Видео

Hey, really love they effect! I wanted to know if it's possible to apply this only on shadow? I'm totally new on this and learning. Thanks.

Absolute chad

So clean! We're gonna use it for sure!

this is extraordinary impressive, but im too dumb/noob for know if i can use it… so just gonna ask in case some bigbrain kind soul has an answer in the future: could this be used to set up a master material for the while game, and create instances to drop the textures maps i made in substance for my characters/props/environment? :)

The visual is really insane! I tried it and it works great! The video with the explanations is just as amazing too haha! I've subscribed, can't wait to see the next videos, whatever the subject!

perfect explanation for newbies!

Thank you very much for taking the time to dive into a topic from a professional point of view and giving people a solid foundation to work from. The minute I realized, oh, this is how we get to that feeling of "shooting in RAW" it makes you realize how many people (myself included) are starting from the defaults giving to them and just piling on band aid after band aid to try and get the look they want. Please keep the in depth information coming. 👍

I am trying to implement height blending based on smoothStep expression but it doesn't look right. till division it's same but when I add 0.5 and sature after the smooth step result. it gives me greyish result. if use constantBaiseScale it offsets the result when I changes the falloff (smoothness) value

ruclips.net/video/eBS3BOI5KnM/видео.html - It bleaches the whole scene when I put 120000.0 lux as the directional light, after this the only time I can 'see' anything is either unlit, or through the PPvolume being enabled infinite extent, is this intended?

I'll try to add it to my path traced anisotropic kuwahara experiment

I have just got to the material function part and was wandering what the orange, green, and purple nodes are. They are named a, b, c, d, e. I cannot find them anywhere?

Others: "Wow. This really helped me solve those issues I had with using decals." Me: "I'm going to avoid using decals on non-flat surfaces like the plague."

Thanks for an incredible tutorial. I've been using an ACEScg 1.2 ocio config setup in Unreal for my cinematic projects and was wondering if the postProcessVolume setup in your tutorial would override any of the visual settings? I've got a log display transform option to check the luminosity and ensuring the image can be as neutral as possible, but not 100% sure if your postProcessVolume is aimed to achieve the same goal without using a standardised working colorspace?

How do you create function nodes at 5:41? are those "in engine" nodes or a collapsed group? Great tutorial btw, very helpful.

You keep this shit up and Epic is going to give you a job offer man. Another awesome video as always! 😎

Amazing job like always. Do you know in this cases why I have in the version 5.x has artefacts in the shadows event the shadows not is dynamic?. Do you think that the problem is something of lumen and nanite world?. More problems with the normal representation when I create dynamic movements using mathematics. Example: Waves in a plane moving the vertex position with a shader. Or using vertex animation texture the shadows not is working well and the normal of this assets have bugs. However when I use this methods In the version 4.x I have not problems.

Actually u can enable CVarDBufferDecalNormalReprojectionEnabled to get the scene pixel normal from last frame in deferred decal materal

This video is tier one. Thank you for going though the drudgery of figuring these technical puzzles so we don't have to .

Hi! Your tutorial is awsome but im struggling to understand. Are the red nodes that you are yousing at 5:42 Named routs also?

Really good video. Additionally you may try to use the Light Vector Z axis to get Decal depth. It was never documented, but it works.

Amazing topic. Great Editing :D

Would be great to mention UE in either the title or the thumbnail.

Epiccc! Never knew this was happening, incredible fix!

the editing/vibe of this video is top notch, makes decal normals feel lively and interesting

Quaternions have entered the chat

Perfect video, I don't know anything about decals yet and I'm pretty sure that after the video, I will know more than 99% of game devs

used to pray for times like this 🙏

Always a good tutorial

Basically. Avoid using normals for decals for lowpoly, non flat mesh.

Love these explorations, top notch work. GG👍

Perfect as always!

already watched it three times and i dont understand a word 😢

i loved

I would love to learn more about forward rendering in Unreal engine which seems to be notoriously under documented. Perhaps for an effect such as "render layers", which is commonly used in FPS games to render the first person arms/weapon "on top" of everything else to prevent it from clipping. In deferred you must scale vertices towards the camera or simply scale the mesh down a lot which messes with lighting. I also tried experimenting with modifying the rendering code in a source build of the engine by adding a half lambert diffuse to direct lighting in a similar way to TF2 and Valorant, but I could never get it to fit well with the indirect lighting. Having someone as knowledgeable as yourself explore these topics would be amazing!

Great talk, guys! This will surely blow up over time, as more and more people find it. As a former color grading artist turned tech artist, this was really fun to follow along with and quite accurate.

Hi, I tried to reproduce your shader and I couldn't make it to work. It gave me a full black result so I guess I mess somewhere. The LaplancianFilter_Normal give me a good result, but I was wondering if it is normal that my LaplacianFilter_Depth always return me 1. When I preview at the Mask(A) I get a full white result, so at the end the max node is also gonna give a full white result and the 1-x is turning my result into a full black image. The material is set to post processing, the blending location to Scene Color After DOF ( beceause I'm doing this with unreal 5.4). Any idea ? I'm a beginner in shader graph, not in coding. I wanted to make your shader work in order to test with other matrix.

Great tut! Proper workflow starts at 8:00

I have a question: What if you use Forward Rendering? Wouldn't that allow us to make a true Cel-Shading using Custom Nodes on the objects themselves?

Exactly the video I was looking for. Will be trying this out this week.

First of love the tutorial. Second thing is that I'm currently working on a project where I need a cel shader where I can freely change the shadow AND highlight colour. These objects will have image textures on them as well. From a quick google search, changing the shadow colour seems to be doable, but I dont get much information on how you would change the colour of the highlight.

I dont know why but its giving me an error about the float2, I did the same code you did in the video so I dont understand why

The results look amazing, but id like to see it in a scene with more building props/details as well as on a character that's more complete. Seeing the effect on solid color character doesn't give me enough data information to understand what it would look like on a finished character. if its possible could you do a video showcasing what the shaders look like in different environment styles or props.. and on different Characters of different style. Id like to see how the shader looks and behaves.

Hey I am an artist coming back after years of unintentional pause ( depression + burnout ) and currently gathering info and inspiration for my own game. My question is tho, I am currently not sure if I am allowed to use stuff that I created from following tutorials like yours. Like a one to one reacreation... sounds iffy but I only know art and do fail absolutely with technical stuff. In most cases I follow tutorials for this.

Very creative approach. And innovative. I think this can bring benefits in many cases.

How to make a polynomial distortion?

why it`s dont work in my procject I don't understand

Hey there, thanks for the great tutorial! I have a question as a total beginner: how would I make the plane face in the direction of my camera normal instead of the camera itself? I am using an isometric view for renderings and I would like all the planes to face the same direction, right now my meshes turn towards the camera which tilts them a bit the more as they get on the edge of my view. Any ideas?

this is the most in-depth technical look into the material editor in unreal engine that i found!

I get lost at 5:52, the node functionInput does not appear to me

I watched this video for the edge detection (after I watched the other two), and this is all mind-bogglingly useful. I have some issues with scaling edges over depth -- you are using linear normalization for depth, but this doesn't distribute the line thickness right (it should be inverse distance for the line thicknes -- two times farther, two times thinner), because kernel sizes are not halving in size, so the lines in mid distance will appear visually thicker compared to objects than foreground lines. But this is not that hard to fix. The main problem I have is what every edge detection I've looked at that handles edge thickness over depth has -- the final edge thickness is contextual, and depends not only on the edge distance from the camera, but also on the distance of the background objects, so that if an object covers two objects at a different distance, its edges are going to be thicker where it covers the closer object. As far as I can tell, this happens because the actual edge is immaterial, it is something that's actually 'in-between' the pixels, and we perform the edge detect on the the surrounding pixels. So what you get is this -- you take the pixels of the near object that are in front of the far object and detect an edge there, then use a big kernel to draw it. Then you move over to the pixels of the far object that are just behind the near object, and also detect an edge there -- but this is a false edge. However, it is drawn using a smaller kernel size, and contributes to the ovarall edge thickness. An occluded object that's even farther away will be drawn with a smaller kernel size, so it's contribution is going to be even less. So the result is the uneven edge of the foreground object whose thicknes depends -- as I already said -- not only on the distance of that object, but also the objects that are behind it. At least that's what I think is happening. Any way to get around that, and only draw edges according to the pixels that are closer?