# Objective
PBR lighting was broken in the new renderer when using orthographic projections due to the way the depth slicing works for the clusters. Fix it.
## Solution
- The default orthographic projection near plane is 0.0. The perspective projection depth slicing does a division by the near plane which gives a floating point NaN and the clustering all breaks down.
- Orthographic projections have a linear depth mapping, so it made intuitive sense to me to do depth slicing with a linear mapping too. The alternative I saw was to try to handle the near plane being at 0.0 and using the exponential depth slicing, but that felt like a hack that didn't make sense.
- As such, I have added code that detects whether the projection is orthographic based on `projection[3][3] == 1.0` and then implemented the orthographic mapping case throughout (when computing cluster AABBs, and when mapping a view space position (or light) to a cluster id in both the rust and shader code).
## Screenshots
Before:
After:
<img width="1392" alt="Screenshot 2021-12-13 at 16 36 53" src="https://user-images.githubusercontent.com/302146/145847314-6f3a2035-5d87-4896-8032-0c3e35e15b7d.png">
Old renderer (slightly lighter due to slight difference in configured intensity):
<img width="1392" alt="Screenshot 2021-12-13 at 16 42 23" src="https://user-images.githubusercontent.com/302146/145847391-6a5e6fe0-22da-4fc1-a6c7-440543689a63.png">
This makes the [New Bevy Renderer](#2535) the default (and only) renderer. The new renderer isn't _quite_ ready for the final release yet, but I want as many people as possible to start testing it so we can identify bugs and address feedback prior to release.
The examples are all ported over and operational with a few exceptions:
* I removed a good portion of the examples in the `shader` folder. We still have some work to do in order to make these examples possible / ergonomic / worthwhile: #3120 and "high level shader material plugins" are the big ones. This is a temporary measure.
* Temporarily removed the multiple_windows example: doing this properly in the new renderer will require the upcoming "render targets" changes. Same goes for the render_to_texture example.
* Removed z_sort_debug: entity visibility sort info is no longer available in app logic. we could do this on the "render app" side, but i dont consider it a priority.
Fixes#3043
`surface_texture.present()` will cause panics if no work is done on a given frame. "Views" are how we queue up work. Without any cameras, no work is produced. This adds a "clear pass" for windows without views, which ensures we clear windows (thus doing work) every frame.
This is a "quick fix". It can be made much cleaner once we make "render targets" a concept and move some responsibilities around. Then we just clear the "render target" once instead of clearing "views". I _might_ have time to tackle that work prior to 0.6, but I doubt it. If "render targets" don't make it in to 0.6, they will be one of the first things I tackle after release.
# Objective
Fixes#3247
## Solution
- Added docs to `Axis<T>`. This one can be extended @alice-i-cecile
- Added unit tests
- Added `min` and `max` values to the struct. @alice-i-cecile Do we need to these variables?
- Limited `set` method usage with `min` and `max` values
Co-authored-by: CrazyRoka <rokarostuk@gmail.com>
# Objective
Fixes#3255
## Solution
- mark the `bevy_gltf` feature as required for some examples
This should be cleaned up when we remove the old renderer
# Objective
Port bevy_ui to pipelined-rendering (see #2535 )
## Solution
I did some changes during the port:
- [X] separate color from the texture asset (as suggested [here](https://discord.com/channels/691052431525675048/743663924229963868/874353914525413406))
- [X] ~give the vertex shader a per-instance buffer instead of per-vertex buffer~ (incompatible with batching)
Remaining features to implement to reach parity with the old renderer:
- [x] textures
- [X] TextBundle
I'd also like to add these features, but they need some design discussion:
- [x] batching
- [ ] separate opaque and transparent phases
- [ ] multiple windows
- [ ] texture atlases
- [ ] (maybe) clipping
Updates the wireframe rendering initialliy implemented in https://github.com/bevyengine/bevy/pull/562 to the new renderer.
It lives in `bevy_pbr2` instead of `bevy_render2` because that way it can reuse the `MeshPipeline`.
# Objective
Make it easier to build and use an asset path with `format!()`. This can be useful for accessing assets in a loop.
Enabled by this PR:
```rust
let monkey_handle = asset_server.get_handle(&format!("models/monkey/Monkey.gltf#Mesh0/Primitive0"));
let monkey_handle = asset_server.get_handle(format!("models/monkey/Monkey.gltf#Mesh0/Primitive0"));
```
Before this PR:
```rust
let monkey_handle = asset_server.get_handle(format!("models/monkey/Monkey.gltf#Mesh0/Primitive0").as_str());
```
It's just a tiny improvement in ergonomics, but i ran into it and was wondering why the function does not accept a `String` and Bevy is all about simplicity/ergonomics, right? 😄😉
## Solution
Implement `Into<HandleId>` for `String` and `&String`.
# Objective
- Rendering before MainPass should be possible, so clearing needs to happen in an earlier pass.
- Fixes#3190.
## Solution
- I added a "Clear" SubGraph, a "ClearPassNode" Node, that clears the color and depth attachments of all views and a "ClearNodeDriver" Node, that schedules the "ClearPassNode" before MainPass.
- Make sure that the 2d and 3d draw passes do not clear their attachments anymore.
### Notes
- It works in the way, that with the current pipeline examples nothing should have changed in their behaviour
- I would like to add an example that adds a pass inbetween ClearPass and MainPass, but I do not understand enough about the new render architecture to do that yet
- Clears all attachment for all views: I do not know enough about rendering in general to say, whether there is a use case for not clearing
- Does not solve #3043 as we still need Cameras/ViewTargets to clear.
# Objective
The new renderer does not support any options yet for wgpu. These are needed for example for rendering wireframes (see #3193).
## Solution
I've ported WgpuOptions to bevy_render2.
The defaults match the defaults that were used before this PR (meaning, some specific options when target_arch = wasm32).
Additionally, I removed `Auto` from WgpuBackends and added `Primary`. The default will use primary or GL based on the target_arch.
# Objective
- there are a few new versions for `ron`, `winit`, `ndk`, `raw-window-handle`
- `cargo-deny` is failing due to new security issues / duplicated dependencies
## Solution
- Update our dependencies
- Note all new security issues, with which of Bevy direct dependency it comes from
- Update duplicate crate list, with which of Bevy direct dependency it comes from
`notify` is not updated here as it's in #2993
# Objective
Implement clustered-forward rendering.
## Solution
~~FIXME - in the interest of keeping the merge train moving, I'm submitting this PR now before the description is ready. I want to add in some comments into the code with references for the various bits and pieces and I want to describe some of the key decisions I made here. I'll do that as soon as I can.~~ Anyone reviewing is welcome to add review comments where you want to know more about how something or other works.
* The summary of the technique is that the view frustum is divided into a grid of sub-volumes called clusters, point lights are tested against each of the clusters to see if they would affect that volume within the scene and if so, added to a list of lights affecting that cluster. Then when shading a fragment which is a point on the surface of a mesh within the scene, the point is mapped to a cluster and only the lights affecting that clusters are used in lighting calculations. This brings huge performance and scalability benefits as most of the time lights are placed so that there are not that many that overlap each other in terms of their sphere of influence, but there may be many distinct point lights visible in the scene. Doing all the lighting calculations for all visible lights in the scene for every pixel on the screen quickly becomes a performance limitation. Clustered forward rendering allows us to make an approximate list of lights that affect each pixel, indeed each surface in the scene (as it works along the view z axis too, unlike tiled/forward+).
* WebGL2 is a platform we want to support and it does not support storage buffers. Uniform buffer bindings are limited to a maximum of 16384 bytes per binding. I used bit shifting and masking to pack the cluster light lists and various indices into a uniform buffer and the 16kB limit is very likely the first bottleneck in scaling the number of lights in a scene at the moment if the lights can affect many clusters due to their range or proximity to the camera (there are a lot of clusters close to the camera, which is an area for improvement). We could store the information in textures instead of uniform buffers to remove this bottleneck though I don’t know if there are performance implications to reading from textures instead if uniform buffers.
* Because of the uniform buffer binding size limitations we can support a maximum of 256 lights with the current size of the PointLight struct
* The z-slicing method (i.e. the mapping from view space z to a depth slice which defines the near and far planes of a cluster) is using the Doom 2016 method. I need to add comments with references to this. It’s an exponential function that simplifies well for the purposes of optimising the fragment shader. xy grid divisions are regular in screen space.
* Some optimisation work was done on the allocation of lights to clusters, which involves intersection tests, and for this number of clusters and lights the system has insignificant cost using a fairly naïve algorithm. I think for more lights / finer-grained clusters we could use a BVH, but at some point it would be just much better to use compute shaders and storage buffers.
* Something else to note is that it is absolutely infeasible to use plain cube map point light shadow mapping for many lights. It does not scale in terms of performance nor memory usage. There are some interesting methods I saw discussed in reference material that I will add a link to which render and update shadow maps piece-wise, but they also need compute shaders to work well. Basically for now you need to sacrifice point light shadows for all but a handful of point lights if you don’t want to kill performance. I set the limit to 10 but that’s just what we had from before where 10 was the maximum number of point lights before this PR.
* I added a couple of debug visualisations behind a shader def that were useful for seeing performance impact of light distribution - I should make the debug mode configurable without modifying the shader code. One mode shows the number of lights affecting each cluster by tinting toward red for few lights or green for many lights (maxes out at 16, but not sure that’s a reasonable max). The other shows which cluster the surface at a fragment belongs to by tinting it with a randomish colour. This can help to understand deeper performance issues due to screen space tiles spanning multiple clusters in depth with divergent shader execution times.
Also, there are more things that could be done as improvements, and I will document those somewhere (I'm not sure where will be the best place... in a todo alongside the code, a GitHub issue, somewhere else?) but I think it works well enough and brings significant performance and scalability benefits that it's worth integrating already now and then iterating on.
* Calculate the light’s effective range based on its intensity and physical falloff and either just use this, or take the minimum of the user-supplied range and this. This would avoid unnecessary lighting calculations for clusters that cannot be affected. This would need to take into account HDR tone mapping as in my not-fully-understanding-the-details understanding, the threshold is relative to how bright the scene is.
* Improve the z-slicing to use a larger first slice.
* More gracefully handle the cluster light list uniform buffer binding size limitations by prioritising which lights are included (some heuristic for most significant like closest to the camera, brightest, affecting the most pixels, …)
* Switch to using a texture instead of uniform buffer
* Figure out the / a better story for shadows
I will also probably add an example that demonstrates some of the issues:
* What situations exhaust the space available in the uniform buffers
* Light range too large making lights affect many clusters and so exhausting the space for the lists of lights that affect clusters
* Light range set to be too small producing visible artifacts where clusters the light would physically affect are not affected by the light
* Perhaps some performance issues
* How many lights can be closely packed or affect large portions of the view before performance drops?
Fills in some gaps we had in our Bevy ECS tracing spans:
* Exclusive systems
* System Commands (for `apply_buffers = true` cases)
* System archetype updates
* Parallel system execution prep
Applogies, had to recreate this pr because of branching issue.
Old PR: https://github.com/bevyengine/bevy/pull/3033
# Objective
Fixes#3032
Allowing a user to create a transparent window
## Solution
I've allowed the transparent bool to be passed to the winit window builder
# Objective
Fixes#3181
## Solution
Refactored `contributors.rs` example:
- Renamed unclear variables
- Split setup system into two separate systems
Co-authored-by: CrazyRoka <rokarostuk@gmail.com>
# Objective
- Checks for NaN in computed NDC space coordinates, fixing unexpected NaN in a fallible (`Option<T>`) function.
## Solution
- Adds a NaN check, in addition to the existing NDC bounds checks.
- This is a helper function, and should have no performance impact to the engine itself.
- This will help prevent hard-to-trace NaN propagation in user code, by returning `None` instead of `Some(NaN)`.
Depends on https://github.com/bevyengine/bevy/pull/3269 for CI error fix.
# Objective
I'm exposing these command encoders so bevy user's can create their own command encoders. This is useful when you want to copy a texture to a texture or create a compute pass manually for example.
Note: I formatted this file which might of changed the order of some exports.
## Solution
Just re-export `CommandEncoder` and `CommandEncoderDescriptor`.
# Objective
- Fix#3188
- Allow creating a `PipelinedSpriteBundle` without an image, just a plain color
```rust
PipelinedSpriteBundle {
sprite: Sprite {
color: Color::rgba(0.8, 0.0, 0.0, 0.3),
custom_size: Some(Vec2::new(500.0, 500.0)),
..Default::default()
},
..Default::default()
}
```
## Solution
- The default impl for `Image` was creating a one pixel image with all values at `1`. I changed it to `255` as picking `1` for it doesn't really make sense, it should be either `0` or `255`
- I created a static handle and added the default image to the assets with this handle
- I changed the default impl for `PipelinedSpriteBundle` to use this handle
# Objective
Fixes recent pipeline errors:
```
error: use of deprecated associated function `std::array::IntoIter::<T, N>::new`: use `IntoIterator::into_iter` instead
--> crates/bevy_render/src/mesh/mesh.rs:467:54
|
467 | .flat_map(|normal| std::array::IntoIter::new([normal, normal, normal]))
| ^^^
|
= note: `-D deprecated` implied by `-D warnings`
Compiling bevy_render2 v0.5.0 (/home/runner/work/bevy/bevy/pipelined/bevy_render2)
error: use of deprecated associated function `std::array::IntoIter::<T, N>::new`: use `IntoIterator::into_iter` instead
--> pipelined/bevy_render2/src/mesh/mesh/mod.rs:287:54
|
287 | .flat_map(|normal| std::array::IntoIter::new([normal, normal, normal]))
| ^^^
|
= note: `-D deprecated` implied by `-D warnings`
error: could not compile `bevy_render` due to previous error
```
## Solution
- Replaced `IntoIter::new` with `IntoIterator::into_iter`
## Suggestions
For me it looks like two equivalent `Mesh` structs with the same methods. Should we refactor it? Or, they will be different in the near future?
Co-authored-by: CrazyRoka <rokarostuk@gmail.com>
# Objective
- New clippy lints with rust 1.57 are failing
## Solution
- Fixed clippy lints following suggestions
- I ignored clippy in old renderer because there was many and it will be removed soon
# Objective
- iOS CI has linker issues https://github.com/bevyengine/bevy/runs/4388921574?check_suite_focus=true
## Solution
- Building for iOS actually requires ~~both iOS SDK for target and~~ macOS SDK for build scripts. ~~I added them both when needed~~ I replaced the iOS SDK with the maOS. This was not an issue on m1 as they are compatible enough to make the build pass.
- This completely confused `shader-sys` which fails to build in this configuration. Luckily as the example now uses the new renderer, I was able to remove the old renderer and depend no more on this lib.
This is confirmed to work:
- on intel mac with simulator
- on m1 mac with simulator
- on m1 mac with real iphone
# Objective
- Add support for `#else` for shader defs
## Solution
- When entering a scope with `#ifdef` or `#ifndef`, if the parent scope is truthy, and the shader definition is also truthy, then the a new scope is pushed onto the scope stack that is also truthy, else falsy. When encountering a subsequent else clause within a scope, if the parent is truthy and the current scope is truthy, then it should become falsy. If the parent scope is truthy and the current scope is falsy then it should become truthy. If the parent scope is falsy, then the current scope should remain falsy as the parent scope takes precedent.
- I added a simple test for an else case.
A sample implementation of how to have `iter()` work on mutable queries without breaking aliasing rules.
# Objective
- Fixes#753
## Solution
- Added a ReadOnlyFetch to WorldQuery that is the `&T` version of `&mut T` that is used to specify the return type for read only operations like `iter()`.
- ~~As the comment suggests specifying the bound doesn't work due to restrictions on defining recursive implementations (like `Or`). However bounds on the functions are fine~~ Never mind I misread how `Or` was constructed, bounds now exist.
- Note that the only mutable one has a new `Fetch` for readonly as the `State` has to be the same for any of this to work
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
# Objective
- Changing the underlying image would not update a sprite
## Solution
- 'Detect' if the underlying image changes to update the sprite
Currently, we don't support change detection on `RenderAssets`, so we have to manually check it.
This method at least maintains the bind groups when the image isn't changing. They were cached, so I assume that's important.
This gives us correct behaviour here.
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
# Objective
- With #3109 I broke iOS CI: https://github.com/bevyengine/bevy/runs/4374891646?check_suite_focus=true
## Solution
- Fix indentation in makefile
- Adds scheme that is now needed
With this, `make install` works on my m1 Mac but still fails on my intel Mac unless I run something like `make install; cargo build --target x86_64-apple-ios; make install; cargo build --target x86_64-apple-ios; make install`. It seems something is off when executing cargo through `xcodebuild` on my setup but not sure what. So this PR will probably not fix iOS CI 😞
This pull request aims to solve the issue of a lack of documentation in the enum WindowMode
# Objective
- Fixes#3136
## Solution
- Added a few lines of documentation that should document what the enum does better
# Objective
- Remove `cargo-lipo` as [it's deprecated](https://github.com/TimNN/cargo-lipo#maintenance-status) and doesn't work on new Apple processors
- Fix CI that will fail as soon as GitHub update the worker used by Bevy to macOS 11
## Solution
- Replace `cargo-lipo` with building with the correct target
- Setup the correct path to libraries by using `xcrun --show-sdk-path`
- Also try and fix path to cmake in case it's not found but available through homebrew
# Problem
Let's say I am writting a simple bevy plugin, and I want to depend on `bevy_ecs` crate instead of depending on the full `bevy`.
So I write the following:
*Cargo.toml*:
```toml
[dependencies]
bevy_ecs = { git = "https://github.com/bevyengine/bevy.git", rev = "94db0176fecfac7e7e9763f2dc7458a54c105886" }
```
*lib.rs*:
```rust
use bevy_ecs::prelude::*;
#[derive(Debug, Default, Component)
pub struct MyFancyComponent;
```
So far, so good. Everything works. But let's say I want to write some examples for using my plugin. And for theses I'd like to use the `bevy` crate, so that I can write complete examples (rendering stuff, etc.) that are simple and look like what the consumer of my plugin will do (`use bevy::prelude::*` and `DefaultPlugins`)
So I amend my *Cargo.toml*:
```toml
[dependencies]
bevy_ecs = { git = "https://github.com/bevyengine/bevy.git", rev = "94db0176fecfac7e7e9763f2dc7458a54c105886" }
[dev-dependencies]
bevy = { git = "https://github.com/bevyengine/bevy.git", rev = "94db0176fecfac7e7e9763f2dc7458a54c105886", default-features = false }
```
And that leads to a complilation error
```
error[E0433]: failed to resolve: use of undeclared crate or module `bevy`
```
Basically, because `bevy` is in the `dev-dependencies`, the macro (of the production code) decides to use the `bevy::ecs` path instead of `bevy_ecs`. But `bevy` is not available there.
## Solution
This PR fixes the problem. I amend the macro utility responsible of finding the path of a module.
If we try to find a path, we first test if this correspond to a crate that the user directly depend on. (Like, if we search for `bevy_ecs`, we first check if there is a `bevy_ecs` dependency). If yes, we can depend on that directly. Otherwise, we proceed with the existing logic (testing `bevy` and `bevy_internal`)
# Objective
Fixes#3160
Unless I'm mistaken, the problem was caused by a simple typo
## Solution
- Fix the typo
Reference documentation: https://git-scm.com/docs/gitattributes#_eol
## Question
Are there other file-types that should be included here?
# Objective
- Shadow maps should only be sampled if the mesh is a shadow receiver AND shadow mapping is enabled for the light
## Solution
- Fix the logic in the shader
See #3165 and #3175
# Objective
- @superdump was having trouble with this loop in the GLTF loader.
## Solution
- Make it probably linear.
- Measured times:
- Old: 40s, new: 200ms
I think there's still room for improvement. For example, I think making the nodes be in `Arc`s could be a significant gain, since currently there's duplication all the way down the tree.
Co-authored-by: Carter Anderson <mcanders1@gmail.com>
# Objective
- Use `std140_size_static()` everywhere instead of manual sizes as the crevice rewrite appears to have fixed all the problems as it claimed to do.
I've tested `3d_scene_pipelined`, `bevymark_pipelined`, and `load_gltf_pipelined` and all three look fine.
