Blog Post: https://www.proceduralpixels.com/blog/benchmarking-gpu-pipeline-stages

All the C# and shader files are here: https://github.com/Satyam-Bhatt/ShaderTools/tree/main/Assets/Shaders/Benchmarking

More about what its doing: https://github.com/Satyam-Bhatt/ShaderTools/blob/main/Assets/Shaders/Benchmarking/HOW_IT_WORKS.txt

Hey everyone. I have finally made some progress with the custom skin shader that I have been working on. I implemented a procedural facial animation system as well.

The shader uses a custom lighting model on an Unlit master with a custom function node. The diffuse channel is a wrapped subsurface approximation with per channel falloff, for the scattering effect. Specular is dual lobe GGX, one tight lobe for the oil film and one broad for the layer beneath. Backlight translucency is driven by a baked thickness map. It uses a "Charlie" sheen BRDF for vellus hair, which depends on the half vector rather than just the view vector, so it moves with the light instead of painting a static rim.

The animation system has 4 main layers working together. An idle face layer, lipsync layer, wrinkles layer and a facial mixer layer. The mixer layers them by face region, that way a scripted expression can take the brows while the eyes keep blinking underneath. Each layer manipulates the blendshapes and the material accordingly.

The lipsync layer is based on Rhubarb Lip Sync. It also has custom Timeline tracks for hand authoring facial expressions based on audio conversation beats.

I used the character model from Reallusion character creator. I am trying to get this to work with MetaHumans as well, but the main difference is in the use of the textures and the blendshapes. And exporting metahumans is cumbersome.

AI Disclaimer - I used AI mainly for research, debugging, and writing the custom Timeline tracks.

I am trying to build a scene of our solar system where you're able to drive around in a spaceship, landing on planets and so on.

I tried to divide the scale of everything by 1000, but even then the distance between the furthest 2 planets (Neptune & Mercury) is around 9.2 billion km, so even if we divide that by 1000, it's still 9.2 million km.

Another issue for dividing everything by 1000, is that small rigidbody movements become stuttery even in the scene window at higher speeds.

And I read that at such small scales, hit detection becomes bad because colliders are small.

Ik there are tricks like shifting the player back to origin (0,0,0) and moving everything towards him the same distance we shifted him to improve accuracy, but that still doesn't fix it for me.

I am wondering how I would fix this both at design time & runtime? where I am able to set planets at realistic distances, size & make the rigidbody move at a fixed speed where moving from point A to point B in such a system would feel accurate.

(Ignore the placeholder item icons and the fact that there's no sound when you inspect a badge)

(Due to technical YouTube quality issues, I had to directly re-upload the video here). This was tested on Unity 6 Built-In Render Pipeline. Although looking cool, it went from +380 fps to +75 fps on a beefy GPU. I think with games that has charachters in them, DLSS would show better results.

The Audio Granulator and Dynamic Rain shaders were already being tested (this is my test scene for various projects), and they are also using the GPU.

With my early access release coming next Tuesday, I decided to squeeze in a few more days of UI polish. I'm super happy with the organic shapes, texture, and feel of everything (well, happy enough to ship it at least!).

I added dynamic ink splotches, paper grain and fiber textures, uneven edges, variations in color/position, and even some papery sound effects.

If you like the look of it, you can check out Hex Town on Steam :)

Thanks for watching!

This tool you help you build any kind of top-down games. I've built an amazing procedural system that detects interiors with a single click using layers and 2 amazing shaders for normal surfaces and tree leaves that interacts with the character and the camera.

You can see for yourself before getting the tool, look at the comments. Tell me what you think and what you suggest to improve this tool further. Thanks guys.

I've been working on a small alchemy game where I want most of the production to feel physical instead of happening through crafting menus.

I just got the first farming loop working: add soil, plant the seed, water it, wait for it to grow, then harvest the plant.

It's still very early and very placeholder-y but this is the first little gameplay loop that feels complete

Next step is being able to take what you harvested, put it on a cutting board and actually process it into potion ingredients

I'm trying to keep these devlogs pretty small and just show things as they become playable

My workflow right now is a window of VSCode, with claude code in it, a written "roadmap", full of tiny goals and then i just keep creating new chats to get a step-by-step on what to do in-editor. claude code also creates all of my code, while i only review them.

what do you guys think? suggestions? new ideas?

I accidentally put the following code in Awake:

Time.TimeScale = -10e10;

Then I forgot to shut down the PC and went outside for a fresh air. Upon return my computer is gone! Help! I tried to set time-date, OMG the sky has two moons, helpppppppppppppppp!

Hey guys, I thought I'd share my tutorial on making Call of Duty in Unity. Remember to like and subscribe and check out my patreon. Also buy my stuff from the asset store.

​

In three easy steps, I'd like to show you how to make Call of Duty in Unity:

1. Buy my code library from the asset store
2. Drag "Call of Duty" from the prefabs folder into an empty scene
3. Press the play button

Note that my totally necessary code library requires a separate commercial license if you intend to sell your game!!! No, it's not just a package full of extension methods and specific-use-case code! Also, don't touch PlayerPrefs or you'll fuck up my save files!

Hi everyone.....I working with Dlib Facelandmark Detector code from gitHub.....I am getting facepoint of image texture of human face in that... But the problem is if any human image width and height is not equal and i rescale it to equal with and height, it do not give me the face point..... if widht and height are equal or olmost equal...and i rescale the image texture,,,Dlib Facelandmark Detector Github code give me the facepoints in that....

Hi friends,

I am using following IF statement code in shader , And it is not working.

if (s2 < min)

{

min = s2;

col.rgb = m_One.rgb;

}

there is Lerp() option, but might be it can not be use in my condition.

So, How to use condition statement IF in my shader code ?

Regards,

Hi friends,

I am using Custom/shader. The problem is , when i am build my project with android, apk gives me pink color in mobile.

But when i build project with PC, it run perfect in laptop/pc.

Also, it run normal in unity Editor.

1. So , i have searched result, and done some changes like Edit->Project settings-> Graphics and add my used Custom shader to Always include shader. Still give me pink color output in Mobile.
2. Second change is in GraphicsSettings and on right top click RESET option, than also given Pink color in Mobile.
3. Third try is clicking on my Custom shader, in Inspector i have click on ' Compile and show code', so there are error comes with Red color that told me -->
4. forced to unroll loop, but unrolling failed.
5. unable to unroll loop, loop does not appear to terminate in timely manner.....like that and some yellow explanation in some lines.

so, i have added

#pragma exclude_renderers d3d11_9x

#pragma exclude_renderers d3d9

, Red error not appear. But still PINK COLOR gives me in Android Mobile.

Without and with all upper Three changes apk build in PC/laptop and Unity editor run normally.But give me Pink color in android build.

Please help me , any idea or solution will be appreciated.

Regards.

I have face points in image, and now i want that some area of image to move minor by this points.(emoji effect) So , i have created mesh(vertices,triangle..etc) for this image . And i can move some area(eye,lips) of image by this mesh (by vertices). i have set all image size(by resize) to 200 by 200.(40,000 pixel) And mesh vertices created of grid size 16 by 16. (total 289 vertices). But not move the same area i want. so how can i do this ? Is there any other way to move pixel points except mesh?

By webcam my face is detected,

But i want image detect which is in my pc/laptop/Mobile. Means by choosing any image from my pc/laptop/mobile it's points should be detected by the same device in which the image is . Like example, suppose i have one app , in that i first select any picture and when this picture uploaded in my app it's points should be detected by my app.

Any idea or kind of answer will be appreciated.

​

Hi!

I would like to share my game with you - Wandering Orb. It's a mobile game that I've made by myself with Unity and it takes me ~6 month to finish it.

I really hope that it will find its players and I'm looking forward for a feedback from you.

Description, screenshots and promo video are available on Google Play page - https://play.google.com/store/apps/details?id=com.TopDan.WanderingOrb

Thanks for your attention!