Ryan Brucks

Testing a method to combine gerstner waves with shallow water simulation. Particles read the waves and feed back into the sim. I'm doing a livestream tomorrow @2pm showing upcoming #UE4 .26 Water tools (waves, caustics, foam etc). Be sure to stop by! #UnrealEngine #GameDev

Wheeeeeee, I had so much fun working with clouds on this. Learned some new tricks, as usual, from @ShaderBits

@gamesstudent That looks like a bug. What do you mean by “RT” though, the default engine pp refract is not ray traced just a simple screen uv offset. Are yoy talking about ray tracing?

@ShaderBits I may be missing something here, but I'm not sure what, I have Niagara particles using the engine DefaultSpriteMaterial, it has nothing plugged in to refraction, but when I toggle the Post Proc's refraction the entire particle gains refraction.

@ShaderBits is there any clean way of getting translucent Niagara particles in UE4.26 to have no refraction other than by turning off refraction in the Post Process? It's really bumming me out.

@Stefan_Maco Not currently as it’s not easy to do. If you just straight up blend wave values it will do crazy looking things that do not feel right. You can do it visually though by just having another waterbody somewhere in the world and using that index and blending waves in the material

@danielelliott3d I actually made a little debug shader to help show these artifacts and debug them too. Will share this soon.

Was trying to work out some of @ShaderBits maths from the great raymarching blog post. The planeoffset seems to be a spherical slice. I've worked out how to do a flat plane and will compare results once I've got the raymarcher all working.

@danielelliott3d The only 100% solution there is icky: take a fractional step to hit the plane first, and another fractional step at the end. So far Ive not added that since most volumes don't have smoke right at the edge. But for those cases maybe just turn off plane snapping.

@danielelliott3d hmm that part shouldnt be necessary since it should do max(t0,0) on t0 before calculating the frac to the plane. But there is another bug with plane offset: if your ray was only going to fit 1 step, and the volume has fog everywhere, it causes another seam by jumping forward

So here is a way to verify I've calculated the right final distance delta to trace to remove artefacts when compositing volumes against scene depth (as shown in @ShaderBits blog and in GPU Gems 3). I'm lerping between the render and the extra step with contrast. They line up!

@danielelliott3d Also what benefit do you think fixing the plane to the nearest vertex vs the center will have? Stabilize the movement a bit? Be curious to see if it helps. Also some of the differences in sampling become harder to notice once the fractional step comes in.

@danielelliott3d You are right I didn't account for the spherical trace from using a normalized vector. Funny thing is I just recently fixed that in ue5 niagara shader. I think dividing thestep size by dot(camvec, normalized planevec) works well.

@milesmacklin @TimSweeneyEpic me too! except I ended up using 2 atan calculations, then I saw an X-Lab presentation showing your version with only 1 atan and I felt dumb :D I actually went and tried to find your post recently to see how to reduce it down to just 1 atan. couldn't find it for some reason...

@TimSweeneyEpic Hah, cool. I derived the same integral a few years ago: blog.mmacklin.com/2010/05/29/in-… blog.mmacklin.com/2010/06/10/fas…

Real-time volumetric fog was pretty mind-blowing when it debuted back in 1997. youtube.com/watch?v=qjgHgg…

My buddy Alan's band Stone Whiskey made some videos. Check em out! youtube.com/watch?v=2IPkW-…

The Unreal and Unity april fool jokes is unintentionally the best comparison of the engines ever made. twitter.com/UnrealEngine/s… VS twitter.com/unity3d/status…

According to sources, VISA requires 149 KWH of energy to perform 100,000 transactions on its global settlement network. The Lightning Network is over 1 million times more energy-efficient than this outdated network. If you are still using VISA I guess that you love wasting energy

@ImaginaryBlend Looks great! We will most likely be hooking up the niagara shallow water solver to the water system by default in UE5. It should enable a lot of fun custom stuff!

Stable and mass conservative shallow-water model implemented in UE4 with only 2 RT(256x256) draw calls. @ShaderBits I feel that you will rise a bar higher in the next engine update anyway :) youtu.be/hGeJgW3KA6o #UE4 #indiedev

That Heather looks amazing. I love some hyper detailed ground cover.

@MagnusL3D Yes! So awesome to see people picking this up already. Great use of the niagara grids.

Fiddeling around with the new simulation stages and render target possibiltys in #UE4 4.26 #Niagara updates. #gamedev

Lol

Wow this is so cool! This was the first game I ever got to work on (not counting ut2004 which was just one level). We had a lot of fun making this back then.

@OrfeasEl I will probably end up doing that. I just made my diagram backwards, as it has some panes of text included with long notes etc. basically tried to make the diagram into the readme instead :D

@ShaderBits Have you considered converting it to an image and attach it to the readme file? Syntax: raw.githubusercontent.com/orfeasel/BotAr… Result:

@LucianoJacomeli On my list, but no promises atm. Its also on our longer term roadmap to support impostor generation as a built in option for HLOD. Don't want to get your hopes up too much since that is likely next year, but I think it will be pretty interesting.

@ShaderBits You will update ImpostorBaker to be supported in new Unreal Engine versions? Your original code works too 4.21. I think only Impostor_ThreeFrameBlend have errors in 4.22 - 4.25