kirk23
-
Posts
721 -
Joined
-
Last visited
Posts posted by kirk23
-
-
I find it too hard and slow to copy/paste into channels manually. Affinity Photo has a natural way to pack channels same as Photoshop by just setting into what channel a group or layer would input too. Do a macro that creates necessary layer structure and then it will always be just one click.
They added "hide" and 'hide other" for selected layers that together with select same tagged layers works kind of similar to layercomps in Psh. You can quickly switch in between color/normal/roughness of your document . The export is still manual although.
They added transform links like chain links in Photoshop. So you can link color/normal/roughness layers to stay always in sync together whatever different group they need to be in. But you have uncheck auto select groups and never select and move a group . If you do it ignores those transform links ( contrary to PS)
They fixed the displacement more or less although you still can't control direction of displacement. Nevertheless you can do a kind of erosion effect on height image by doing a series of 1 pixel displacements gradually fading one after another ( if you record a macro) . Same as the slope blur in Substance Designer.
Affinity photo still makes black holes to any open tga or any file with alpha where alpha is zero. No fix for that. You can avoid to make one while export by adding some 0.00001 alpha value beyond what 8 bit file can handle but the black hole is still there when you open the file back.
-
I need an option to draw 32 bit UV stripes in Designer to use them as a deforming rails for scattering small alternating pictures within same stroke after export/import to Substance Designer etc. Well, since it's not an option in Designer. Just let a brush be any bit depth we need .
Something like this with a brush being 32 bit exr please.
Would actually be super cool if we could use nozzle brushes in Designer too. The ones that could stay tweak able post fact and where we could replace the nozzle pictures , choosing another saved brush. Not rasterized instantly. It's a huge missing point of Designer. I loved how I could do it in Xara 20 years ago.
-
I mean things like compound or luminosity masks have been perfectly possible before . Mostly just a new interface.
So what's relly new ?
I found so far.
1. bitmap fill ( although it doesn't make the fill stay on place when you do boolean on vector shapes like it was in Xara , neither it allow to feather edges while keep the fill intact so not that much helpful
2. Transform links . A great feature indeed until you select a group accidentally and break everything apart again contrary to Psh. Still works fine if you auto-select layer, not groups. No pixel content linking from groups/a kind of live rasterization into a "virtual layer" . I hoped for that very much. Could make the soft rival any node based solution if would be working that way. Sadly isn't.
3. Normal map flip rotate and intensity live adjustment layer. Nice but i would prefer height to normal map one. Without it all those rotate and flipping is pretty useless. Lighting filter makes normal maps from height with wrong gamma in 16 bit and with edge artifacts.
4. Hide and hide others in layer panel. Cool indeed. Now we could select same named or tagged layer and hide others. A kind of poor man replacement for layercomps in Photoshop . I'd love they add auto-select same tagged in selection drop down too.
That said layercomps are so much better and so much more convenient. Still no way to export such "layercomps" automatically in export persona.
5. Live displace. they kind of fixed it. Now works same as in Phototshop if done over 16 bit grayscale depth image. Too bad you can't select a vector of displacement . I offered to use red/green as direction and blue as distance and they make some red/green choice but seems it doesn't work that way.
Would love they would do gradually fading series of 1px displaces . A kind of a version of "slope blur" from Adobe Substance Designer. We could probably do it with macro but it would be a huge messy stack . Would love it be just an option in displace filter
It's all I found so far. Anything extra? Hoped for a an improvements in procedural filter but nothing new I could found there.
Anyway thanks Seriff for improving the soft . Adobe does nothing at all for a decade already pursuing that useless AI features that helps you with nothing. Hope they would fill the rivalry and wake up too.
-
Would love to see 32 bit UV X /Y gradients rectangle picture as a curve driven image brush to be then used as a rail / curve deformer for any kind of pattern, texture or curve based scatter after export to a 2d compositor.
Using only 8 bit png as an image brush doesn't work at all , not enough precision for UV. What the purpose of having 32 bit vector editor with only 8bit brushes/gradients?
-
Not a text on a curve based but rather just a brush like nozzle brush in Photo but keeping all objects editable and convertible . maybe as symbols
-
My main tool for my job during last 10 years is Adobe Substance 3D Designer . But I can't express enough how unhappy I am with it . I hate it deeply. It's a huge time sink for nothing. Extremely artist unfriendly. Everything seems done most inconvenient way ever possible.
Absolutely everything, even UI .
And I know it well deep enough. Had to redo every node to my owns through years . Had to construct a tone of my own pixel and value processors + custom functions and yet it's a huge pail of crap where every subtle artistic tweak takes you hours for figuring out this notorious "macaroni monster", "exposing" parameters and tweaking functions math. Finding lost dependencies is another crazy challenge there.
So Serif please do your version . A one that could work in cooperation with Affinity photo or rather Designer . An advanced procedural layer. I'll buy it happily whatever it would costs if it would be just slightly more convenient or just would update quicker like your embedded docs vs smart objects in Photoshop. Not sure why I still even pay Photoshop subscription. New release few days ago and nothing helpful at all. Only that useless AI shite.
- DelN and Frozen Death Knight
-
2
-
Imo we need not each layer exported but rather a compositions of layers . Like layercomps in Photoshop. It could be working just fine if we could do copies of slices and make them remember each own set of layers , effects , blending modes including ones inside embedded or linked files.
Exporting form Affinity is so much a pain in your ass really.
-
4 hours ago, drstreit said:
Disclaimer: I am complete amateur with creating seamless patterns, so a question to your expertise: Would you mind to have a look at https://barium.ai/dashboard , maybe register and try out some patterns and tell me if that is something you see as being useful?
It uses the Stable Diffusion model (the only one to my knowledge that you can easily deploy locally), so its quite accessible and extendable.
Well, that's new for me. I have tried it. Nope, still useless toy but at least it tries to do height channel . The problem it's blurry as hell , as well as diffuse textures too while in videogames the textures should be like a pixelart , not a single pixel wasted for nothing . Results are repeating as hell either and looking super surreal .
In a word even Adobe Sampler AI is way better , at least it can do pretty well a limited number of subjects . A sand with with small pebbles for example. For some uncertain reason it's only matter AI texture generators could do well.
My guess if AI do well in the field we would hear a huge buzz about it from every game developer and CG artist. But all the buzz is about Substance Designer mostly and how cool it is with its procedural pipeline. Wish we would have a taste of it in Aphinity. Just a few touches to its procedural live filter maybe so we could really use it that way. Not just a name. IMo it should go ahead of all those AI toys
I recall there was a buzz few years ago about some game where they used AI assisted animation where AI did 80% of the game animation. It looked so crappy players complained non stop. Now Nvidia tries to do something like that . Perhaps we will see something good from this field.
-
On 9/25/2022 at 12:41 PM, DelN said:
Hi kirk23,
I haven't used 3D software for a few tears. I tried Max in its early years, Cinema 4D for a few years but that was 5 years ago now, Bryce (but that was many. many years ago, and it was very amateurish (I couldn't afford the £2,000 for Max then). it was a very surreal interface and the results were not very good. I loved Vue and Plant Factory (for creating plants), but they were bought out by another company (who absolutely decimated the number of users they had by not updating the software, not giving information, just hiding away behind their website for a few years. When they did announce an update, it was to be on a RENT-only (they call it, 'subscription' like Adobe. But it means 'Rent'; if you stop paying your subscription - you-re out, mate!)
You certainly know your stuff!😊
I'd be interested to know what program you use to create your seamless textures (or do you find them or purchase them online?). Personally, I have always used Painter - from version 5.0, years ago - but Affinity Photo has the amazing Tile layers, which I find easy to use, also I use 'Affine' to create seamless textures and amend them, ie. painting out the join lines (vertical and horizontal). But I guess you have checked out these methods and know more about it than me, working in the gaming industry.
But I just thought I'd mention it.
The 'Dream Textures' plug-in for Krita and AI-text to image software in general is perhaps not the right tool to be creating the quality of seamless textures you require at the moment. The technology is still in its infancy. But it's coming along leaps and bounds and many people are working on the software in the background and they're getting funding from big companies who see the potential of this new and exciting technology.
I'd love to see some of your work. Send a link. I'm sure it would help and encourage other young Affinity Photo/Designer users who want to get into the industry.
And I agree with you entirely... coders - not all of 'em - create what THEY envision the end product to be (despite being given a detailed brief and having endless meetings), and the prototype they have spent many months developing is often far from user friendly and is often lacking in design and functionality. 'Ah well,' as my mum used to say, 'it's all for the sake'...
To share any work I have to get my company permission first. Don't have any portfolio for 20 years already, just no time to do something outside of the job . But it's mostly scene environment stuff anyway. Nothing spectacular.
As of seamless textures I use Substance Designer and Painter mostly. They are industry standard now . Since Photoshop released pattern preview I use it to do photogrammetry based textures tilable sometimes, thanks to Photoshop patch and content aware move tools + record-able actions. It is simpler and you have more control than with any of AI tools I tried so far.
Perhaps they are in its infancy state , yes but I am waiting for 10 years already and nothing really helpful yet. just toys.
Affinity Tile layer is nice but kind of useless because of a number of Affinity limitations. it's destructive and you absolutely need non-destructive tools to do tilable textures now.
There is no convenient ones unfortunately . Adobe Sampler was a try to do one my guess but it's a pathetic fail imo.
Also a half of my tilable textures is a bake from 3d software. I use ClarisseFX . It has lots of easy tools to scatter small things around and nice baker . Much quicker and better quality than that non-stop pain in your ass Substance Designer is .
-
On 9/21/2022 at 6:53 PM, DelN said:
'Stable Diffusion' AI tool now available in Blender.
Dream Textures...
I haven't tried this but the video itself don't give me any incentive to even bother trying. It perfectly illustrates my disappointment in AI tools for textures . Totally useless . Even so simple subject as brick texture is visually repeating like hell. Bricks are deformed etc.
And all this before the fact that color RGB textures is a last thing you care about when create any material for games or movies. It's depth/height texture that is a base to everything else. A thing you derive color , normal, roughness and everything else . A core thing of any material. And AI so far are pretty bad with generating height . Just try Adobe Sampler AI.
I am really looking for de-lighting AI but so far haven't seen anything good. Removing lighting and shadows from photogrammetry source images is what could be a perfect AI purpose but I have yet to see anything useful. Well, actually not from source images , they would instantly turn to be useless for photogrammetry after that , rather from final texture. Any modern albedo/color texture should be deprived of any shadows or shading information at all. It should look very flat and incontrast , just a hue variations and that's all. And all AI tools I have seen so far just do ancient old hipass on them , nothing good.
Perhaps it's not even an AI issue, rather people who code it. I have to talk with graphics programmers a lot at my job and usually it's like a talk in between blind and deaf. Programmers who spent their whole life codding visual stuff often don't see an absolutely obvious things . That is always make me kind of stuck since I often even can't find right words to explain something because its so damn obvious and self revealing .
Same reason probably why all the non-AI tools for creating CG content are looking like been created by mad scientist . You spend your whole life trying to figure out their crazy logic and deal with their monstrosity .
I once heard a story that still amazes me. An artist turned businessmen in US invented lead paint tubes in 19 century . A metal known since antiquity. Before artists used pig bladders to keep their paints from drying . Doing small holes in such bladder evry next day. And they did it for DAMN five if nor more centuries . I very much hope AI wouldn't take five centuries to be useful
-
So far I used a few AI tools :
Simplygon - nope, it's still easier to do lods manually. Simplygon, while now a part of Unreal, still couldn't work efficiently or does it for a few typical subjects only.
AI selection in Photoshop - nope , works ok only for kittens and faces, nothing I ever needed. Couldn't recognize a thing beyond those few subjects . Other AI tools in Photoshop is totally useless for me .
Ai keywording services for DAM soft. - do a huge mess of keywords . Same issue with recognizing subject as with AI selection in Photoshop. Works fine for people faces and that's all. To bad I never needed that.
I have been subscribed to what now Unity art engine for a year.
Probably the best AI tool for CG and textures for the moment . Hardly used it at all during the year because other then for a few very specific tasks it was pretty inconvenient. Slow , not good enough preview, always doing anything but what you actually wanted. Lacking any perception how repetitive the final material is looking. I preferred to just use patch tool in Photoshop instead of all that cool AI material regeneration.
As much as I hate Substance Designer it still miles ahead and much more useful without any AI even considering its pixel processor and Fx nodes are surely codded by mad scientists.
AI material from photo generator in Adobe Sampler. Well , for high quality materials not a chance. Works fine only on very limited subjects and the soft is inconvenient as hell either.
Videogames materials have lots of limitation you have to constantly find workarounds for. Not a chance AI could do it for you so far.
AI unwrap in Substance painter, Never worked right. Works like a rookie doing its first UV unwrap in Blender . Totally useless.
In a word after all that fuss I am so disappointed with AI . My impression it's good only for deep fakes and that's all.
-
I se
43 minutes ago, drstreit said:And last sentence: As I like to visit the Basel ART each year, I habour some doubts if your expectation of creativeness, exploring totally new fields of imagination, boldy go where noone has gone before that you demand from our field is really the barrier that every artist has to face these days...
I stopped visiting things like that couple decades ago. IMO it's just a visual information noise . I agree it's a kind of fan to create . I instantly recall a nice smell of paint from my old art school, a cool feeling of easy and limitless creative freedom . But now I think about it just that : an information noise.
And where AI soon would be dominating probably
Now try to do a good looking tree models for a mobile game. From speed tree software for example, a best available. Looks like crap , right . textures looks like crap too.
it's where you to need your brain, need to recollect all that boring descriptive geometry courses , vector 3d math and that short 3d graphics theory course in your art school you thought you would never need in your supposed glorious art career etc.
IMO it's where true creative and inventive power necessary and where AI wouldn't help you at all. So far at least.
Years after my most regret is that I myself have paid not enough attention to all those short ancient geometry courses in art school they considered one of "free arts" . My thoughts was it's not I am going to use ancient Romans mathematics and geometry to build Pantheon dome or something, neither architecture was my specialty. Surprisingly ,years after It turned out it's a knowledge I had a good use of ahead of anything else.
-
Few years ago I was super exited about AI . Hoped it could help to get rid of lots of annoying routine. Would make human interaction with a computer much more simple and let us focus on creative tasks.
Nothing of this have happened since. I admit AI perhaps succeeded in a few very narrow tasks like coloring b&w images and de-noising but mostly AI is a few useless toys , not helping a bit. In many cases it rather annoying like AI selection always selecting something you don't need really . Same with AI usage in other fields of computer generated graphic: AI LODs , AI unwraps , AI 3d modelling etc where it's often stupid like a rock and also rather annoy you than helps. Pretty much like AI support in your bank.
My impression it will take hundreds of years till we get any helpful AI really and it dangers only those kinds of jobs where you couldn't stress your own brain at all. You indeed should get rid of such job asap . Including some brainless art generation tasks. Like those where you could call anything an art and that in fact was just a happy and fan random canvas smearing or random photo or fonts compositing .
When it comes to something where a good taste, a good sense of nuances , something of a true inventive nature like 3d animation, video games and movie graphics, fashion and costume design, something with clear understandable criteria where you have to find a way to overcome limitations , any field where you should solve unsolvable and hopeless tasks and nobody but you could invent how to do it , in such fields AI is still totally useless and been a huge disappointment during last decade.
Any content creation software is still tremendously inconvenient and a huge pain in your ass to use , and at the same time not that flexible at all. It all looks like mad scientist nightmare , totally artist unfriendly , illogical and over-complicated. The adjective "convenient" next to "software" still sounds like dry water , an oxymoron basically . I wish developers would focus on this issue first.
- Stun Damage and Old Bruce
-
2
-
52 minutes ago, Alfred said:
I know that the term “albedo” relates to reflected light, but I don’t understand what you’re asking for here.
Was that a typo? Did you mean to write “from [the] illumination factor”?
yes, I meant something like this http://www.cgchannel.com/2019/06/tandent-vision-science-launches-lightbrush-cloud/
I had this soft on mac . It never worked that good as on the picture and now is dead I believe. Something AI could do probably
-
I'd like an AI that could separate albedo color form illumination factor , remove unwanted shadows, increase resolution , recreate Z depth . But still haven't seen any of such things in any other programs that could do it good enough to be of a real help.
Even AI selection that could be really helpful works very so so in Photoshop for example and doesn't help a bit really. Rather annoy you like AI technical support in a bank.
The whole AI neural networks thing is a total disappointment through recent years. Other than deep fakes it looks like incapable for anything useful.
So I wouldn't waste time and focus on gazillion very basic missing things they could improve without AI involved. The soft still has so huge gaps in features I still can't use it for my job and have to use Photoshop to fix files from Affinity photo.
-
19 minutes ago, Pšenda said:
But then a DAM (file manager) would be required, not a file viewer.
Well. if windows explorer could write keywords not only for jpg and Finder could do hierarchical tags it could be well enough probably without any special DAM . But my guess it's nothing Affinity team could do about.
-
My guess OP mean something with basic key-wording , rating, metadata and a search based upon , something at Adobe Bridge level at least. Not those horrible file managers both Mac and Windows couldn't make useful for decades for some uncertain reason. But I definitely prefer it would be on OS side rather than Affinity one.
-
Nvidia has a free standalon soft available to convert files to DDS . Perfectly same as Photoshop plugin. But of course it's an extra pain in your a... by saving a second time each time.
And we need first to have that zero alpha black holes multiplied on rgb colors fixed IMO before even think about using AP for textures
-
Basically a layer rasterizing content of a certain group. To use the rasterized result elsewhere. This would allow to use linked and embedded layers as sources . We could also be able to have complex dynamically calculated masks instead of copies of huge stacks of layers.
It would simplifies the main layer stack a lot. Would make it same flexible as any node based editor . To speed up calculations it could be working on demand only as an option , with "update linked " button or something , if too many layers would make it too slow .
I'd love also a certain color code for chain image in layers panel/pane. The whole pixel content is red, opasity only is blue etc.
Also please , add transforms linking as an option for the linking . It's so much missing there. It's actually the first thing that should have been getting this whole linking idea.
-
I'd love we have it working same way Pattern view works in Photoshop. Instead of destructive way currently only for a single layer
-
Layercomps is a great tool in Photoshop. Together with scripts it allows to keep track and save multiple versions of same document automatically. by just single button . While in Affinity it's all same annoying routine.
if layercoms are patented or something just make overlapping slices at export persona keeping each its own set of layers .
Or just make " select same tagged /named" inverted command so we could select all layers except certain tag /name and switch them off. Ideally in all linked, embedded docs too in sync .
Would allow to get rid of so many tedious save /export routine . That export persona doesn't help me a bit currently
-
We already have live displace filter and it could be a perfect replacement to any manual mesh warp doing basically exactly same thing : warping the image .
But weirdly it works all the way wrong vs same thing in evry other soft. Doing kind of opposite shifting on the left and right side. Not sure why.
All they need to do is fixing that displace filter and make it sample form 3 channels of "layers behind" Red and green channel to get shift direction /vector and blue channel for a distance of shift. Sure it could be an hour of code tweak since otherwise it's all already working.
-
The ability to open layered exr files is cool but missing one important thing. We need cryptomatte tool to produce masks , preferably interactively working . Better as procedural filter with sampling capability that would produce masks by clicking on something only when necessary. Would make it another competitive advantage over Photoshop.
-
The way I could displace/distort pixels in direction set by RG of some vector /normal map and distance from Blue value . In a word a sampling mechanism from another layer /image or rather groups of layers.
That only thing would turn it into really procedural image editor . Maybe not as advanced as Substance Designer but I would instantly switch to Affinity anyway. I so much tired of that nodes only things and all the quirks of those pro tools where I really need just simplicity and convenience.
Thank you for ... feature
in Feedback for the Affinity V2 Suite of Products
Posted
Why everyone is so excited about that shape builder while it's always been possible before with boolean operations? Imo it's just a new interface for the perfectly same thing. Am I missing something?