jpg is used). History: I'd just been playing with masks via the spare channel. Duplicate, Threshold to create b/w. In Channels, right-click and create spare channel. Then load to curves. Then load from blank channel to remove mask. I've tried repeating this to no avail. The bug however seems captured in the .afphoto. Maybe you can figure it... before.afphoto

Thanks, @carl123 I had Affinity hang after trying to select two layers in the usual way. On APh restart, the macro panel had the 'record' button disabled and 'stop' enabled. No great shakes, but probably a bug.

Olivio shows useful tips, but I'm not sure how this would be used for unidirectional blur on irregular-shaped object. Yes, it is a sharp edge. It's like using second-curtain sync on a flash photo.

Thanks, though a little academic at the moment as Affinity, in their wisdom, decided to remove the HSV checkbox from the HSL dialogue in 1.7. I've moaned and they said they'll restore it at some time...

Yes, thanks GarryP. Just did this on the leading edge. Yet a checkbox would be so much easier...