kirk23
-
Posts
721 -
Joined
-
Last visited
Posts posted by kirk23
-
-
Like Corel Painter has . Only please do it accessible and exportable in 16 bit at least . And having depth info for nozzles too.
Plus a special separate blending modes for the depth: add, sub and depth combine/height blend when depth channel mixes using lighter (MAx) blending and at the same time provides a mask for other channels. ( the math is simple like 2x2 really)
Plus an auto masking feature for brushes when they would paint in cavities or bumps with a curve based tweaks.
It all could almost work that way right now . Well, except nozzles. Just requires a painful setup and insane stack of layers. So do it a proper and elegant way like being just an extra channel please.
ideally 8 bits in RGBA and 32 bit in the depth at the same time. You are already supporting exr files for nozzles . just let the soft read the depth from them too.
With your nice live lighting filter it could work super natural altogether. Well, maybe an extra feature of casting shadows from the bumps could be nice too.
Could be a huge advantage over Photoshop and even Painter where the depth is inaccessible and mostly useless . Not sure why they can't do anything right with it for decades.
ps. I am not asking for another Zbrush here , just simple things really. would be super helpful not only for impasto painting but for texture works too , focus stacking etc.
-
That workaround only works nicely in 32 bit linear mode. You can get more or less ok result with gamma fix in 16 bit integer mode although. It's still not 100% mathematically accurate and honestly I don't see why they can't do it .
-
17 hours ago, DarkClown said:
I with for a non-destructive(!) Frequency Seperation filter.
Any frequency separation technique is just a fancy name for blur / kernel math down to what software actually does really . You already have plenty of ways to recreate it live starting from Hipass live filter which itself is also re-creatable by blur, blending and opacity sliders. People do it since ancient time of first iteration of pixel processing software.
Whatever some tutorials suggest it's all the same basically just with different level of convenience.
In one word try Hipass live filter applying it separately for each of RGB and blend option curves
-
Like accessing/sampling beneath image to modify like some other live filters do? Like displace one. Can I do my own displace with procedural filter? I mean sampling not the pixels the filter is applied to but what's beneath that layer ?
-
On 11/14/2022 at 3:28 PM, Kal said:
V2 may have brought some cool new features, but it has only brought modest improvements to a few of the features which matter to me the most, while other issues have been overlooked completely. Having waited so many years for the first major update, I have to say, I'm pretty disappointed.
v2 is a nice step forward indeed and in the right direction. The problem it's not big enough step. With that pace we would have to wait our whole lifetime before we get truly modern non-destructive image editor. Non-destructiveness is what I need for the first hand and we still can't link pixel content without rasterizing bitmap first killing its link to its original source.
Procedural filter is still in a baby stage , no sampling other layers except beneath one , nothing truly helpful beyond noise texture.
I just hope Affinity wouldn't stop to develop the program itself and do only AI filters like Photoshop seems to per-occupied now. I am so much more disappointed by recent Photoshop release.
For simple like 2x2 image compositing we still have to buy Nuke and deal with this typical node puzzles even if we don't need any animation at all .
-
Interface changings are usually just annoying pain in your a.. Never helps a bit . I couldn't recollect a single case for any soft at all. Contrary it always makes you new puzzles . Never understood why software devs insisting on this. Software interfaces of 20 years ago was already perfectly fine with me.
It's not like in a typical market when you buy same crap again because of new shiny box and cool buttons.
-
I recall I did it with a text along a path approach. But honestly it was an absolute torture to work that way.
Have anyone found a better workaround?
-
I tested that vectorQ on iphone. Haven't noticed anything special. Perfectly same as inkscape. Can't recognize fonts or just simple shapes like rectangle or oval . Lots of redundant path nodes , no optimization. Is it better on real thing like mac?
ps. would love a kind of AI vectorizer , a smart one, but doubt we would see it any time soon.
-
Affinity photo exports those files with alpha channel just right . V2 at least. In v1 it was also perfectly possible with a simple trick.
The problem is not what Affinity export. The problem is that when you open the file back to Affinity photo it multiplies black over rgb values of pixels having perfectly zero alpha, Only over those ones. End this needs to be fixed for sure.
ps. V2 doesn't do it for exr files although. floating point images being opened perfectly right too, without those black holes
-
1 hour ago, deeds said:
That's Procreate's 3D model painting features.
How is it vs substance painter or 3d coat? Looks very basic from youtube videos.
-
22 hours ago, ThirtyFiveThousand said:
If you want to put a logo on a 3D object, then why not just put it on a 3D object?
I meant not 3d object but rather 2d image, computer generated . Basically rendered picture. Why I need to modify 3d scene and render again when it's simple and easy to just distort a logo to match something perfectly using depth or UV AOV if they have been included in your render. No guess as with manual mesh wrap. Try to use that mesh wrap to make a logo following something complicated , a waving flag for example. Still the mesh wrap is a nice addition too indeed.
-
11 minutes ago, pixelstuff said:
This is really where artificial intelligence can perhaps make a big difference
Yeah but currently it can't to do it even with the sky reliably based on Photoshop. Along a horizon line where colors are somehow earthly already , a dusty haze of some sort it doesn't work that good. People , cats are totally ok. And that's all. Anything I ever needed for my job . no chance.
Resent new Photoshop version with better cloud AI option didn't make a slightest difference. I wish they would rather focus on their super outdated layer system, 32 bit support and slow like hell smart objects.
-
-
Mesh warp in nice indeed but I would prefer they would do direction control (both manually and by texture input) in live displace. Displace is so much proper way to deform 2d Image. Even better idea probably is to use UV coordinate AOV image to place bitmap on a necessary shape. Another live filter probably. Like everybody do it in 2d compositing . You want to put a logo on 3d rendered car in a movie post render , nobody would ever want to do it with mesh deform especially when you need it in 2k frames.
-
Does anyone know if there is an AI selecting tool that can select say ,mossy areas on rocks image , or rusty spots on something metal still having little color difference, or peeled paint layer on same colored another layer. Easy and quickly . That would be huge AI help.
Or something that can remove lighting , shadows and keep just diffuse color?
-
When you record a brush stroke in a macro it allows to chose diffrent color. Could you allow as to chose different brush please so it would be executed with a new brush , but keeping all the nozzles at exactly same position and scale . Or be able to stay as separate objects if brought to Designer . So the brush could scatter small bitmap objects.
-
Now as we have an option to make bitmap fill for a vector shape. Is there a trick to feather edges and save it as a style ? I recall someone posted how to do it with certain combination contour effect and blending options together with layer blend options but i totally forgot how . Any ideas?
-
1 hour ago, ThatMikeGuy said:
Care to mock that up?
I just offer the way how cryptomatte usually works evrywhere . In Blender compositing pane for example . You have a split view and in one half you see cryptomatter info layer with all objects filled flat with random color id . you select something there and get b$w mask in right pane
-
-
Please don't make it like exr_IO . let it be just a cool split screen interface where you could select an object and just have a mask . Not gazillion of masks at once
-
Couldn't care less about dots vs checkboxes
-
1. Can we have nozzles in Designer please. Spline bound . Please, please,please 🙏 With lighter blending in between each of them rather than different strokes. With a blending choice set for each of RGBA channels separately. So it could be add for the green , MAx(lighter) for Alpha . multiply for red etc. All saved in a brush version.
Also please make image brush in Designer to have several crop lines so we could set several special blocks that would alternate along the spline randomly. Come-on , we have similar feature in former Seriff product. Why it felt under the feature cut at all?
I recall Expression had it back in early 90th.
2. LIve displace needs direction control, badly. I see some red/green vs sobel choice but see no difference at all . Am I missing something? A sound way would be using red/green as direction vector and blue as shifting distance multiplier IMO. Or gray slope as direction.
3. We almost have layer comps now with HIDE and HIDE OTHERS . Now could we have some auto -exporting mechanism for this. Slices that can remember layer visibility maybe?
4. Bitmap fill as drag and drop could be super helpful if the fill would stay in same place when you reshape the shape and do boolean operation using shape builder. Make it by first selected for example so the zero UV coordinate would always stay in top left corner by default of the first selected shape . Look at Xara for example .
5. inner glow FX having erase blending mode working to current object only . Aka edge feather slider.
6. live rasterization adjustment layer we could use as a source for pixel content linking. into masks for example or as "virtual" layers. Usable as a source for live pattern layer maybe.
7. Cryptomatte live mask for 32 bit mode. Not like exr-IO plugin with a mess of gazillion layers but rather just live object picking from a colored preview with your typical split view.
8. UV red/green gradient into picture placement live layer . To reapply other materials on CG rendered picture. maybe just a displacement version.
I believe those thing would make Affinity a real competitive advantage .
-
1.transform links aka chain links in Psh
2.hide vs hide others working together with "select same tagged" , smells like layer comps of a sort.
3. Displace live fixed , not quiet yet although . Still no control over displace direction . Still can be used as a slope blur from Substance designer (with macro recorded)
-
On 11/12/2022 at 12:57 AM, drstreit said:
the quality is doubling every 4 weeks
I probably use other AI but see no quality changes for years already. They are all same ridiculous as AI assistants in a bank. Still wait for something AI based would appear and make life of videogames industry artist easier . All I ever seen in UV unwrap, 3d modelling, texture creation are so stupid and so bad I stooped to even care . No hope at all. Even Photoshop AI selection that could be used theoretically to select some photography features to make proper roughness texture from same photo doesn't work AT ALL with Ai selection. No use whatsoever.
Normal Map Generator
in Feedback for the Affinity V2 Suite of Products
Posted
No, my version is no different in that regard . You can load separate depth too but can work with just what's beneath + "texture" slider that adjusts normal map intensity/contrast. The only improvement I offer is the gamma fix for 8 and 16 bit per channel images . in 32 bit linear mode you don't need the gamma fix.