Deathrey

I spent around 65 hours researching, implementing, and writing for my next "Gotta be less than 2ms" newsletter article. In this article I explore Julian Love's famous Diablo 3 VFX talk from GDC 2013. I go over how you can implement the techniques from his talk into a modern engine like Unreal Engine 5 and some of the problems you may face when doing so. I show the process from start to finish on how to recreate some of the FX demos that were shown and provide the IlluGen project files so that you can have source materials to try and replicate the effects for in engine verification purposes. I had a lot of fun investigating this and I certainly learned a lot more than I had anticipated. Comment on the Linkedin newsletter version here: linkedin.com/pulse/investig… Read the full article here: jangafx.com/insights/diabl… #realtimevfx #gamedev #indiedev

My "No Graphics API" blog post is live! Please repost :) sebastianaaltonen.com/blog/no-graphi…

@toreler Branch out after completing first loop only if the outcome is the same for whole warp, not just one thread. Specifics needed to say more.

Shader gurus out there! Is it possible to optimize the following? I'm stumped. I have two for loops. Both do texture samples, but the first one can return if it never finds what it's checking for. So often the code ends after the first part. I improved the first loop, reducing registers, etc. But the overall performance dropped instead of increasing. Since the first loop is more optimized, the GPU starts prefetching more in the second loop. Which in turn wastes fetches on passes that end up returning after the first loop.. 🤦 Is there a way to improve this that I don't know of? Feels wrong to leave it "unoptimized" even if it gives the better overall result.

@GhislainGir You can stay simple and brute force. Start guess at two segments, evaluate with scoring like mean deviation. Pick segment with worst deviation and subdivide it into two. Relax points along spline, refit and repeat until mean/peak deviation is acceptable.

I managed to write C code that performs bezier fit/least square to approximate a fixed number of points with a single bezier segment given the start/end points, but coming up with a solution that best solves the entire path with a varying amount of control points eludes me.

Any math genius out there? Looking for ways to approximate a cyclic path made of discrete positions with a bezier curve made of multiple continuous segments.

In preparation for my talk at Unreal Fest Bali this week, I've made a small blog post on how to create a custom FMaterialShader in Unreal: victorcareil.com/making-materia…

@GhislainGir But a match exists. You store such bit pattern, that on output of floating point 1- op, gives you bit pattern that you initially wanted to encode. You need to make a lookup table for whatever program that stores your data.

@Deathreycg Hmmm, I don't see how. Applying a 1-x on a float that contain arbitrary bits seems to scramble bits in a destructive way. Doing one more 1-x operation to reverse the operation doesn't return the original bits.

#UE5 any chance for this to become an option so we can bit-pack data into the V channel? :D

ue5-main #diff-16714c117b9f52fe868d4eda12d22ce37d9cff913fbc157f73d818d4e653c5b5" target="_blank" rel="nofollow noopener">github.com/EpicGames/Unre…

@TimSweeneyEpic Basis change time. Get rotated!

Indeed Unreal Engine is moving to Left-Up-Forward coordinates everywhere, starting with UEFN, and coming to UE5-6 in an incrementally-adoptable way through UI settings and C++ helper functions/macros to ease the transition. This will align Unreal with Y-Up, right handed standards of USD and glTF. Why? Because future 3d tools and ecosystems will be increasingly interoperable and standards-based. There are a lot of missing standards we’ll need to propose, and Team Unreal will be far more successful proposing new things if we adopt and add to existing standards and conventions. The USD-glTF-Maya-Houdini quadrant is the center of mass for complex code-art-pipeline tooling that is highly sensitive to coordinates. (Flipping coordinates when exporting from AutoCAD or Blender is easy enough; changing a movie vfx pipeline is not). Coordinates based on project settings sound like a have-it-your-way compromise but are a combinatorial mess when projects are a mix of code modules and content packages from many independent authors. The best time to make this change would have been 1995, but I believe the second best time is now with the launch of Scene Graph in UEFN.

Mark your calendars 🗓️ ! ✨🧱 BRICKADIA 🧱✨ lands in Early Access on July 11th! To celebrate, we just dropped a new montage trailer (100% real gameplay!). The most chaotic, creative multiplayer sandbox is calling... are you in?

External pressures have caused us to buckle and recognize AI as king. We're deleting the node graph from our tools and making our tools prompt access only or PAO for short. You need to type what you want to get it, and just have to live with the results that come out on the other end!

Happy holidays and may you never be invited as dinner 💚

#ResistanceIsBrutal now has a steam page! Battle thousands of enemies trying to bring an end to your miserable life. If it looks interesting please consider wishlisting! store.steampowered.com/app/3378140 #screenshotsaturday #solodev #indiegame #indiedev #UE5

update on Asher's Slime game💚 The dungeon's popularity is plummeting! Only the Slime can make it fun again. Lots of magic sodas are needed to get the job done. #indiegame #gamedev #techart #slime #game

Latest gameplay video of the graveyard map for my #solodev game #ResistanceIsBrutal, built on #UE5! #gamedev #indiedev

PSA - Use the link below to get free & perpetual licenses for all 17,000 MegaScan assets. They become paid next year. This is what your library should look like after clicking the button. It's a very good button as buttons go.

Announcing Voxel Plugin 2 Runtime Nanite Nanite Tessellation Runtime Lumen Unlimited Materials Release early 2025 #ue5 #gamedev #indiedev

@BenSimsTech There is no guarantee that after texture filtering, stored components make up vector with length of one, hence the normalize that you deem unnecessary.

Found another unnecessary piece of Unity shadergraph code which won't be optimised away. Normalizing a normal vector after reconstructing the Z component from the squared length of the X and Y components is unneccessary, since the Z component is guaranteed to make the length 1.