#archived-hdrp
1 messages Β· Page 18 of 1
also not sure if this matters but ive tried changing lit shader mode to forward instead of deferred and it didnt seem to do anything either
Can you confirm what the input textures contain at this stage though?
#archived-hdrp message
it looks like the normalbuffer one doesnt have the fps layer rendered in front
Yeah, so that might be the cause...π€
Can you see where the normal buffer is rendered normally? Perhaps there's a "GBuffer" or "Base pass" or something earlier in the queue?
this i see is right before the custom pass
That's "Dbuffer" If it's anything like in Unreal, it's only for decals.
Also it seems like the output is a depth buffer.
yea theres no gbuffer or base pass that i see in there
Can you expand the forward depth prepass and take take a screenshot of any draw call inside(with more event details visible - not just the output)?
Aha. This looks like the pass that renders to the normal buffer
this is my particle effect i have running in the scene thats the only thing that uses bloodshader
There are 12 draw calls there
oh then that was just one that happened to be for the blood effect?
when i disabled it its like the same but says a lit shader
Yes. The details are per draw call, not a pass or category of draws. When you click the category it probably just shows the details of the last draw call
so basically i need to get the pass to run before that ? or update the normal buffer after when my custom pass runs
You need it to render to the normal buffer. Doesn't matter where, as long as it's before the Ambient Occlusion pass. Though I'm not sure how to achieve that.
yea i feel like ive tried everything i can think of within the options of the editor
Maybe try playing with the render queue here:
i went through all of those and that one produces the best results some either make it worse or dont render it on top of everything
Other than that, can't you use a camera stack or a separate camera?
nah not for hdrp especially i also steered away from that in urp cause i was having issues with lighting
it sucks cus its so close lol almost got lucky and had everything work no prob
Donno. A custom pass really feals like a wrong approach for something like that. I'd try the camera stacking approach:
https://docs.unity3d.com/Packages/com.unity.render-pipelines.high-definition@17.3/manual/graphics-compositor-window-reference.html
i could get camera stacking working on hdrp but i think id still have the same lighting issue and inherent performance cost
I'm having this weird issue with my material's height map. How can I fix this. The material is an hdrp lit material btw
Yeah youβd need to modify the render pipeline so it caches some things between the cameras.
I havenβt looked into the details very much but thereβs a high base cost to rendering anything at all in HDRP (this is why real time reflection probes are unusable even at 32x32) and I suspect much of that work can be shared
<@&502884371011731486>
!softban 791971685213536266 bot spam
zynxer_ was softbanned.
hey unsure if this is the right place but for using HDRP/LayeredLit, does the main layer take the splatmap's alpha channel? cuz that's not working for me rn
Sanity check: the depth value in a depth texture only relates to the Z coordinate (relative to camera), rather than relating to the distance to camera (i.e. relating to all XYZ coordinates relative to camera)?
E.g. as long as a point is at the same distance on the Z axis, it will have the same depth, regardless of the X and Y coordinates of that point?
It should, yes.
Thought so, yeah. But it's weird, as it's hard to confirm on the internet, where everywhere I look it just carelessly mentions "distance to the camera" without making the distinction/specification that it's only the Z-axis distance.
LinearEyeDepth in Common.hlsl handles the conversion
I think that just handles the conversion between non-linear value distribution and linear value distribution.
I don't think it gives you the Euclidian distance between the pixel and the camera, only the Z-axis distance.
Thinking of going from urp to hdrp, anything crucial i should be aware of?
Aka anything urp has that doesnt work in hdrp
Camera stacking doesn't work in HDRP
thnx, its been a smooth transition so far, im tryna figure out what settings i should keep enabled or disabled. i wanna keep disabled the non-vfm settings aka those who hit performance hard for a small visual upgrade
That's subjective, but personally I keep the shadow filtering on low on all quality settings.
Under volumetric fog, there's a setting for the maximum amount of volumes you can have on-screen and by default this one's really high. You can save quite a bit of VRAM there.
I have it set to maybe... 4? Used extensively
u mean the max local fog which is set to 256 by default?
Yeah, that one
256 is insane. That's 256 fog volumes on-screen at the same time
Who uses that many?
bruh there are so many values i need an extra person just to go through them all
are there other settings that are set to unnecessary high values?
You can shrink some of the atlases for things you're not using
and i havent even touched the graphics tab, im only looking at the quality > hdrp
Yeah most stuff is in there
Like if you're not using point light shadows, you don't need to reserve so much memory for them and you can set their atlas to the minimum value
what about some odd checkboxes like stream reflections from gpu and disk, should these be enabled for optimization?
I'd probably leave that one off unless you're having a memory issue
There's another one called screen space shadows - it has no quality impact, but it's something that can either help or harm performance depending on the situation. So you might want to test in a build whether it helps your game.
If you're using volumetric fog I'd not use the quality presets, you can get better performance and stability with custom values.
yea, i enabled all screen space toggles to test them, i also wanna see what exactly high quality line rendering does, considering the camera already has anti aliasing
That's for the hair system
i would never guess
any idea what the buffer format does and what i should set it to? i have left the default value for now
More color accuracy. I've set it to the lowest.
I think it might make a difference on HDR displays? But if you're targeting PC there's almost nobody using those so I'll just take the extra performance, thanks.
i have the hdr settings off cause i work on an oled screen
You'll need to test HDR support at some point. It's possible to break it with improper tonemapping
is it the allow hdr display output check in player settings? or is there another one?
also i noticed it hase the auto windows graphics api disabled, should i enable it?
If you're not using ray tracing or wave intrinsics (you'd know if you were using that) I'd recommend setting it to DirectX11 only.
when i enable the hdr output in player settings it says its not enabled in the pipeline but dont see anything hdr related there
I don't know
Who wanna build a game doe fr
!collab
:loudspeaker: Collaborating and Job Posting
We do not accept job or collab posts on Discord.
Please, use Discussions to promote yourself as job-seeking, advertise commercial job offers, or look for non-commercial projects to participate in:
β’ Collaboration & Jobs
which setting affects the resolution of the clouds?
Cloud map resolution
cant find it either
wait i think i had to add override to the volumme
still dont see resolution tho
its weird cause some pp effects are enabled but some others like screen space GI i need to override to manually enable
Yeah thereβs the default volume
I donβt recommend screen space GI without ray tracing. Itβs not nearly as good and performance cost is similar
what about the rest SS options like shadows, AO and reflection?
by default volume u mean the slot in hdrp tab or in graphics tab
Screen space shadows are a performance option. It can either harm or help performance, you'll need to test this
SSAO and reflections are good
You'll need to test if you want to support the ray traced options. I personally don't because RTAO doesn't help with foliage, and I don't like the sharpness of RT reflections
camera relative support on VR getting closer for HDRP: https://github.com/Unity-Technologies/ScriptableRenderPipeline/commits/xr/hdrp/camera-relative
currently you have to turn that off to get any shadows in VR, this can't come soon enough π
Hello there. So i try to use the new visual effect graph comming with this hd render pipeline. And after one month i still havnt found how to make my own 3d vector field textures in unity. The examples come with vector field but none explain how to make them ? Any idea, tutorials ?
The only tools i found are 2y/o 40 buck assets. There must be a tool in the new unity right ?
@blissful geode you know about https://github.com/Unity-Technologies/VFXToolbox ?
I can see that they have vector field examples there too (https://github.com/Unity-Technologies/VFXToolbox/tree/master/DCC-Tools/Houdini/Examples/VectorFields) but I don't know houdini myself
this paid plugin also can convert fga to vfx graph textures: https://assetstore.unity.com/packages/tools/particles-effects/vectoraygen-official-vector-field-plugin-fga-3d-texture-support--97239
but the actual conversion must be trivial
considering fga is just 2d array of values
I looked a bit in the repo, and it look like vectors fields examples are made from houdini just as you said. Exporting vector fields shouldn't be hard but before spending ton of time learning a whole new software, I would like to know if someone else actually used the visual effect graph and made vectors fields successfully inside unity.
It is pretty unclear how to make them for me and I mean... If unity released a tool and a demo using vector fields (The unity logo with milion particle, the genius from the lamp, the demo vortex sun graph...), there probably be a way to make them with the said tool ? There really isn't any doc or tutorial about this anywhere.
Still, thank you @turbid matrix !
well, the shape things are probably using SDF for the actual shapes and some may use vector fields for the actual effect
I do agree that it would be nice to have some plugin that did all that baking inside Unity
and I'm sure Unity will do something about it eventually, it's still very early days on this
Speedtree v8 shader on the way for LWRP at least: https://github.com/Unity-Technologies/ScriptableRenderPipeline/commit/c0b5acdd7a1c3b09356780b034a829948eb559d4
hope they get HDRP done for it soon as well, altho Speedtree itself is sure taking it's time on getting their v8 Unity modeler out...
There was a render pipeline for PS1/PSX visuals but the last time I used it, it was outdated/not working.
@zenith hedge The one made by Keijiro? I think the API used to make that is out of date now. π¦
yeah if I remember it was just turning off lots of stuff lol to get back polygon jitter and stuff
@dawn sorrel Sadly, yeah.
it's probably not too hard to recreate tbh
There's also some free PS1 shaders for Unity on Github but they're not that customisable.
With the new Custom Render Pass API in LWRP, It shouldnt be too difficult to replicate a similar visual. π
@dawn sorrel I'd love that. A classic pipeline that renders Unity games like PS1 games with utmost authenticity, but then ways of making the experience cleaner, but still reminiscent of PS1 games. For example, weak vertex jitter, minimal affine texture mapping, more than 16 bits of colour. So it feels similar only without the graphical hindrances.
Im not sure what you mean by 'utmost authenticity, but a cleaner experience'. Do you have a specific example for the 'emulation'?
By utmost authenticity, I mean as if you were developing a game for PS1 hardware, shown exactly how it would be on a n actual PS1/PSX. The alternative is like when you remember PS1 graphics in your head, only it's not as outdated, if that makes sense.
Oh, I understood that! I just didn't understand the cleaner experience part. π
Nostalgic things are usually seen through rose colored glasses, it might be very hard to make it still look appealing
So ideally the cleaner/more cleaner version is supposedly how one would see it through rose-tinted glasses, lol.
For anyone curious with what we are talking about: https://github.com/keijiro/Retro3DPipeline
He also wrote a Retro 3d shader about 4 years ago. Looking at the shader, it should still work today for the Built-In Renderer; and not too difficult to rewrite for LWRP (I haven't written a shader for HDRP so have no idea how it compares): https://github.com/keijiro/Retro3D
There's one by Dsoft too. I think that's the name.
Is there any way to get HDR Pipeline to not eat so many shader keywords?
It, plus poststack v2 eats somewhere in the order of 180 keywords.
@full fulcrum you know that PPv2 is going away on 2019.1 (for HDRP)?
Oh for crying out loud.
they already got PPv3 integrated on HDRP's master branch
it works on recent alphas
ppv2 will not even work there
it's literally replacing it
That's gonna be a mess when it comes to people writing plugins for LWRP/HDRP/Legacy
yeah, I agree
Just about everything in HDSRP seems to be designed to cause people to lose hair.
I'd love to know how they plan PPv3 to be extended
now there's no info about it yet
and current PPv2 docs literally say it'll work on LW and HD RPs too π
Yep
which is still true today
but not on next version
anyway, just wanted to give you heads up if you weren't aware
Yeah thanks, that's really useful.
on PPv3, everything is moved to HDRPs scene settings
We're running a user generated MMO built around Unity, which we're planning to migrate to SRP later this year once it gets a bit more stable.
you know the thing which has HDRP sky, reflections etc setup in
and porting everyones content automatically between versions has been ... shall we say fun, and HDRP has just made that more so.
well, HDRP is currently planned to get out of preview for 2019.3
but it was supposed to be out preview originally even full year earlier so you can't really count on such long targets to hold
Visually it's a very nice compelling upgrade
But upgrade-wise, it's been absolutely terribly managed so far. One of the nice things about Unity has been historically, upgrades just work.
E.g. going from Gamma->PBR, the old shaders still work. (we've been upgrading Unity projects since 2.1/2.2)
well, to be fair, it's in preview
This one, .... everything breaks. And breaks again.
it's same with current ECS
Yeah; but even stuff like Surface Shaders -- there's no good reason for them to break.
things move around much
Surface shaders just output metalness/diffuse/etc values - that should easily remap to the new renderer.
I can understand the odd feature like _GrabPass going by the wayside, but there's been a lot I've gone "grrrrr." over with this.
also stuff like PostStack V1->V2 lost a bunch of useful functionality (like tonemapping pre-pass when outputting to LDR)
Mainly just kvetching! But it's annoying.
there are some funky things coming up as well
this was spotted on github https://github.com/Unity-Technologies/ShadowMapsVoxelizer
that's probably related to this package Unity is now working on SRP branch: json { "name": "com.unity.voxelized-shadowmaps", "description": "This proposes a data structure for precomputed shadows.", "version": "0.1.0-preview", "unity": "2019.1", "displayName": "Voxelized ShadowMaps", "dependencies": { "com.unity.render-pipelines.lightweight": "5.2.3", "com.unity.render-pipelines.high-definition": "5.2.3" } }
I'd assume this is pretty much equivalent to UE4's relatively new volumetric lightmaps
like, not the implementation but the use case
I don't think Unity itself has mentioned this officially anywhere, but then again they haven't really spoken publicly either about the DXR raytracing things that are now merged to HDRP master
also woop woop https://github.com/Unity-Technologies/ScriptableRenderPipeline/pull/2718 π
noticed that being broken a while ago (PR fixes HDRP SSR in forward)
already tested it too, it does work again, altho it still doesn't work with MSAA
would love to know someday if that's actually planned as supported combination π
it is
they got HDRP camera relative VR on the works too (https://github.com/Unity-Technologies/ScriptableRenderPipeline/pull/2719)
that'll be nice as we don't need to disable camera relative in the future for it
that used to break some shaders in past
I haven't made custom experimental merges anymore, could probably do some soon again as there are some nice new things coming up that I'd love to try
I did bunch last fall where I just manually merged all cool upcoming HDRP feats in experimental HDRP branch so people could try them out early on
the feats weren't fully polished but was fun to get heads up on upcoming things, like how they worked
and could also report some bugs really early on
I fixed some small things on my own branch and when Unity eventually merged the new things to master I could PR the fixes right away few times π I didn't tell them I had the fixes just waiting there for days before
I mean, I had to do the fixes to use my experimental branch anyway, didn't do it for sake of sending them to Unity
Hello everyone, I know that Lightweight render pipeline is still in preview, so I shouldn't use it in a commercial project, right? I'd like to use shader graph but I guess it's risky... what do you think?
Beeing in preview "only" means that the support on it is not as extended as for released features: non high user pain bugs won't be backported for example. But you still can use it for comercial projects.
And you can upgrade during your project development, maybe it would came out of preview ?
We've tried to start our big internal project on HDRP, but it failed on many points, forcing us to switch back to legacy. I'm not sure about LWRP, looks like a bit more complete/ready for production than the HDRP.
@atomic dove worth noting that LWRP will be out of preview really soon, they target 2019.1 for that
so even if you start using it today, not much will change to the point when they make it go out of preview (as they can't make any big changes anymore at this point if they want to make it stable)
I could be horribly mistaken on their plans π that's my assumption tho
From my own observations, I wouldn't be surprised if LWRP becomes the default render pipeline, and people just stuff more HD features into it over time.
Since it's the one that's more broadly compatible (in almost every sense of the word)
LWRP should be out of preview in 19.1, one of our goals for 2019 is to close the feature gap between LW and built in renderer so that LWRP will become a replacement for built in.
Hey @remote forge that sounds good - I have one question, re: shaders. Is there any reason why you guys didn't just expand the Standard shader to make the LWRP/HDRP Lit?
It seems like 90% of what we're doing in all three shaders is just describing real world properties; and ideally we'd fill in every map we can possibly put, then have the engine pick which ones the hardware can support.
It'd be nice to have just a "Real World" shader that had a ton of properties/parameters, of which Unity doesn't support a bunch .... yet. (or not in all renderers/etc) - but long-term it's a stable material definition.
Its a complex situation. Builtin renderer standard shader is a bloated beast atm, and has some key incorrect lighting calculations which we cannot change for fear of breaking live projects
(among other general design issues)
There are two things here which are connected, but different. There is the interface to the shader (that is the shader lab properties that exist in the shader), and then the shader itself (specific passes, specific data processing, and specific operations).
The first part here, the interface does have some similarities between pipelines but LWRP isn't a strict subset of HDRP (i.e things do not overlap 100%). We didn't want to confuse people here and make a mega shader where only some things work when using LWRP. On a practical side, these are also run from different packages and worked on by different developers so keeping it in sync and high velocity is pretty hard.
The other part (the shader side) is not possible, each pipeline has different passes that exist as well as different levels of complexity in the actual shader code to target differnt hardware levels.
Then HDRP and LWRP shaders (and their libraries) are purpose built for maximum performance/quality for their respective hardware targets
(Stramits answer was better than mine π )
@elfin osprey / @remote forge I'm thinking more in terms of defining "One shader to rule them all" similar to how say iray does their shader defs
well, not nessecarily shader.
more one material definition
Shader Graph ^
Yeah, I know - is dual-compatibility with HDRP/LWRP a long-term feature of Shader Graph?
in practice, developing and maintaining a system like this leads to sub-par shaders on all targets
I know, I'm trying to balance more long term projects.
We've been building/upgrading a MMO since 2008 on Unity
(we were like #2 after that Nickelodeon one)
Yes, the long term goal for shader graph is generation of equivalent shaders for any and all render pipelines that have defined backends for it
We've migrated through just about every renderer Unity has, and the HDRP/LWRP split is ... a bit frustrating.
Why do you think that (really curious as we designed them to solve very differnt problems)
Well, different platforms, different visual goals - we run really cross platform, target almost every build target on offer.
Some of them, like desktop PCs - we want to use HDRP.
Not all desktop PCs though - older ones, LWRP seems best.
But phones, switch, modern consoles, etc.
Each is going to fit a different pipeline.
Visual disparities aren't a huge breaking feature for us - just the ability to offer the best the platform can support.
I know that's not everyones use case - but with things like the Switch out there, I suspect it could be more common than you might think.
if you only use PBR and Unlit Graphs, they mostly compile on both LWRP and HDRP afaik?
I think there are some LWRP specific nodes but I don't use LW myself
it's kinda weird concept but I get why someone would want that
but as side effect, there's now PBR and Unlit but also HD Lit and HD Unlit graphs (names might be different but you get the idea) on recent SRP git for HDRP
To claify, our intention is that these HD specific master nodes are temporary and will be replaced by a more robust, render pipeline agnostic, solution in the future.
In the short term we needed them to get all the HD features to our users (internal and external) in a timely manner.
Fair enough, and that's a useful bit of insight - I tend to operate on a fairly long term view with Unity, e.g. I'm more concerned about 'what comes after HDRP' and making sure we're not rewriting thousands of shaders
(I already need to come up with a hack for converting Surface Shaders to SRP...)
was just going to write "surface shaders for SRP when"
Seriously, I'd love it.
Right now there's no easy way to make a shader that works on legacy, and SRP.
yeah they were great
I could still make small modifications to HDRPs shaders manually, that's fine but maintaining stuff like that is PITA when new versions come out
this is what all Unity's tech demos etc do atm
like BotD, new FPS Sample, I bet same applies to MegaCity at some extent when it comes out
and if there's bigger change in SRP, it takes a ton of effort to port things again
like for example, to get BotD environment running on current HDRP version, you'd have to figure out a ton as there were huge changes between 2.0.5 what they use and current 5.x
I mean, if you want all the custom tech they did there
I'd still love to know what's the deal with occlusion probes π
like is that already forgotten thing, is voxelized shadowmaps partially replacing it (it's fundamentally different thing but has similar effect)
or well, replacing is wrong word as OB's were never integrated on official packages
nice to see that camera relative VR patch is finally on HDRP master - no need to manually edit the SRP files for it
Is it still true that LWRP doesn't have sprite renderers? Does HDRP have them? Or are 2D games stuck in the legacy pipeline?
I found this https://forum.unity.com/threads/are-sprites-not-supported-by-the-srp-shader-graph.526280/
so you probably can work around it with shader graph shader
I dunno about the current state, I don't use LWRP, I only know what happens on bleeding edge HDRP
(they don't have Sprites working out of the box on HDRP)
We dont support sprite output in shader graph right now im afraid. It can "work" but we dont support sprite renderer features like masking etc
The question is not necessarily about Shader graph, I'm interested in general 2D sprite workflow in these pipelines and whether they will eventually work well with 2D (more than just giving minimal support for sprites be viable) - I get that right now it's not a good idea.
Sure, I was just responding to Olento's statement. 2D will be supported in SRP world, but I cant speak for timelines on that, or the specifics.
Understood. Thanks!
One message removed from a suspended account.
@misty charm checkerboard rendering like used on consoles?
One message removed from a suspended account.
One message removed from a suspended account.
@misty charm isnt checkerboarding usually done as a gpu feature?
ie. its not done via RT's etc by hand
its something you enable on the GPU
It's something you could definitely use a render pipeline to do - but I think you may have to wait for the experts π
One message removed from a suspended account.
@true zealot no but i mean to get an actual 4k checkeboard rendered texture
you have to ask the GPU to do that
idk remember the name of the optiion you use
One message removed from a suspended account.
its called multi resolution shading i think
it requires lower level access to the gpu than unity provides via the API to enable
i dont know of any way to enable it from unitys C# api
but im not a rendering guru so i might be wrong
heres some tests some guy did https://forum.unity.com/threads/rendertextures-merging-issues-for-custom-checkerboard-rendering-shader.456890/
donnu how useful it is
One message removed from a suspended account.
from a reddit post: As far as I have seen, not without the sourcecode of the engine. The scriptable render pipelines do not give you enough platform dependent options to do it yourself. :( asking if you can do CB in Unity
One message removed from a suspended account.
afaik yes
One message removed from a suspended account.
i have never seen a single example of working CB in Unity
and afiak the reason is you dont have access to enough platform dependant stuff
again I could be wrong, but I did look into this a few months ago out of curiosity
im sure you could fake it half decent with RTs and stuff
One message removed from a suspended account.
but the whole idea behind CB is to increase performance with as litle visual quality loss as possible
and whenever you start jumping through hoops
you usually give up performance
and kind of negates the original point anyway
One message removed from a suspended account.
One message removed from a suspended account.
afaik no, the SRP is still fairly hard to get into if you want to make your own from scratch
i tried... (was a while ago) and gave up, just too much information/documentation missing
One message removed from a suspended account.
the basics of SRP is simple, there is some examples somewhere
but other than like getting a simple mesh to draw, you're on your own afaik
@misty charm Catlike coding is currently making a tutorial on writing your own SRP :) https://catlikecoding.com/unity/tutorials/scriptable-render-pipeline/
One message removed from a suspended account.
One message removed from a suspended account.
SRP 4.7.0-preview incoming for 2018.3: https://github.com/Unity-Technologies/ScriptableRenderPipeline/releases/tag/4.7.0-preview
it's at least on staging atm, probably soon on regular registry (unless it's there already)
that should at least fix the HDRP unlit graph on shader graph
@turbid matrix nice
@calm elm thanks, im gonna go through that one
thank god someone had energy to figure all of this crap out without any documentation π
and tell the rest of us who are way to lazy
4.7 is on PM (might need to enable showing preview packages etc ofc)
ah, cool
wonder if Unity will release 5.x for recent LWRP changes soon, there are some HDRP changes but most of them are still on the other WIP branch
Has anyone figured our how to get baked reflection probes to look better on highly reflective surfaces? The pipeline asset's refection cube resolution setting seems to only affect realtime reflection probes.
Yeah I've tried that extensively I even bumped up the cookie sizes for good measure. My generated exr cant exceeded 128 in size. Not sure what is going on
Closed and reopened Unity. Deleted and recreated my reflection probe. Didn't help.
lol looks like i cant recreate my problem in another scene so far
The mystery & magic of Unity
Ah yes, keep you guys posted if an answer develops.
heh, this SRP batching thing doesn't quite do what I'd expect π
without batcher:
that's just some placeholder model from blendswap on my test scene, not really game ready and has tons of individual materials, they are almost all set to HD Lit
I think the mainbody is still sharing only same material across all submeshes and it still breaks only on some parts
@turbid matrix Could you maybe send your scene so I can test it ?
sure, it'll take a while to take the extras out but can strip it to smaller repro project probably (if it still breaks there)
you prefer I submit it through issue report or just share link here?
You can send it in PM
will do
would kinda wish it would break on some simpler model, this isn't quite typical mesh but it could be the reason why it breaks too
every other mesh on my test scene is fine
I can also make whole Unity crash by removing "wrong" parts from the mesh prefab on prefab editor but that's probably totally different issue related to prefab workflows
this is 2019.1 alpha tho
Or could be an issue cause by the combination of the SRP batcher and the prefab view
You like living dangerously π
I guess it could be
well, you have to be on bleeding edge to try latest hdrp master π
And that's not even the most unstable and risky, trust me π
Hey, small question... I dont think my game requires the HDRP but i want to use the tools it provides (decals, particles, etc.) Should I use it anyways or use the LWRP?
HDRP team has been pretty cautious on recommending HDRP's use in production so far, it's even right now in the middle of big change and there will not be a officially stable release until late this year (if their current targets hold)
LWRP however is targeted in getting out of preview for 2019.1 so it's safer bet
that being said, I'd still try them both and see how they work for your project and use cases
and there isn't anywhere near as many showstoppers on current HDRP as there were last spring when they started promoting SRPs π
(by promoting, I mean blog posts, GDC and Unite talks, tech demos etc, HDRP team specifically have been really clear at the same time it's not officially production ready)
re: SRP Batcher, that issue is now on issue tracker here https://issuetracker.unity3d.com/issues/srp-batcher-doesnt-support-32-bit-vertex-index-buffer-mesh-on-dx11
tbh, I'm surprised to find the HDRP to be pretty much stable for me even thought it's not production ready.
I still keep breaking things from it π but many things that break now are due to new systems or restructuring of it
they've done quite few such along the way
but in general, it's pretty usable already
biggest issue from my own perspective is on scalability. afaik they still have lots of systems that run in preparation for some features and they are all for nothing if you don't enable those feats. You'd want to have really bare bones "simplified" setup for low end hardware so you don't have to put outrageous min spec for your game
of course a lot of the perf depends on the assets and levels you have but still, there's still overhead in the HDRP itself that could be cut out for weaker systems
Epic has "simple forward" mode on UE4 which they've used on their own games, it basically forces everything to medium tier mobile level graphics to guarantee they can run their games on really weak gpu's
their main renderer still has certain basic overhead which will slow weaker gpu's to crawl if used there
that being said, I have not done myself any serious benchmarking on HDRP yet so would really have to test these on final levels to get better understanding of the situation
Is it a renderer with submeshes?
Should report a bug as a case like that would be easy to spot as anyone into perf typically does not go for submeshes
that was basically just a mesh with 13 etc materials π
it wasn't really game ready thing like noted earlier, just a placeholder
I see yeah so a ton of submeshes. I can see that breaking real good
well, it only needs 2 materials to break and >65k mesh (unless you force 16bit indices) π
I can see that more probably scenario than the mesh I have right now π
@severe leaf Kink3d (who's here on the discord) is developing https://twitter.com/MatthewDean3D/status/1063919555242688512 and has indicated they're planning LWRP support, so there may be decal options for the LWRP in the nearish future π€·
Released: kDecals. Forward projector Decal system for #unity. Create Decals at edit or runtime with automatic pooling. Multiple render modes and a script API for custom Decals! Package in releases tab if you want to try it. #gamedev #madewithunity
https://t.co/oVoGsC80d2
I honestly don't get this thing https://forum.unity.com/threads/onprerender-onpostrender-counterpart-in-srp.535875/
I know Unity has made it clear that their old callbacks had issues but they've been postponing this thing for so long now, basically everyone wanting to hook in custom things is left hanging unless they want to implement the hooks manually (which isn't big deal for people who work on projects but it huge limiting factor for people making assets for sale)
how I see it right now, almost every time Unity itself uses HDRP for some demo or extensive sample, they add similar hooks manually as even they can't work without
positive thing here is that we get the sources and are able to do it ourselves but would be nice to have some built-in solution too
Vertx, just to quickly point out that is a private project and not affiliated with Unity in any way π If I add LWRP support to it there is no guarantee it will make it back to LWRP, however I dont think I need to alter the render pipeline itself to achieve this. π
Oh yeah I totally get that! I'm just recommending your work π
Danke π Just have to add the disclaimer sometimes ya know
Does HDRP have support for volumetric lighting in 2018.3 ?
yup
@sharp hatch yes, it's built-in now
if you've ran the FPS Sample, it uses it on the level too
i have, and ur right yeah
talking of HDRP and features, spotted new branch on github for adding dynamic resolution for HDRP (they just added it to LWRP)
@turbid matrix do you know if there's any tooling for packing the HDRP Lit maps?
M = metalic
AO = ambient occ
D = ??
S = Specular
what is D here?
depth?
i suppose S is smoothness, not specular
but ye
ye not specular
I'm also going to tell you something that you didn't know
realized that
that's exactly the same packing that old standard shader has used so far, besides having the detail mask in blue channel
if you don't need detail mask and have used the old packing efficiently, your textures work as is in HD Lit etc
look closer to it π
they had separate slots for these
but they only used specific channels from them
it was designed so that you'd only have one texture for these all
oh yeah sure its the same channels
but I doubt many users realized it
oh yeah that im aware of
but i want a tooling to combine them into one texture for HDRP
there are few 3rd party utilities that do the texture packing but I dunno, I'd probably write custom if you need these for some specific purpose, it doesn't take that long to write
yeah no its usually pretty quick, just figured i would save myself the effort
Alloy has their own packing tool for example (free on github, I also have it ported to package manager for on my github so one can just drop in the github link to packages/manifest.json and it should work
but Alloy is overkill for this if you don't use the shaders
i mean its just remapping channels to a new texture
just figured if something was available i would not have to write it myself
I actually don't know if the material conversion tool does the packing too
it totally could
you know the one that does ship with HDRP
I'd check that first if I actually needed this
@turbid matrix for reference im importing arch visualization assets into hdrp
I'd just write a script that you can batch run automatically for set of textures if there's a lot of them, would save a lot of time
The material conversion does the packing
you mean the legacy to HDRP converter?
Yep
@scarlet hull i can not get it to work
or rather, does it trigger automatically?
all my old materials are just pink basically
You need to call it by hand, it's in the edit/renderpipelines menu
ah
yeah, that's the one I referred on the one they ship with
I've just ran it to get some 3rd party samples to show something on HDRP, but I've never checked if they took in the additional maps
that conversion tool has been since the early 2018.1 versions I think (never looked what it did behind the curtains)
It kind of existed before, but wasn't doing the repacking job
Also packs the detail maps btw
so technically, if you need a custom batch packing tool, you could hack it from that too π
hmm what the hell is going on the edges of the table leg there
does it do that when you run the game (in game view)?
TAA is there to help smooth spec issues out, but many of the issues do originate from the materials themselves
ye
HD Lit has some specular AA options I think, altho I'm not sure how much of them were on 2018.3
I don't think they've even merged all in from current github PR's even
StackLit has more options for this
so if you do Archviz, definitely check that one out
it's not meant for realtime games as it's heavier
@turbid matrix im not doing arch viz per say, just using arch viz assets to try to push the visual quality a bit, figured i would use the HDRP for even better quality
For the packing stuff : the material converter does the packing through a class I wrote : TextureCombiner. I wouldn't be very hard to make an UI to use it individually, I just don't have the time to do it, as the converter now works
here's another asset
if you look at the steel bars under the chair
its the same issue
@scarlet hull I was more of thinking if people need to batch convert a ton of similar texture setups (in non-Unity format), having a batch script that used some hardcoded values for iterating through all those would save tons of time, it doesn't even need fancy ui then
like, imagine you have PBR material library that's all authorized some specific way and you have hundreds of these materials there
no fucking clue whats going on
it does not look like that in the standard deferred pipeline
Other way to do it : make a new shadegraph material that uses this PBR packing and the same property names, and just change the original materials shader π
sure, but a lot of these library materials aren't really packed in efficient way to begin with
would be just huge waste of texture samplers and memory
pretty hard to tell what's going on from those images π
but I'm not a renderer guy to begin with, I just use these things π
ditto
@turbid matrix do you know any good tutorials on the HDRP? How to set it up from scratch properly, deal with importing (mega-)scanned assets, etc stuff like that
for quixel megascans, you should use their own bridge tool, it packs those materials so they can be used with HDRP automatically
other than that, I guess this post? https://blogs.unity3d.com/2018/09/24/the-high-definition-render-pipeline-getting-started-guide-for-artists/
ah nice gonna read that
getting some weird pixel like artifacts between the tiles also on a floor
oh thats just the pom setting
okey so it doesnt transfer material properties perfectly using the convert tool
what ?
@scarlet hull well i used the conversion tool, and it messed up the height map extrusion setting
so the value in the inspector was like 0.015
but that wasnt the actual value being used
that setting
Well, the height map extrusion value can't really be "converted"
im not complaining per say, just saying what happend
the editor read 0.015 as it does there, that's after i set it to 0.015 by hand
but the actual value being used was MUCH higher
more like 1.5 instead of 0.015
The displacement algorithm is totally different. On the standard shader the height value is use to do an uv distortion, while in HDRP it's raymarched per pixel parralax mapping
Guess I could take a look at it
@scarlet hull Again, the issue wasn't that the value of 0.015 was WRONG, it was that it showed 0.015 in the inspector... but it used some other value
i touched the inspector value once
and it worked after that
also for completeness sake
that weird.... specular aliasing or w/e you wanna call it
went away after i upgraded to 4.6.0-preview
it looks correct now
i was on 0.4.1-preview before
Cool
just upgraded and it looks correct, so idk what was going on and probably doesnt matter
a lot has changed even from that 4.6, things are moving really fast on HDRP land
how do i get the later version?
4.7 isn't on your package manger?
no
it should at least have fix for Unlit shader graph, it's broken on 4.6
ah, maybe it's still only at staging registry and github
for newer HDRP, you have to use 2019.1 alphas
Still in staging
2018.3 still gets some things backported but dramatically less now
as HDRP has gone through big changes recently that can't really be backported anymore
If you want to have a newer newer version, you'll need the 2019.1 alpha
if you stick with package manager versions, HDRP 5.2.3 seems pretty similar to 4.6 at first glance
Huh ... can't even remember ... changes to fast
there are some additional shader graphs
@scarlet hull When is it going to slow down and become 'stable' ?
Moving fast is great
But it's hard to do something with it
Or well, try to base a project on it, if it moves so fast
2019.3 is what they target internally afaik for getting it out of preview
unless that has shifted again (I'd expect 2020 tbh)
all the bleeding edge stuff is still only at github atm
they've been working on hdrp specific wip branch for a while which contains all the new cool things
but also things we can't use atm, like DXR
oh yeah, just waiting on the 2060 to become available here
so i can play with DXR a bit
outside of Unity for now
i dont really game, so just want the cheapest RT capable card
there's that raytracing stuff, HDRP specific new postprocessing, VR improvements, more shadow filtering options etc
yeah, me too
might get RTX 2060
my current workstation is a beast of a machine
eexcept the GPU
GTX 660 π
i dont do shit with rendering or game at all
no new splash screen for 2019.1? π¦
same as 2018.x π
Will maybe come for the final release ? I guess it's not the priority right now π
yeah doesnt really matter, just always fun to see which one gets picked :p
tbh
I hope the splash would at least give option to use the same as on their youtube videos
that's really nice, but the one we get is just tiny static image that can be zoomed a little on splash
okey on 2019.1 alpha with hdrp 5.2.x something
the scene camera has acceleration now π
GODDAMNIT
gonna take a year to get used to that
scrollwheel thing?
that's great
I just wish it went further than 2.0
but you can always hold shift
I just don't do that automatically when I need it for some reason
i can get used to this
it's actually similar to ue4 scene navigation now
AAND SEGFAULT
π
well, alpha
most of my random crashes have been for having the new memory profiler just installed there (I could be paranoid about that, I swear everything is way more stable as soon it's gone)
You sir are very fast
I just checked if there's update on github for hdrp-master so spotted the tag for release
currently trying to solve an issue on my project
basically console gets 999+ errors asap I try to examine the camera on editor
anyway, about that 4.8.0-preview, here's changelog for those whom interested:```[4.8.0-preview] - 2018-01-16
Fixed
Fixed remove sync with render thread when updating decal texture atlas
Fixed issue with TerrainLit not compiling with depth only pass and normal buffer
Fixed geometric normal use for shadow bias with PBR master node in forward
Fixed error message when having more than one directional light casting shadow
Fixed issue with MicroShadowing and SpecularOcclusion
Fixed issue with vertex animation in shader graph not compatible with SRP Batcher
it only seems to affect HDRP, there's no changes for other parts of the SRP on the changelogs
oh snap
the camera error thing I have when using HDRP seems to be specific to happen now on even stock HDRP 5.2.3
@sharp hatch you still have the alpha open? if you click on camera object, do you also get flooded with error messages like NullReferenceException: Object reference not set to an instance of an object UnityEditor.Experimental.Rendering.HDPipeline.HDCameraEditor.SceneViewOverlay_Window (UnityEngine.GUIContent title, System.Action`2[T1,T2] sceneViewFunc, System.Int32 order, UnityEngine.Object target) (at Library/PackageCache/com.unity.render-pipelines.high-definition@5.2.3-preview/Editor/RenderPipeline/Camera/HDCameraEditor.Handlers.cs:61) UnityEditor.Experimental.Rendering.HDPipeline.HDCameraEditor.OnSceneGUI () (at Library/PackageCache/com.unity.render-pipelines.high-definition@5.2.3-preview/Editor/RenderPipeline/Camera/HDCameraEditor.Handlers.cs:23)
@turbid matrix let me try
like, with the scene view opened
yup i do
scene view has to be active for that to happen
not the exact same one
ah
yeah, there's like 999+ errors in no time
it has to be a14 etc related change
I first thought it only happened on bleeding edge github version but I've now tracked it to happen even on default HD Template scene
I'm also pretty sure this didn't happen on a12 or older, but would have to install one to be sure
never used anything before a14
in moments like these, you'd wish the hub would actually let you install older versions as well :p
need to go to the archive now to test where it broke
uh
π
why would it want to remove old unity installation
hub itself lets you install multiple minor versions
I just ignored the warning for now, seems to still install
@turbid matrix Β―_(γ)_/Β―
also installed to a new folder so it's not that either
ok so, it's a14 thing
it doesn't happen on a13
im just bitching in #π»βunity-talk about the half finished state of things
I'll file a report for it
(not from your bitching)
well, HDRP isn't supposed to be ready yet
@turbid matrix i know
but half of the features are "it's not ready yet"
ecs, burst, srp/hdrp/lwrp, new networking, etc.
all is 'not ready yet'
π
the other alternative would be that Unity quietly worked with these for years behind the curtains and didn't get our feedback on things
I dunno, I very much still prefer this approach
honestly that sounds like a much better approach :p
current approach makes their development so much more transparent
and if people don't like WIP stuff, they can just use what's marked as stable
i'm fully aware of how works :p
it just seems to trigger people that they see the preview ones sitting there for some reason
or what the idea is
I sent this feedback along recently to Unity - not sure how it is for other studios, but its difficult for my team to meaningfully engage with new features. Spending time to provide feedback for WIP stuff seems like a real luxury
only real complain I totally get is ditching old systems before new ones are fully out
yeah, that too ^
@plush pilot yeah nobody working on a real project has time for that
I see stuff in 2019 and think "cool, my games in 2021 might be able to use that"
ye
that's not even that far fetched π
especially if we think of state of ECS
it's not going to be finished this year, like have all engine parts ECS'fied
i'd be surprised if it's ready before 2021
that's less than 2 years until 2021 so
but it creates this problem
where the engine constantly taunts you with
the new features that are not quite ready
that you cant use yet
and right now it's a SHIT TON of features which are not ready
you should have used UE4 between 4.0 and 4.10 :p
unity really seems to like throwing away and starting over
networking - thrown away twice even, old rendering system, go/mb system vs ecs, etc.
i mean i understand it's still there
I'm personally all for what they do now
but how can you ask people to put effort into using system which we dont know when/why they will be deprecated
and not worked on any more
it's finally driving the engine further instead of doing the same old thing they've done for past 15 or what years
i very much prefer iterative improvements that are stable
over this shotgun approach of re-develop every core feature of the engine at once
unet aside, rest of the "stable" systems still work
also, mind you LWRP is supposed to be out of preview for 2019.1
yes, but when will they stop being updated? do they have more development resources assigned to them for further improvements, etc ?
if you look at the packages, they don't even say preview anymore
same with shader graph
all of it creates huge question marks basically
about what to use, will it be future proof, etc.
all kinds of crap
it might work for hobbists, etc.
I think the messaging around "the v1 of this feature is deprecated and depriortized, but the v2 of this feature is only in preview mode in the latest" isn't fantastic
then again, they did remove the preview status from the hub and it still feels like early beta with it's limited functionality π
but like trying to build something that is future proof on top of Unity at the moment feels like an impossible task
even if it's an asset or a game or whatever
the whole engine is in flux
Cryengine is currently at similar situation too, things get messy when you have to redo a lot of old systems
yeah but nobody uses cryengine π
and same happened on UE when they did 4.0
j/k ofc
it literally took like 2 years for UE4 to feel fully featured
they didn't even have UI at first
like, game ui
given the options of having stuff in preview mode available for comment vs not and just waiting till its ready - I agree the former is much more preferable
other than the hardcoded slate setup that wasn't documented anywhere
sure, but that situation is different, they didn't have a bunch of existing projects IN UE4
just important that they're calribrating how much commentary is reasonable
they did a clean cut vs UE3
a lot of people moved over from UDK
and that silence doesn't imply that everythings right π
@turbid matrix yes but that was their choice right? We can either stay with the old deprecated stuff, we know what it does, we know where it is, etc. or we can go with the new unsure thing
also, you still can use the old systems in Unity, it's not like the engine is broken without new but still in preview features
currently the engine is being heavily modified while people still work in it
it's a difference
I can't think of any other thing that UNet where they cut the cord too soon right now
it's not about that
but even in case of UNet, it was far from ideal to begin with
it's about that knowing which parts of the engine to use for a project is not easy
since you dont have a clear word on say when is the old rendering API going stop having updates done to it
well, officially, you'd use the parts that are marked as full releases
nobody from HDRP team will tell you to use it on production
they actually say the opposite
you're missing the point im making
right, maybe what fholm is getting at is production cycles are long
you start a new project on the legacy render pipeline because its the approved one
take SRP vs old rendering
and you ship your game 3 years later on something thats deprecated
SRP is new stuff, its not ready
Old stuff works, its ready
ah
but there is NO WORD on when its going to switch
if you target on shipping in 3 years, definitely use SRP now π
haha
and sure, if there was ONE feature that was like this
but right now we have:
- Rendering
- Game Logic (ECS)
- Networking
in this state
vfx
input too :p
yeah
see what im getting at now?
whole engine is in flux, and it feels like there's no direction, no clear roadmap when things will switch over, etc.
just a jumbled mess of old vs. half finished new features
you'll love it when we get HPC# Physics
yes sure, so 2023?
(just trying to point out that there will be more WIP systems to this)
but I'm not really understanding what you'd prefer instead
slow steady cycle, incremental and stable changes over time
Unity is keeping legacy systems in place until they can be replaced fully
instead of shotgunning the whole engine
you can't make huge changes like these in incremental fashion
that's actually what's brought all the technical debt to Unity in the first place
then keep it in internal development until it's closer to production ready
they've just kept adding things
I like that they're being pretty bold with the scope of the changes they're making vs letting the engine rot
but agree with the overall sense that it's kind of an awkard place to be in atm
yeah
not sure what the answer is other than buckle up
but to backtrack a bit
and get to RP's since that's what this channel is about.. HDRP is pretty usable even on 2018.3 right now
things being unstable on 2019 alpha.. well, that's alpha
yeah not complaining over that
I'm guessing Unity didn't do betas and alphas before because they were worried people would expect stable things
just dropping that here in case some others face similar issue, no need for others to track down what caused it
oh wow
it only spams more errors if you hover the mouse cursor over the scene/game view WITH camera selected and HDRP installed
that's like super specific combo π
I feel like the approach they're taking is a sensible one, and fairly clearly communicated (and I know this isn't RP specific, but we can move if someone has a specific question)
The old pipeline works, and all the legacy systems are still there and functioning, they also receive bug fixes, just not updates
so that part of the engine stagnates somewhat
but everything else is steaming along as fast as possible, isn't recommended for games being released any time soon, and therefore is a fantastic early preview for those who are extremely invested (us weirdos)
whilst also giving them ample bug reports
The only better way it could go is if the legacy pipelines were stably updated with new features during this time, but you can completely understand how infeasible and ultimately wasteful that endeavour is
yeah, that's also pretty much how I feel about it
altho I don't feel they should put any effort on legacy besides keeping it running
it would just make the switch harder in the future as there wouldn't be as much reason for users to switch to the new thing if old one is almost as good
plus those extra resources could just be used to make the new thing get out faster
right now the biggest pain point for people seem to be the development time
not that the legacy systems would need updates
I think I'm fine with that provided the replacement has exited preview
Actually there is something they could do better: documentation
you mean in preview?
in general, all across these new systems
I dunno, there isn't really many that have gotten out of preview π
they're very inconsistent in format, location, and descriptive accuracy
they start to document them more when they are getting closer to release
deprecating a system before the new one has exited preview presumes some success on the new direction
like for example, SRPs have gotten recently a lot more docs
which obviously we hope is true, but they've built and scrapped v2s before: see HLAPI
yeah, but that's also pretty much only thing they've done that for
and they really didn't want to keep patching UNet
which I kinda can understand
still sucks tho
multiplayer wise, Unity isn't in best position and is has never really been
their own systems have been half baked (earlier setup) or bugged (UNet), paid third party offerings have been sketchy, now we have few free ones that are updated pretty much by few individuals so they lack resources...
@true zealot - yes, our documentation definitely needs to be better across the board. I can't say much other than: We know. We're working on it.
In the meantime, if there's every something specific you see missing from docs (a use case, an example) or there's an error, please let us know.
it's definitely a lot better than it used to be! I can give an example of what I mean by inconsistent location: click View documentation on the Visual Effect Graph in the PM and it takes you to a page without documentation
so I then have to google and hope to hit the official docs
Yes! So that's a peeve of mine too.
We're working on tooling for Package docs and the overall Manual, to try and consolidate them, and improve searchability etc. I can't give a specific date - but I can assure that us writers are pining for it.
In the meantime, I'll try to research why that link even exists if there are no docs. :)
Not really focused on render pipelines but that's the closest channel for that.. Do you guys know if there is a way to add the DynamiclyScalable flag to a RenderTextureDescriptor? For some reason that setting is not available on it, and it's a pain : (
Documentation is one of the hardest jobs given that they're documenting something that's practically out of date the day the ink dries.
Specially with alpha/beta/experimental
A nice to have, I think maybe. Not got a use for it atm
Updated FB demo:
Some people have been asking for it, so here it is.
Error downloading object: Assets/Demo/Video_Timeline_Profiles/Camera01 (2) Profile.asset (e5c6ace): Smudge error: Error downloading Assets/Demo/Video_Timeline_Profiles/Camera01 (2) Profile.asset (e5c6acef153ee3d87b6442b96b103f85c839f4a4ff0c0095cbbfdc8d35274373): batch response: Rate limit exceeded: https://github.com/Unity-Technologies/FontainebleauDemo.git/info/lfs/objects/batch
whups π
can also do this if you're lazy (like me)
You sure? it didn't work out for the FPS assets
gave stubs π
I don't think the zip gives you the LFS part either
tbh, they should have put this earlier in the readme
that was half way through a git clone btw.. i cloned again and it worked π looks great!
now make one with realtime gi! π
mine is still cloning
LFS while using git from command prompt is suuuuper slow for some reason
this is my internet connection while downloading with git LFS:
I mostly get 1Mbps peaks where it mostly just transfers some kilobytes π
yea lfs is slow in general for me .. its supposed to be faster if you use git lfs clone, but feels the same tbh
and it would normally go to around 50Mbps on regular downloads
yeah, it's the same for me too
hm well i did clone the fps sample just now with git lfs clone, took only 30-40 mins.. maybe there's less/bigger files idk
The demo now additional decals fabric, particles, volumetric and planar reflection and is use to illustrate some technology of HDRP
so should look quite different now
We are starting work on integrating PPv3 on LWRP and then this will be supported.
I totally thought PPv3 on HDRP was super specific to HDRP itself
Plz, plz re-use the PostProcessProfile class / types from V2
I've written now four conversion utilities going between old school ImageEffects -> CinematicEffects Beta -> PPv1 -> PPv2
I may break if I write another. π
they wont
I know because I'm using v3 already from github
if you've used HDRP, you probably know they have their own scene volume asset for it
it stores sky, fog etc stuff
they moved them out of the lighting
PPv3 has moved PP volumes contents to that scene volume asset
it's similar concept tho
@full fulcrum
π’
let me grab a screenie
Yeah I can imagine, I've poked at HDRP a little bit, I vaguely recall that
I just really, really wished we didn't have different components for different SRPs
For us, it's a pain since we need to write migration scripts depending on which SRP we're using (we're planning on using both LWRP and HDRP for different platforms)
but for asset store creators .... O_o
this:
will create this:
which has that profile there which is just a scriptable object, like in PP volume has it's own this as well
you add components there just like in PPv2
they got most PPv2 effects ported to that already
there are some extras, like panini projection
I wonder if it re-uses the same classes/types.
they have their own setup for this, but structure could be similar
you don't install PP package anymore with this btw
there is no PPv3 package
this is now integrated in the SRP package
us that scene setting thing just for HDRP or LWRP as well?
I've heard a bit of confusion here and there regarding HDRP not being too compatible with Intel... I'm going to assume this isn't actually a thing, but does anyone have any idea if this is true or not?
Not true
@devout thorn Exept if you have a REALLY old intel GPU, it should work. Probably slowly, but should work.
are you sure about that?
because what I've read from peoples experiences, there have been gpu's that have failed to run the game
let me find some posts, I don't quite trust my memory on specifics π
I just published a game with HDRP: https://nothke.itch.io/trails-of-tenere
I had a few complaints that it appears that some player's GPUs don't...
hmmmm, I'm reading the messages on the itchio thread, the users who had it failing aren't really super specific on their GPU specs
so it's totally possible that these users are really running the thing on some ancient laptops with lesser than DX11 integrated GPU
I have few old integrated intels here (all DX11 worthy) that I could test this with tho, but I'd not target a game to run on these today
@devout thorn @scarlet hull I could bet that forum post is the root source of people being iffy on integrated gpu's + hdrp as even I remembered it despite having a terrible memory π
I also think there was some other thread that discussed about this and some bare minimum HDRP GPU requirements
I have tested HDRP on intel integrated GPU, and it worked.
but... people do try to run your game on toasters anyway. I've been wondering about what to do with it myself a lot as I afaik HDRP does have a lot of initial overhead atm
It's all I can tell. It's pretty hard to test on EVERYTHING
sure, especially in PC space π
it's like trying to make sure your android mobile game runs properly on all phone models out there π€
you mean you haven't tested on the Direct3D Wall of GPU History? 
I'm actually rethinking about the HDRP scalability thing, Unity staff usually brings up the point that everything works differently on HDRP to other pipelines so swapping the pipeline at runtime for weaker platforms isn't recommended
but...... what if you actually made custom scaled down version of HDRP itself that literally only implements the bare bones setup using the same scene controls, lighting setup etc (and swap to that on weaker platforms)
I can already tell that's probably not worth the effort, but just a thought (still far from impossible)
I hope HDRP itself can do this at some point
thing like this could also be an asset you'd sell on store, like slimmed down variants of individual HDRP versions for letting people to use the same scene setup as with stock HDRP and let people just swap the pipeline based on quality settings
I dunno, I'd buy that π
as different topic, I see that Unity is bringing voxelized shadowmaps to LWRP on 2018.3 too: https://github.com/Unity-Technologies/ScriptableRenderPipeline/commits/vxsm-18.3/lw
this makes me pretty curious as all the work is being done in LWRP and there's no HDRP implementation at all atm
Cheers for the clarification guys @scarlet hull @turbid matrix
hmmm
I have this weird reoccurring issue with texture streaming now in 2019.1
but since it seems to happen totally randomly, it's hard to write a bug report for it
basically what happens is that sometimes objects with standard HD Lit shader and Alpha clip don't actually clip the alpha, but this only ever happens if I use texture streaming
if I manually touch the alpha clip threshold, it renders the asset (tree leaves) properly again
Yes, that's not the best reproductible case you seem to have here
I have actually an idea, I'll try something and report back
worth noting that I've seen this only in editor so far but I haven't really done that many testbuilds with texture streaming enabled
oh well, can't repro on my minimal project, I had this theory that when I saw this happen, it usually happened after some library refresh but I can't really trigger it in any way
Hey @turbid matrix I remember you saying that you used light layers on HD in 2019.1. I can't find where I can customize it :0
I haven't used it, I just know the settings on HDRP Asset...
Enable the light layers in your HDRP Asset, and then use them in your lights / probes / objects
yeah, but does it like automatically work if you just use different layer per light ?
that's what I'd try first but haven't seen any info on the setup
Yes ?
@dawn sorrel ^
I'm trying but light layer is grey
And I don't see how to select a layer after that
Or is it the game object layer?
(sorry btw, someone called me)
@dawn sorrel it's gray because you haven't enabled the feature
Oh, where do I do that? π
basically everything in HDRP works like this, you get the main toggle for some functionality first, like a master switch and rest are just options you can config for different passes or change how it works etc
Light Layer : Choose which rendering Layer this Light will affect. This Light will only light up Mesh Renderers with a matching rendering Layer.
so.. I'd assume it means the gameobject layers
Hum I tried but I guess I did it wrong, I'll dig into it ^^
There is a layer index inside the meshrender component now, maybe not the same
oh
Okay found rendering layer on the mesh renderer
there's light layers specifically
but I need to put my light inside a layer now
I have no idea how you can set the light itself to these layers
If I find it I'll tell you ^^
ah
oh wow
this was like not obvious at all
wonder if it works like this on older HDRP
but I figured how it works on latest π
That little icon x) Gosh nice one
yeah, I was wondering where the additional settings are now
there used to be regular checkbox for it in past
but they changed the design
this will not be obvious to people....
I don't think I have special use cases for these myself but good to know the current setup regardless
@dawn sorrel what's your use case?
Not really mine so I won't be able to explain (I should ask), but it's to have a specific rendering on a decor. I think he didn't want the main object to reflect on the decor.
oh wow
basically, this would let us use those occlusion probes with shader graph shaders "properly" without radically custom SRP
at least I think it would
light bake just crashes the editor on 2019.1 when used with occlusion probes... oh well π
========== OUTPUTTING STACK TRACE ==================
0x00007FFD5059A251 (ntdll) RtlUserThreadStart
========== END OF STACKTRACE ===========```
it's really not giving me much to go with
I think I got that one
I'll not really put more time into this, will see what Unity will do with the callbacks when they eventually port this to 2018.3 (altho it's totally possible their old callback still work as is on 2018.3)
for example I noticed they use different convention for FPS Sample callbacks
but it would require some restructuring on whole occlusion probes scripts logic
Hope it will reach HDRP as well. Currently the tiny weapon trick is unusable for me.
it'd be as usuable as it would be in VR
(ie none)
you can't really do the fake gun on top rendering in VR either :p
or was that the point you were trying to make? π
No, but I need to render some things in front to prevent minor polish clip issues. I have exausted all over avenues and used to this fine with my own engine π
some people seem to think the separate rendering pass is some standard thing that every engine gives out of the box, without realizing that out of modern engines, Unity has been one of the few that has allowed that
for example the enemy will thrust with his sword and you will block it but his sword passes through your shield in an ugly way, or world objects can penetrate it slightly. This is an issue in VR as well, and touch ups are needed π
ue4 doesn't support any of this stuff for example
you could make it all physically correct but then the game would play quite bad in my case π
and there doing custom passes is like 10x harder than with say, Unity SRPs
I'm saying it's a valid thing to ask for, and I can script the SRP myself to add this, but I don't really want to as Seb's said it's subject to change quite a bit til 2019.3
I wouldn't want to make big changes to SRPs either
so a native feature for tighter opaque rendering control is nice, it's purely for polish though
To be fair on UE, it doesn't really support anything much at all except for fortnite these days
If it's not suitable for their core plans, it's unlikely it'll get added
ok so
SSR just got fixed on latest hdrp-master
it broke a while ago when they added raytracing stuff
I also kinda wonder if the HD guys even want any reports for this wip stuff, considering they probably know already many of these things being broken
but that being said, I actually think there's something fundamentally broken about the reflect sky option
this is the main reason I never have it enabled:
I get similar issues on all test scenes where I've tried it
here's better image to describe the issue:
(don't mind the SSR settings here, it's setup here this way for debugging reasons, the effect is still there on more sane reflection levels)
Anyone else getting a problem with LWRP 4.7 and 4.8 where overlay cameras (like UI cameras etc.) are clearing the render buffer to the slid background colour?
Really? That seems pretty crazy...
I believe you that it happened, I can't believe that Unity actually did it. It hamstrings so many techniques essential for game development.
I'd disagree on that, you can do most things without that hack
for example, engines like UE4 do just fine without it
that being said, if you read the last post on that thread, they are making custom passes more approachable so people can work around the limitation
I agree that there are ways to work around it, with render textures, etc. but to eliminate stacking before those replacements are in place is a bit crazy. For instance, uGUI can't be rendered in perspective now with SRP, and you'll always have that giant canvas because you can't put it into a screen space camera anymore UNTIL Unity gets around to implementing those things, whenever that is. And it feels that the other solutions are more hacky than just being able to stack cameras on top of each other, it's intuitive.
Anyway, not looking for a fight, I'm just flaggergasted at this sudden move.
Thanks for the link though, I'm checking it out now.
ah the HUD/UI thing does make sense
you wouldn't want things in front of it
I do get the benefits from the stacking too, reason why I feel it's not all that necessary is because it usually enforces bad practices. If they'll make a separate pass you can inject easily, it actually solves this and makes the process more obvious to user (instead of having magic happening behind the curtains like it's been so far)
As long as there's easy to find code examples I'm happy
re: removing feats before replacements: apparently happened now with shader graphs custom nodes
based on the random things I've seen here from Unity staff, I'd guess they are replacing the custom c# node code with custom node inside the shader editor, like it's done in ASE, ShaderForge, UE4 materials etc
that being said, I don't even know if they have a release where it's removed, so my point isn't super strong here π
Current Custom Node API for Shader Graph is being removed in 2019.1.
@turbid matrix how about UI particles on SRP project? They don't work properly when using screen space UI, you need to use camera screen space in order to have the z order set correctly. Did they provide a fix for that?
@exotic robin tbh I didnt know that was a thing :D