kirk23

Where we could put a spline over an image and make it deform underlying image . Its just a few clicks in Blender for example . Simple like 2x2 . Why not do it here . It doesn't require 3d axis really . Like projecting image on Spline generated UV/mesh and deform it in sync of spline editing gradually within set distance. Would be better that Photoshops puppet wrap which always been a toy not a tool really .

Like displace in couple pixels distance repeated along underlying vector/flow map number of time gradually fading out or not . aka "Slope blur"

Photoshop does it so right . why not here.

Since Photoshop introduced it I can't do it in Affinity anymore. Those options even when look weird provide so much less repeating, copy/paste from neighboring place look.

and ability to make it work in layers alpha too same as with rgb as an option . Better color adaptation of edges with a sliders . Meaning how much blur it applies to channels intensity from original layer and in what distance from edges. Could be so much better that Photoshop both patch and content aware move . No need to fill the previous place with blurry something .

I tried that Generative fill and it always does something low res and never what you actually asked it to do . Still feels like a toy mostly . For cool instagram pictures. I would be more interested in more clever content aware fill better recognizing patterns and less repeating but it's all the same as before . Ai selection is still same annoying selecting random things around that you need manually clean after anyway.

I miss this so much after Photoshop. The only reason I still pay their subscription. This and pattern preview which is much more convenient than pattern layers here. And perhaps scripting. As of every other thing I prefer Affinity Photo. Tried their recent "generative fill" in photoshop betta. What a useless thing. As well as AI selection that selects nothing you actually need .

I love those options in Photoshop. It eliminates those repeating patterns in inpained areas I often see in APhoto inpainted areas.

I disagree. Both Designer and Photo are still seriously lacking in features. I only happy they are not pursuing AI path. Still struggle to find anything useful AI I could use for my actual job.

With a set amount of colors /gradient stops. So as input we have just RGB image then the filter turns it into grayscale one + gradient map filter . Thus we would have color info stored in one channel and a gradient. Working like sort of indexed colors. So we say the gradient should have 100 stops for example and it makes 100 gradation of gray with corresponding stops in the gradient.

I wish there would be "same tagged" and "same named" in auto select options

Thanks a lot NotMyFault The problem is although those 8 bit per depth is not enough , even 7000 value I used is marginally enough actually and still gets stepped sometimes when you convert this depth into a normal map. The original purpose of all this was to composite materials and to get usable normal map too. I am close to give up on the whole idea 🙁 It still would be possible if keep the color and height on totally different layers and in different groups but producing necessary dynamic masking for color ones would turn the whole layer stack into a huge undecipherable and unusable mess. I think v2 with transform links makes it a bit easier but still it will be a monstrosity. I tried it couple years ago.

Well, at least I figured out I can use symbols from designer and a macro that would do 8 extra copies and shift them around . First one up and next ones clockwise if selected down from the top in the layers pane. Share them here if anyone need it. You can save the symbol as an asset and then replace the image with your own. symbol1.afphoto pattern shift 2048.afmacro

it's basically what I try to do but get something weird. floor4.afphoto

Thanks a lot for the example. That expression syntax is something I don't understand quite well. Those spirals are 16 bit integer screen captures from Zbrush . So black pixels are just 0. My guess it could be a proper render from 3d soft where black pixels get deep negative value but my purpose is not to simulate actual deep Z buffer , rather just a small depth to do depth based blending of materials. Small pebbles and a sand and such matters. So having an option for just a subtle up and down and your formula works perfectly. Another challenge is how to make this subtle up and down by a brush with manual input . Something that would multiply on integer part of values or only on after comma part for color editing . So far nothing I tried is working. Ideally a combination of hand painted mask and levels/ HLS filter that would do it for integers /height or after comma part(color) separately. Hoped just procedural "floor" layer after any normal layer would do the trick but seems it doesn't. ps. I added mask on top of the spiral image and it solves blacks RGBGrab05.afphoto My goal is to make a non-destructive version of what Zbrush calls 2,5D canvas.

Here I tried to recreate a technique from Substance Designer used to height blend objects. Each of two layers have its height info stored in integer part of floating point values, and color values after comma. So a pixel with value 7000,34 consist of 7000 which is height/depth info and 0,34 as a color info. Then I use floor operation like R-floor(R) to see only color , like 7000,34 -7000=0,34 in top procedural filter . It all works only if on screen resampling is set to nearest neighbor so pixel values stay the same. Now a question: how can I modulate the height of layers with real time visual feedback. Like sending some object forward or backward by modulating only integer (height) part of pixel values. What should be an expression for this in procedural filter? floor2.afphoto

Thanks walt.farrell it's exactly what I was looking for

or any specific way? I mean not during export but rather for what we see on screen.

I vote for Ai trained to do two things. 1 Shadow removing and de-lighiting . 2 de-noise height maps from Reality capture

It's so much more convenient than pattern layer in Affinity. Best feature in Photoshop for years . I'd love to see it in Affinity photo . Working same non-destructive way for a hole document. Affinity Photo already have affine filter. It just needs to be live as a first step.

I use my homemade method. In LUT adjustment layer I hit "infer LUT" and select original photo of the color charts and secondly edited photo of same color charts where I edited all colors squares to their proper numeric ( target) values. Like spectral red be 255,0,0 etc. Those values is written on the back of charts card. DONE . You have LUT transforming layer/preset you can apply to all photos you did in same session/lighting setup you took your original image of color charts card. IMO it works no worse than automatted X-rite soft for image/camera profiling I used before . Before I misplaced my X-rite color "pasport" charts somewhere and bought non-name cheap charts card

Would be nice to have a node interface where we could just build our own FX . Maybe just using 3d party , Filter forge , Substance Designer or whatever and have it not be rasterized instantly but rather like live filter or FX

Well, I do want but they are pretty useless in Affinity . It splits them into a mess of individual layers and what I want is basically to work with all those info accessible from a single layer . CMYK provides a nice compromise. Would be nice if we could use cmyk brush nozzles too . It's tremendously more convenient when you operate with a single layer at once instead of of a mess of linked ones.

Yes, we have . All same 5 channels . The issue is that with 8 bit per channels you couldn't keep enough precision for the depth to make a nice impasto . It would look ugly as hell . Same with UV channels . they are unusable in 8 bit

I would be happy if they would just copy Xara or defunct Creative House Expression approach for this where you could use vector spline to place vector or bitmap objects along in random or any other order. It's so a basic option in many vector editors.