@@bigmartin343 It is, but it's not explained in the video. You could, though, use the methods in this video to have the mesh follow vectors still computed from the camera, but modified to only rotate around Z. You can get the camera's X vector and normalize its X and Y components, removing Z, to have a normal that doesn't point up or down. Then 0,0,1 would be your Z vector. And Y, you can derive from X and Z by getting their cross product with Cross(X, Z). The order of the vectors in the cross product matters, so if Y looks flipped, try Cross(Z, X).

@@user-rd3jf1uh1c Fuzziness is the border blurriness. Expansion is to extend the mask towards its borders

@@RVillani If I have "sharp" color mask, (sharp mask need to get coord for textures), BUT may I "blur" border in this mask to make soft transition between textures?

@@user-rd3jf1uh1c if your masks are blurred in the texture, then you can use fuziness to blur the resulting masks from the shader too

@@WinterOkk thank you!

Yeah, just make sure both normals (mesh's and WorldAlignedNormals) are in world or tangent space when you merge them with BlendAngleCorrectedNormals. And do the Division stuff before merging in the WorldAlignedNormals, otherwise you'd be affecting those too.

Thanks for letting me know! I assume it's a plane with a healthbar material, then

@@RVillani It is a widget i spawn over the head of enemies in world space, not screen space, because i want the healthbars to partially disappear when they stand behind cover

​@@HakiPalatina Cool!! Glad I could help :D

Thanks! And... dammit Autodesk :/ It's as if they're happy as long as they have customers' money. Oh, wait..

Are you using Strata? I'm not familiar with it, but with it disabled, I can set shader models in the material in UE5. Also, even in UE4 you couldn't mix translucent and opaque models in the same material. You need separate sections in your mesh with separate materials for that, as I show at 1:01:35. I think you can only mix opaque shader models, like Default Lit and Subsurface. As for the blend, I also show it in the video, with material functions. Without material functions, you still have to create material attributes to blend them with BlendMaterialAttributes (1:02:01). To create material attributes, you can do like I did in the material function, at 1:02:15, with the MakeMaterialAttributes node or SetMaterialAttributes. Then you set the different shader models in each Material Attributes and blend them with BlendMaterialAttributes.

@@RVillani No no strata. Using the exact same setups demonstrated here just in unreal 5 the blend mode is not something you can change by the shader model. Looks like you could do it in 4 but not 5. Masked or translucency is not something you can change it seems even if wrapped in a material function and blended using the node you mention and show in the vid. We only get access to a preview blend mode and it makes you choose one or the other once the blend.

@@RVillani ah wait got it now lil bit different for unreal 5 you have to declare the final render parent materials blend mode to be translucent mode no matter what. Then of course set its lighting mode to surface forward shading to get the default pbr properties back.

Se nao me engano, se vc usa só o Y, ele tenta sempre apontar o Z pra cima. Se vc quiser controlar, precisa usar o Make Rot from YZ e dar um Z válido pra ele, que mude quando a rotação chegar nos 90 graus onde inverte. Outro esquema é usar o YZ quando não tá perto dos 90 graus, e trocar pra YX ou mudar o Z quando estiver perto dos 90 graus. Meu usuário do discord é o mesmo do meu canal aqui.

​@@RVillani Eu te adicionei la. Eu acho kkk Nao manjo de Discord Eu tentei usar YX e YZ, mas nao rolou. Eu ja adicionei um monte de print string pra tentar entender o que ta rolando mas num ta rolando :/

Thank you so much, Sir! ❤❤

then Pee again, so PP :P

lol

next tutorial: how to download more ram

Thanks, my friend! Let me know what you think

That's what the video is about, so I need you to be more specific so I don't just repeat what's in it. What questions do you still have left after watching it?

Sorry, I just can't understand the math you explained at 5:06 I really love the video thanks for sharing, So you took pixels of different colors i.e RGB values and you multiplied them with the vector in those directions.you get three more values and then you added the x values of all 3 coordinates I really can't get my head around this , like why we should multiply that and add that,, sorry I think my math is very poor :(

​@@sreedhar7266 haha no worries. The only thing you have to feel bad for is making me listen to my terrible English from back then (not that it has improved much lol). For each pixel, we want a final color that's the combination of all 3 planes. Each plane represents an axis: X Y and Z. But you need to know how much to use from each of those planes proportional to the intensity of each axis in a pixel's normal. If a normal is [1, 0, 0], you want 100% the X plane, and nothing from the other 2 planes. If a normal is [0.8, -0.3, 0.1], you want 67% of X, 25% of Y and 8% from Z (I'm rounding to make it easier). The total combination of all planes must always be 100%. If it's less, you're making a darker color, and if it's more, you're making it brighter. To figure out how much of each plane you need to have a combination that sums 100%, you calculate how much each axis of the pixel's normal represent from their total value. To do that, you divide each one by their absolute total. Absolute means you always get their values as positive. Starting from my [1, 0, 0] example. If you sum 1+0+0, total is 1. Then, if you divide each individual axis by the total, 1/1 is 1 (100%), so we'll use 100% of the color from the X plane for that pixel. The other two will be 0/1 = 0, so we'll use nothing of the Y and Z planes for that pixel. Now, for [0.8, -0.3, 0.1]. Sum is 0.8+0.3+0.1, total is 1.2. Remember we ignore the negative sign. Now, divide each axis by the total to find out how much they represent of it, up to 100%. X is 0.8/1.2 = 0.66666 (67% rounding up), so we get the color from plane X and multiply by 0.66666. Then, Y is 0.3/1.2 = 0.25 (again, ignore the sign), so get the color from Y plane and multiply by 0.25 (25%). Finally, Z is 0.1/1.2 = 0.083333 (8% rounding down), so we multiply the color from the Z plane by 0.0833333. Now, let's say all projection planes had white (1,1,1) at this pixel. If we multiply each of those white colors by the values from any one of these examples and add them together to make the final mixed value, they'd result in (1,1,1), because we made it so each of the axis of the normal represents a part of a total of 100%. For a more colorful example, that's probably better to visualize it in the mind, let's say plane X gave us magenta (1, 0, 1), plane Y gave us yellow (1, 1, 0) and plane Z gave us cyan (0, 1, 1) for a given pixel. And let's use our second example normal to compute the combination of the 3 planes, [0.8, -0.3, 0.1]. From the math in that example, we know we want 0.6666 of the X plane, so we multiply its color by that value -> (1, 0, 1) * 0.66666 = (0.6666, 0, 0.66666). Then, Y is (1, 1, 0) * 0.25 = (0.25, 0.25, 0). And Z is (0, 1, 1) * 0.083333 = (0, 0.083333, 0.083333). When we add those resulting colors together, we get the total of each channel = (0.916666, 0.333333, 0.749999), which is a pinkish color, because X was magenta and it was the most pronounced axis for the example pixel normal. Also, even adding those 3 colors together, that could easily add up to more than 1, we never got values greater than 1, because this math uses each color proportional to a slice of the 100% combination of the axis values. Does that make sense?

@@RVillaniOh my god!! Thank you so much man! Now its all clear. Neat explanation. You are amazing man!

Are you Lerping the transforms already? What's Vtubing?!?

@@RVillani I mean like VR rig. VR rigs are smooth and almost zero glitches b/w bones. With kinect sometimes the bones are twisting right. Does control rig a viable solution ? Most of VR/Mocap for UE on YT making Control Rigs. How can I overcome with this problem ? Any suggestions would be helpful. Thank you in advance. Anyway I am not Pro in Skeletal meshes or rigs.

@@OriginRow These are not simple issues, unfortunately. I advise you to get help from a technical artist. Lerping the transforms from frame to frame is one of the easiest ways to smooth the jittering out. Twisting of the hands is something Kinect always does, unfortunately. For that, I'd recommend getting the correct hand orientations from the VR controllers.

thanks! Could be many things. The objects in play mode could be having stencil or custom depth disabled, for some reason. Or your player camera is overriding post process settings and disabling the PP material. Or something attached to the camera is drawing custom depth covering all the custom depth in the scene. Or you were testing from inside a Post Process Volume that's not set to infinite bounds, but in play you're outside of it. Or in your material you set some parameter during play that disables the effect. Or even something else specific to your project that I'd never guess haha You'll have to debug. While playing in the editor viewport, you can press F8 to eject from your player into level-editing mode. Then you can select things in the viewport or in the World Outliner and inspect their details to see if the expected depth/stencil settings are set on the meshes. You can also select your pawn/character and inspect its camera to see if it's overriding any post process settings. While ejected, you can also change the viewport mode from Lit to other view modes and see if stencil and custom depth look as expected. To go back into possessing the character, press F8 again.

Vlw! XYZ são 3 eixos sempre perpendiculares um ao outro. Se um gira, afeta os outros 2. Quando vc usa um eixo só, a Unreal precisa escolher um segundo eixo pra completar a rotação, senão os outros 2 eixos rodariam aleatoriamente. E esse segundo eixo sempre vai apontar pro mesmo eixo no mundo. Por ex, se vc falar pra Unreal apontar o X pra qqer lugar, ela vai tentar manter o Z do objeto pra cima, q é o Z do mundo. E o Y vai ser perpendicular ao que resultar dos outros dois. Mas se vc apontar o X pra cima ou pra baixo, em 90 ou - 90 o X tá apontando pro Z do mundo, e o Z do objeto tá deitado. Então quando virar 90.0001, o Z ficaria pra baixo. Mas como ele é sempre direcionado pra cima, o objeto dá um 180. O esquema é semprr usar 2 eixos. Senão a Unreal sempre vai escolher o segundo por vc e tentar apontar ele pra um ponto fixo.

3:15 tem um exemplo com o X e Unreal fazendo o Z ficar pra cima

@@RVillani cara obrigado pelo retorno. Mas eu repliquei os seus exemplos e continuo nao entendendo o pq que meu teste nao funciona. Sera que vc nao podia me dar uma forca com isso no privado? Eu nao encontro nada em lugar nenhum sobre isso, todos os exemplos que eu vejo pessoas usando vetores sao iguais com excessao ao seu. lol

Eu consegui reproduzir o mesmo efeito! 🙌🙌🙌🙌🙌🙌🙌🙌🙌🙌 Mas ainda estou tentando entender o pq que funciona kkkkk Cara obrigado pelo video, eh esclarecedor ainda que eu nao tenha entendido 100% a hora que der o estalo eu vou dominar essa p* hahahaha Valeu

@@ddvictor Tem coisa que pra mim foi assim tbm. Tive que tentar, dormir, tentar, dormir... Até o Tico e Teco clicar e entender o q tá acontecendo haha

:/ that sucks

@@RVillani I found a solution on a forum. You can copy the modifier, delete it and then paste it back just before exporting :D. Clearly a bug >_>. It worked for me. I spent almost a day on this. Leaving it for future generations as knowing Autodesk it will never get fixed.

@@robertturaa9561 this video is 7yo and people still thank me for it. So, indeed, never 😕

But that is screen mapping. The screen is the camera viewpoint. What are you looking to achieve, exactly?

@@RVillani I'm looking into using the camera as a projector to display AI-generated images of a 3D environment based on scene z-depth. The idea is to project these images back into the scene using the same camera used for rendering. I came across a technique similar to this in a Blender tutorial here. thanks!

@@hassanb.eclipseHassan so you mean projecting the texture from a specific point and direction in the world (the camera that did the rendering) and see that projection from a separate viewpoint (i.e. the player camera view)? If so, I think UE has a texture projection node, but I don't remember the name. It's something weird like ZAlignTexture. I used it once to fake caustics projection at the bottom of a pool, 6 years ago. Wasn't intuitive. I'll see if I can find out more, but in the mean time check out material nodes with Texture, Project, Align or Z in their name.

Thank YOU for the kind words! :)

Hmm, the problem was fixed when I swapped from using the normal "Camera" component to using "CineCamera" component. The normal camera probably has some rendering/post process settings which caused my issues.

I finally got my nodes right, so what i did was: Custom depth - Scene Depth, then put that into ceil and then clamp bethween 0 to 1. Then I masked R and then did a "oneminus". This then got put into the multiply node that was put into the lerp alpha :)

Nice one! Sorry I didn't answer before. I think I missed the comment notification :( There were some people who asked something similar, I think. But KZfaq's comment ordering is weird, so it's a bit hard to find it. Thank you for the compliment on your first comment as well! 🫶🏻

Thank you for letting me know! 🫶 I really strive to teach the why, not just the how. That way you can get this new knowledge and apply it to a different problem.

@@RVillani you nailed it with this one. I shared it to a couple of colleagues. I really hope you can afford to do this more often

@@NJoint Thank you for the kind words! I hope so too. I was very frustrated with my latest videos. The audio was bad (don't know how I didn't notice that while editing) and I feel like I'm not explaining things in a clear manner. Because of that I've been holding off on making more content until I figure out why I became so terrible at explaining this stuff.

thanks for letting me know 🫶

@@RVillani I now have a realtime shadowcatcher in UE5.4 with Motion Design. No more Composure plugin😜

What do you mean with shadow catcher? Like something that receives 3D shadows but on top of a video/image background?

@@RVillani exactly that!

@@Mr_Dee that's cool! So I guess you used the screen mapping stuff

I used instanced static meshes with a material that rotated them to always face the camera in a specific way. But now I'd just make it in Niagara. It would be much easier.

I set this up back in UE4. I have never tried it on UE5, but with the new Lumen settings, I bet there's more options that need changing. Like Lumen's raytracing quality, reflections count and whatnot. I'm not sure all the required options are there now, but you could try playing with them in the Lumen's section of the post process settings. Honestly, I'm not even certain this can be achieved in real-time anymore with Lumen. Perhaps only on the path tracer now.

@@RVillani Oh, yeah, I even tried with UE 4.27.2 because of this reason, but I could not manage to do it. The corners of the cube reflect what is inside but stretches all the reflections. Not sure what it is, I think it could be the path tracing?

Maybe the offset values are miscalculated somehow? Could be many things. I advice you preview each node in your sequence of nodes to see if they display expected results. If you need to see specific numbers, you can use the debug nodes and preview them (Debug Scalar, Debug Float2 and so on). That way you debug where things are going wrong in the logic and find what needs fixing.

If you were asking how to do that, that's pretty much what I'd have suggested. A stencil value for "ignore this". I don't think there's another way if you can't tell those objects apart. Thanks for that comment!

I think so. But I need to reconfigure my setup for recording. I had to format my PC recently and haven't set that up again yet :/

That was back in UE4, before Lumen. I'm not sure that's possible with Lumen as I haven't tried. I hope it is! But it will definitely require all the heaviest hardware raytracing settings, if it can do that.

No i was using with Ray Tracing no Lumin, Well i will try to create this in UE4. Many Thanks 😊