Nanite team also made a custom rasteriser- basically, gpu rasteriser suffers on small triangles, so they wrote a software rasteriser that works on triangles that are smaller than 25*25 pixels or so. It whould be such a cherry on top
TIL! Thanks for the info. The audio working group channel confirms this is the case (current audio doc includes this comment as well hackmd.io/@solarliner/bevy-audio-v2 for anyone else interested)
@@chrisbiscardi Hi! That is correct, we're switching to making a custom audio engine for various reasons (as listed in the above design docs you liked), hopefully it is the correct choice and means easier time implementing new audio features, better user-facing API, and more stable especially in WASM.
Why? What we’re seeing here is per-object motion blur, not the annoying type of motion blur that is affected by camera movement. I think it actually looks really good.
On the one hand it's so cool to see this engine getting all the fancy 3d features and fireworks like god rays, volumetric fog but on the other hand you have 2d which again was abandoned and doesn't even have the most basic 2d lighting, not to mention some normal maps or something :/
fwiw, 2d isn't "abandoned" and there are people thinking about and working on it, including some interesting 2d-specific transform and such features. New feature development is often driven by contributor interest and time, and 3d features tend to have more people interested in upstreaming those contributions at the moment (also pcwalton, which is incredibly productive and responsible for a lot of the "volumetric fog" type features this cycle, is working on 3d rendering features, which means an uptick in 3d compared to other areas) for 2d lighting there are a few different crates that can be used depending on what approach you're looking for. * github.com/jgayfer/bevy_light_2d/ * github.com/443eb9/bevy_incandescent * github.com/goto64/bevy_2d_screen_space_lightmaps * github.com/zaycev/bevy-magic-light-2d (Jarl uses bevy-magic-light-2d -- www.youtube.com/@Jarl-Game-com )
after doing thisweekinbevy for the entire cycle, and helping out some with the release processes and candidate issues it was reasonably easy for me to figure out when the release was actually going to happen and record a video at the right time.
my understanding is that meshlets as a feature are based around preprocessing large assets. So you can have a really large detailed mesh and process it instead of using displacement maps.
For most basic games absolutely, if you don't mind writing them in code instead of UI/graphs. For advanced cases it depends on how much you depend on specific Unity/Unreal functionality. Bevy is very modular and you can add any missing features on top with very little overhead, but depending on which features you're adding it might end up being a lot of work. You can also use large chunks of bevy (ecs, color, ...) without using the whole engine suite if you're rolling your own engine from scratch.
@@chrisbiscardi You have, but when I click on source, the implementation for Distribution is missing. I finally found it, it's generated by a macro in the glam crate, but it's not uniform. They need to remove the implementation before someone accidentally uses it.
ah yeah, I see what you mean. Yes the Bevy implementation defers to glam for the actual implementation. There's some additional information in the PR that implemented it around needing to do some more testing: github.com/bevyengine/bevy/pull/12857 I asked in the bevy math channel in the discord if there was ever followup to check the uniformness. How did you determine it wasn't? Would you be open to posting how in the bevy_math channel in the discord?
There will be no editor until 1.0.0 version. Also, if you need editor then this is not for you. What I like the most about Bevy is that it gives me opportunity to make my own dev tooling that I need with eGUI. and having game + editor panels is nice dx.
@@RootsterAnonIt's the other way around actually. There will be no 1.0 without an editor. This release lays a lot of technical groundwork for the editor and more will follow (most notably a new Scene format and the Bevy Remote Protocol). Several people have started prototyping stuff for the editor. It just takes time for everything to come together.
@@paulkupper194 I don't want them to focus on editor, I want them to progress with features and new scene format as you mentioned etc. Editor you can have in few different flavors if you want like custom one, blender, etc. lets wait for all of the features to be in presentable state then official editor can emerge if really needed. I would like API to be more stable so we can migrate easy from version to version, docs to be up to date with latest api changes and good examples. Editor can wait. :D
@@RootsterAnon 1. You're very selfish, you're thinking about what you want not what Bevy needs, but for Bevy to be taken seriously, it needs an editor that you like it or not. 2. An editor written in Bevy is the perfect tool for dog fooding, just like Godot does.
Nanite team also made a custom rasteriser- basically, gpu rasteriser suffers on small triangles, so they wrote a software rasteriser that works on triangles that are smaller than 25*25 pixels or so. It whould be such a cherry on top
Really nice and quick summary of the release. Great video!
FYI I think the audio working group decided to create a new audio solution to better fit into the ecs paradigm rather than using kira audio
TIL! Thanks for the info. The audio working group channel confirms this is the case (current audio doc includes this comment as well hackmd.io/@solarliner/bevy-audio-v2 for anyone else interested)
@@chrisbiscardi Hi! That is correct, we're switching to making a custom audio engine for various reasons (as listed in the above design docs you liked), hopefully it is the correct choice and means easier time implementing new audio features, better user-facing API, and more stable especially in WASM.
@@SolarLiner looking forward to seeing audio take shape as this progresses!
I love Bevy
Always appreciate these round up summary videos, keep it up and thank you!
Nice, can't wait to disable motion blur.
Why? What we’re seeing here is per-object motion blur, not the annoying type of motion blur that is affected by camera movement. I think it actually looks really good.
@@ThePC007 Lol, motion blur is contentious, no matter how well it's implemented. No harm in having it as an optional feature.
Thanks for the overview! Looks like very interesting stuff just landed.
holy shiet, that was packed. Bevy is getting better and better.
On the one hand it's so cool to see this engine getting all the fancy 3d features and fireworks like god rays, volumetric fog but on the other hand you have 2d which again was abandoned and doesn't even have the most basic 2d lighting, not to mention some normal maps or something :/
fwiw, 2d isn't "abandoned" and there are people thinking about and working on it, including some interesting 2d-specific transform and such features.
New feature development is often driven by contributor interest and time, and 3d features tend to have more people interested in upstreaming those contributions at the moment (also pcwalton, which is incredibly productive and responsible for a lot of the "volumetric fog" type features this cycle, is working on 3d rendering features, which means an uptick in 3d compared to other areas)
for 2d lighting there are a few different crates that can be used depending on what approach you're looking for.
* github.com/jgayfer/bevy_light_2d/
* github.com/443eb9/bevy_incandescent
* github.com/goto64/bevy_2d_screen_space_lightmaps
* github.com/zaycev/bevy-magic-light-2d
(Jarl uses bevy-magic-light-2d -- www.youtube.com/@Jarl-Game-com )
haha wow looks like you got this video out within an hour of the announcement, well done! 🎉
after doing thisweekinbevy for the entire cycle, and helping out some with the release processes and candidate issues it was reasonably easy for me to figure out when the release was actually going to happen and record a video at the right time.
mega cool
With meshlets, will be we able to do heightmap/displacement maps with a ton of detail?
my understanding is that meshlets as a feature are based around preprocessing large assets. So you can have a really large detailed mesh and process it instead of using displacement maps.
Does Bevy have the ability to replace unreal or Unity?
it depends on your use case and requirements
@@chrisbiscardi Intractive Digital Twin apps Developement
For most basic games absolutely, if you don't mind writing them in code instead of UI/graphs.
For advanced cases it depends on how much you depend on specific Unity/Unreal functionality. Bevy is very modular and you can add any missing features on top with very little overhead, but depending on which features you're adding it might end up being a lot of work.
You can also use large chunks of bevy (ecs, color, ...) without using the whole engine suite if you're rolling your own engine from scratch.
There's no random sample for Quat. anyone know a fast uniform one?
@@CjqNslXUcM are you looking for the FromRng implementation on Quat?
docs.rs/bevy/latest/bevy/math/struct.Quat.html#impl-FromRng-for-Quat
@@chrisbiscardi yes, where is the implementation?
I don't understand the question, I've already linked you to it.
@@chrisbiscardi You have, but when I click on source, the implementation for Distribution is missing. I finally found it, it's generated by a macro in the glam crate, but it's not uniform. They need to remove the implementation before someone accidentally uses it.
ah yeah, I see what you mean. Yes the Bevy implementation defers to glam for the actual implementation. There's some additional information in the PR that implemented it around needing to do some more testing: github.com/bevyengine/bevy/pull/12857
I asked in the bevy math channel in the discord if there was ever followup to check the uniformness. How did you determine it wasn't? Would you be open to posting how in the bevy_math channel in the discord?
Substates are literally what i wanted ever since i played with Bevy. Absolutely horrible the old system!
Still no Editor?
There will be no editor until 1.0.0 version. Also, if you need editor then this is not for you. What I like the most about Bevy is that it gives me opportunity to make my own dev tooling that I need with eGUI. and having game + editor panels is nice dx.
@@RootsterAnonIt's the other way around actually. There will be no 1.0 without an editor. This release lays a lot of technical groundwork for the editor and more will follow (most notably a new Scene format and the Bevy Remote Protocol). Several people have started prototyping stuff for the editor. It just takes time for everything to come together.
@@paulkupper194 I don't want them to focus on editor, I want them to progress with features and new scene format as you mentioned etc. Editor you can have in few different flavors if you want like custom one, blender, etc. lets wait for all of the features to be in presentable state then official editor can emerge if really needed. I would like API to be more stable so we can migrate easy from version to version, docs to be up to date with latest api changes and good examples. Editor can wait. :D
@@RootsterAnon 1. You're very selfish, you're thinking about what you want not what Bevy needs, but for Bevy to be taken seriously, it needs an editor that you like it or not. 2. An editor written in Bevy is the perfect tool for dog fooding, just like Godot does.
@@francois199 That is just reality of the situation we are in. You can't have editor without stable api. It's that simple.