Re: RFR: 8234920: Add SpotLight to the selection of 3D light types [v14]
On Mon, 7 Jun 2021 13:29:01 GMT, Kevin Rushforth wrote: > In that case, using computeSpotlightFactor in the GLSL shaders and > computeSpotlightFactor3 in the HLSL shaders seems like the way to go. If we > are settled on this, the two you aren't using should be commented out or > #ifdefed out. I think so too. Did you test on Win? Does @arapte want to test these functions too? > The more interesting question is whether the gap between what we have now and > what is possible for simple lights justifies additional shaders. Maybe not > for point lights, but we need to at least consider it when you add > directional lights. The question of adding shaders is a bigger one than just the light types. I plan on adding emissive (or self-illuminating) color, in addition to the existing map, which will probably require creating additional shaders as we do for diffuse and specular colors/maps. Another change will be removing the limit of 3 lights. I have a patch that does it by removing the shader creation per light number. We will need to think how to redistribute the shaders since we can't create one for every light number. Either remove those or, if possible, divide then into 1 light, 2 lights, 3 lights, and 4+ lights. Not sure how to include different numbers of lights in the same shader syntactically. - PR: https://git.openjdk.java.net/jfx/pull/334
Re: RFR: 8234920: Add SpotLight to the selection of 3D light types [v14]
On Sun, 6 Jun 2021 16:19:46 GMT, Nir Lisker wrote: > > So it looks like `computeSpotlightFactor` is the winner. Does that match > > your observations? It might be worth some further tuning. > > Yes, on Ubuntu. On Win, `computeSpotlightFactor3` was the best for me. I > detailed my findings in the topmost comment. Oh, right. I didn't read that carefully. In that case, using `computeSpotlightFactor` in the GLSL shaders and `computeSpotlightFactor3` in the HLSL shaders seems like the way to go. If we are settled on this, the two you aren't using should be commented out or `#ifdef`ed out. > What further tuning are you thinking of? I don't have anything specific in mind. I was just wondering if there might be additional optimizations that one could do the `computeSpotlightFactor` method. Even if so, it could probably be done in a follow-on bug. The more interesting question is whether the gap between what we have now and what is possible for simple lights justifies additional shaders. Maybe not for point lights, but we need to at least consider it when you add directional lights. - PR: https://git.openjdk.java.net/jfx/pull/334
Re: RFR: 8234920: Add SpotLight to the selection of 3D light types [v14]
On Sat, 5 Jun 2021 17:27:13 GMT, Kevin Rushforth wrote: > So it looks like computeSpotlightFactor is the winner. Does that match your > observations? It might be worth some further tuning. Yes, on Ubuntu. On Win, `computeSpotlightFactor3` was the best for me. I detailed my findings in the topmost comment. What further tuning are you thinking of? - PR: https://git.openjdk.java.net/jfx/pull/334
Re: RFR: 8234920: Add SpotLight to the selection of 3D light types [v14]
On Sat, 22 May 2021 02:27:41 GMT, Nir Lisker wrote: > Did you test the different `computeSpotlightFactor` methods in the shader? I > got some 4 fps difference between the best and worst cases. Here are the performance numbers on my Mac (using the discrete `AMD Radeon Pro 5300M` graphics chipset) using the using the LightingSample test program with a mesh of 1000 quads rendered with 3 point lights and no attenuation: | shader compute method | fps | | | | | computeSpotlightFactor | 10.75 | | computeSpotlightFactor2 | 9.40 | | computeSpotlightFactor3 | 7.70 | So it looks like computeSpotlightFactor is the winner. Does that match your observations? It might be worth some further tuning. - PR: https://git.openjdk.java.net/jfx/pull/334
Re: RFR: 8234920: Add SpotLight to the selection of 3D light types [v14]
On Mon, 24 May 2021 05:02:45 GMT, Nir Lisker wrote: >> modules/javafx.graphics/src/main/java/javafx/scene/SpotLight.java line 55: >> >>> 53: * {@code falloff >= 0}; values outside either of these ranges can >>> produce unexpected results. >>> 54: * >>> 55: * >> >> Should we have cone depicted in the diagram ? > > Not sure how to do that in a clear way. If you want to create an image I will > use it. I shall try to create a diagram. - PR: https://git.openjdk.java.net/jfx/pull/334
Re: RFR: 8234920: Add SpotLight to the selection of 3D light types [v14]
On Fri, 16 Apr 2021 13:26:56 GMT, Ambarish Rapte wrote: >> Nir Lisker has updated the pull request incrementally with one additional >> commit since the last revision: >> >> Combined rotation and direction > > modules/javafx.graphics/src/main/java/javafx/scene/SpotLight.java line 55: > >> 53: * {@code falloff >= 0}; values outside either of these ranges can >> produce unexpected results. >> 54: * >> 55: * > > Should we have cone depicted in the diagram ? Not sure how to do that in a clear way. If you want to create an image I will use it. - PR: https://git.openjdk.java.net/jfx/pull/334
Re: RFR: 8234920: Add SpotLight to the selection of 3D light types [v14]
On Fri, 16 Apr 2021 13:33:36 GMT, Ambarish Rapte wrote: >> modules/javafx.graphics/src/main/java/javafx/scene/SpotLight.java line 46: >> >>> 44: * >>> 45: * The light cone is defined by 3 factors: an {@link >>> #innerAngleProperty() inner angle}, an {@link #outerAngleProperty() >>> 46: * outer angle}, and a {@link #falloffProperty() falloff} factor. For a >>> point whose angle to the light is {@code a}, if >> >> -> >> A {@code SpotLight} is a {@code PointLight} that radiates a cone of light in >> a specific direction. >> < p > >> The direction is defined by the {@link #directionProperty() direction} >> property. >> The light cone is defined by 3 properties: an {@link #innerAngleProperty() >> inner angle}, an {@link #outerAngleProperty() outer angle}, and a {@link >> #falloffProperty() falloff} property. > > Also Will it be illustrative to mention these three properties using li tags > with a statement that briefs the use of property. > Something like, > - {@link #innerAngleProperty() inner angle}: Angle of the inner cone where > light intensity is always maximum. > - {@link #outerAngleProperty() outer angle}: Angle of the outer cone where > light intensity attenuates based on fallof. > - {@link #falloffProperty() falloff}: The attenuation factor that controls > the attenuation of light intensity from edge of inner cone to the edge of > outer cone. I used bullet points, but the descriptions of the intensity behaviors are described later anyway in more detail, so I kept these short. - PR: https://git.openjdk.java.net/jfx/pull/334
Re: RFR: 8234920: Add SpotLight to the selection of 3D light types [v14]
On Thu, 15 Apr 2021 02:21:50 GMT, Nir Lisker wrote: >> Added a SpotLight only to the D3D pipeline currently. >> >> ### API discussion points >> >> - [X] Added `SpotLight` as a subclass of `LightBase`. However, it could >> also be a subclass of `PointLight` as it's a point light with direction and >> extra factors. I saw that `scenario.effect.light.SpotLight` extends its >> respective `PointLight`, but it's not a perfect analogy. In the end, I think >> it's a questions of whether `PointLight` will be expanded in a way which >> doesn't not suit `SpotLight`, and I tend to think that the answer is no. >> >> - [X] The inner and outer angles are the "diameter angles" as shown >> [here](https://docs.microsoft.com/en-us/windows/win32/direct3d9/light-typeshttps://docs.microsoft.com/en-us/windows/win32/direct3d9/light-types). >> I, personally, find it more intuitive that these are the "radius angles", >> so half these angles, as used in the spotlight factor formula. Do you think >> I can change this or do you prefer the current definition of the angles? >> >> - [ ] The current implementation uses an ad-hoc direction property (using a >> `Point3D`). It crossed my mind that we could use the rotation transforms of >> the node to control the direction instead, just like we use the >> translation/layout of the node to get the position (there is an internal >> Affine3D transform for lights, not sure why `AmbientLight` needs it). >> Wouldn't that make more sense? When I rotate the light I would expect to see >> a change in direction. >> >> ### Implementation discussion points >> >> - [ ] I've gotten advice from a graphics engineer to treat point lights as >> spot lights with a 360 degrees coverage, which simplifies a few places. We >> can still try to optimize for a point light by looking at the light >> parameters: `falloff = 0` and `outerAngle = 180`. These possible >> optimization exist in `ES2PhongShader.java` and `D3DMeshView.cc`, and in the >> pixel/fragment shaders in the form of 3 different ways to compute the >> spotlight factor (the `computeLightN` methods). We need to check which of >> these give the best results. >> >> ## Performance >> >> Testing 3 point lights and comparing this branch with `master` using a 1000 >> division sphere, 200 meshes, and 5000 meshes. >> Using an AMD RX 470 4GB GPU. >> >> In this branch, there is a possible CPU optimization for checking the light >> type and using precalculated values (in `D3DMeshView.cc` for d3d and >> `ES2PhongShader.java` for opengl). On the GPU, I tried 3 ways of computing >> the spotlight factor contributions (`computeSpotlightFactor`, >> `computeSpotlightFactor2` and `computeSpotlightFactor3`) trying out >> different branching and shortcuts. >> >> ### Results >> The CPU "optimizations" made no difference, which is understandable >> considering it will not be the bottleneck. We can remove these if we want to >> simplify, though maybe if we allow a large number of lights it could make a >> difference (I doubt it). I don't have a strong preference either way. >> >> The sphere 1000 tests always gave max fps (120 on Win and 60 on Ubuntu). >> >> **Win 10** >> Compared with the `master` branch, this patch shows 5-10 fps drop in the >> mesh 200 test and ~5 in the mesh 5000 test. I repeated the tests on several >> occasions and got different results in terms of absolute numbers, but the >> relative performance difference remained more or less the same. Out of the 3 >> `computeSpotlightFactor` methods, `computeSpotlightFactor3`, which has no >> "optimizations", gives slightly better performance. >> >> **Ubuntu 18** >> The mesh 200 test always gave 60 fps because it is locked to this fps, so we >> can't measure the real GPU performance change. >> The mesh 5000 test shows 2-6 fps drop from master, with >> `computeSpotlightFactor` > `computeSpotlightFactor2` > >> `computeSpotlightFactor3` at in terms of performance (~2 fps difference >> each). >> >> **Conclusion**: we can expect a 5 fps drop more or less with 3 point lights. >> `computeSpotlightFactor3` on d3d and `computeSpotlightFactor` on opengl gave >> the best performances. > > Nir Lisker has updated the pull request incrementally with one additional > commit since the last revision: > > Combined rotation and direction No, but I can do that. - PR: https://git.openjdk.java.net/jfx/pull/334
Re: RFR: 8234920: Add SpotLight to the selection of 3D light types [v14]
On Thu, 15 Apr 2021 02:21:50 GMT, Nir Lisker wrote: >> Added a SpotLight only to the D3D pipeline currently. >> >> ### API discussion points >> >> - [X] Added `SpotLight` as a subclass of `LightBase`. However, it could >> also be a subclass of `PointLight` as it's a point light with direction and >> extra factors. I saw that `scenario.effect.light.SpotLight` extends its >> respective `PointLight`, but it's not a perfect analogy. In the end, I think >> it's a questions of whether `PointLight` will be expanded in a way which >> doesn't not suit `SpotLight`, and I tend to think that the answer is no. >> >> - [X] The inner and outer angles are the "diameter angles" as shown >> [here](https://docs.microsoft.com/en-us/windows/win32/direct3d9/light-typeshttps://docs.microsoft.com/en-us/windows/win32/direct3d9/light-types). >> I, personally, find it more intuitive that these are the "radius angles", >> so half these angles, as used in the spotlight factor formula. Do you think >> I can change this or do you prefer the current definition of the angles? >> >> - [ ] The current implementation uses an ad-hoc direction property (using a >> `Point3D`). It crossed my mind that we could use the rotation transforms of >> the node to control the direction instead, just like we use the >> translation/layout of the node to get the position (there is an internal >> Affine3D transform for lights, not sure why `AmbientLight` needs it). >> Wouldn't that make more sense? When I rotate the light I would expect to see >> a change in direction. >> >> ### Implementation discussion points >> >> - [ ] I've gotten advice from a graphics engineer to treat point lights as >> spot lights with a 360 degrees coverage, which simplifies a few places. We >> can still try to optimize for a point light by looking at the light >> parameters: `falloff = 0` and `outerAngle = 180`. These possible >> optimization exist in `ES2PhongShader.java` and `D3DMeshView.cc`, and in the >> pixel/fragment shaders in the form of 3 different ways to compute the >> spotlight factor (the `computeLightN` methods). We need to check which of >> these give the best results. >> >> ## Performance >> >> Testing 3 point lights and comparing this branch with `master` using a 1000 >> division sphere, 200 meshes, and 5000 meshes. >> Using an AMD RX 470 4GB GPU. >> >> In this branch, there is a possible CPU optimization for checking the light >> type and using precalculated values (in `D3DMeshView.cc` for d3d and >> `ES2PhongShader.java` for opengl). On the GPU, I tried 3 ways of computing >> the spotlight factor contributions (`computeSpotlightFactor`, >> `computeSpotlightFactor2` and `computeSpotlightFactor3`) trying out >> different branching and shortcuts. >> >> ### Results >> The CPU "optimizations" made no difference, which is understandable >> considering it will not be the bottleneck. We can remove these if we want to >> simplify, though maybe if we allow a large number of lights it could make a >> difference (I doubt it). I don't have a strong preference either way. >> >> The sphere 1000 tests always gave max fps (120 on Win and 60 on Ubuntu). >> >> **Win 10** >> Compared with the `master` branch, this patch shows 5-10 fps drop in the >> mesh 200 test and ~5 in the mesh 5000 test. I repeated the tests on several >> occasions and got different results in terms of absolute numbers, but the >> relative performance difference remained more or less the same. Out of the 3 >> `computeSpotlightFactor` methods, `computeSpotlightFactor3`, which has no >> "optimizations", gives slightly better performance. >> >> **Ubuntu 18** >> The mesh 200 test always gave 60 fps because it is locked to this fps, so we >> can't measure the real GPU performance change. >> The mesh 5000 test shows 2-6 fps drop from master, with >> `computeSpotlightFactor` > `computeSpotlightFactor2` > >> `computeSpotlightFactor3` at in terms of performance (~2 fps difference >> each). >> >> **Conclusion**: we can expect a 5 fps drop more or less with 3 point lights. >> `computeSpotlightFactor3` on d3d and `computeSpotlightFactor` on opengl gave >> the best performances. > > Nir Lisker has updated the pull request incrementally with one additional > commit since the last revision: > > Combined rotation and direction Did you test the different `computeSpotlightFactor` methods in the shader? I got some 4 fps difference between the best and worst cases. - PR: https://git.openjdk.java.net/jfx/pull/334
Re: RFR: 8234920: Add SpotLight to the selection of 3D light types [v14]
On Thu, 15 Apr 2021 02:21:50 GMT, Nir Lisker wrote: >> Added a SpotLight only to the D3D pipeline currently. >> >> ### API discussion points >> >> - [X] Added `SpotLight` as a subclass of `LightBase`. However, it could >> also be a subclass of `PointLight` as it's a point light with direction and >> extra factors. I saw that `scenario.effect.light.SpotLight` extends its >> respective `PointLight`, but it's not a perfect analogy. In the end, I think >> it's a questions of whether `PointLight` will be expanded in a way which >> doesn't not suit `SpotLight`, and I tend to think that the answer is no. >> >> - [X] The inner and outer angles are the "diameter angles" as shown >> [here](https://docs.microsoft.com/en-us/windows/win32/direct3d9/light-typeshttps://docs.microsoft.com/en-us/windows/win32/direct3d9/light-types). >> I, personally, find it more intuitive that these are the "radius angles", >> so half these angles, as used in the spotlight factor formula. Do you think >> I can change this or do you prefer the current definition of the angles? >> >> - [ ] The current implementation uses an ad-hoc direction property (using a >> `Point3D`). It crossed my mind that we could use the rotation transforms of >> the node to control the direction instead, just like we use the >> translation/layout of the node to get the position (there is an internal >> Affine3D transform for lights, not sure why `AmbientLight` needs it). >> Wouldn't that make more sense? When I rotate the light I would expect to see >> a change in direction. >> >> ### Implementation discussion points >> >> - [ ] I've gotten advice from a graphics engineer to treat point lights as >> spot lights with a 360 degrees coverage, which simplifies a few places. We >> can still try to optimize for a point light by looking at the light >> parameters: `falloff = 0` and `outerAngle = 180`. These possible >> optimization exist in `ES2PhongShader.java` and `D3DMeshView.cc`, and in the >> pixel/fragment shaders in the form of 3 different ways to compute the >> spotlight factor (the `computeLightN` methods). We need to check which of >> these give the best results. >> >> ## Performance >> >> Testing 3 point lights and comparing this branch with `master` using a 1000 >> division sphere, 200 meshes, and 5000 meshes. >> Using an AMD RX 470 4GB GPU. >> >> In this branch, there is a possible CPU optimization for checking the light >> type and using precalculated values (in `D3DMeshView.cc` for d3d and >> `ES2PhongShader.java` for opengl). On the GPU, I tried 3 ways of computing >> the spotlight factor contributions (`computeSpotlightFactor`, >> `computeSpotlightFactor2` and `computeSpotlightFactor3`) trying out >> different branching and shortcuts. >> >> ### Results >> The CPU "optimizations" made no difference, which is understandable >> considering it will not be the bottleneck. We can remove these if we want to >> simplify, though maybe if we allow a large number of lights it could make a >> difference (I doubt it). I don't have a strong preference either way. >> >> The sphere 1000 tests always gave max fps (120 on Win and 60 on Ubuntu). >> >> **Win 10** >> Compared with the `master` branch, this patch shows 5-10 fps drop in the >> mesh 200 test and ~5 in the mesh 5000 test. I repeated the tests on several >> occasions and got different results in terms of absolute numbers, but the >> relative performance difference remained more or less the same. Out of the 3 >> `computeSpotlightFactor` methods, `computeSpotlightFactor3`, which has no >> "optimizations", gives slightly better performance. >> >> **Ubuntu 18** >> The mesh 200 test always gave 60 fps because it is locked to this fps, so we >> can't measure the real GPU performance change. >> The mesh 5000 test shows 2-6 fps drop from master, with >> `computeSpotlightFactor` > `computeSpotlightFactor2` > >> `computeSpotlightFactor3` at in terms of performance (~2 fps difference >> each). >> >> **Conclusion**: we can expect a 5 fps drop more or less with 3 point lights. >> `computeSpotlightFactor3` on d3d and `computeSpotlightFactor` on opengl gave >> the best performances. > > Nir Lisker has updated the pull request incrementally with one additional > commit since the last revision: > > Combined rotation and direction Here is the performance of mesh on my macOS system with a discrete graphics chip using the LightingSample test program with a mesh of 1000 quads rendered with 3 point lights: | Run | fps | | - | | | Baseline (FX 15) | 25.8 | | After Attenuation added (FX 16) | 15.4 | | SpotLight support (this PR) | 10.6 | This was the same program in all three cases, so no SpotLight and no attenuation. This is the same "worst case" stress test that we used when evaluating attenuation. Hopefully we can optimize and get some of the performance back with optimization. In any case, I think this adds additional motivation for a future Directi
Re: RFR: 8234920: Add SpotLight to the selection of 3D light types [v14]
On Fri, 14 May 2021 22:29:25 GMT, Kevin Rushforth wrote: > You may be right about the scale problem being preexisting. I'll double check > when I test on Mac next week (both Retina and external screen). I can confirm that the scaling problem on macOS Retina a preexisting bug (not related to either this or the earlier addition of attenuation). So I'll file a separate issue for this. - PR: https://git.openjdk.java.net/jfx/pull/334
Re: RFR: 8234920: Add SpotLight to the selection of 3D light types [v14]
On Thu, 15 Apr 2021 02:21:50 GMT, Nir Lisker wrote: >> Added a SpotLight only to the D3D pipeline currently. >> >> ### API discussion points >> >> - [X] Added `SpotLight` as a subclass of `LightBase`. However, it could >> also be a subclass of `PointLight` as it's a point light with direction and >> extra factors. I saw that `scenario.effect.light.SpotLight` extends its >> respective `PointLight`, but it's not a perfect analogy. In the end, I think >> it's a questions of whether `PointLight` will be expanded in a way which >> doesn't not suit `SpotLight`, and I tend to think that the answer is no. >> >> - [X] The inner and outer angles are the "diameter angles" as shown >> [here](https://docs.microsoft.com/en-us/windows/win32/direct3d9/light-typeshttps://docs.microsoft.com/en-us/windows/win32/direct3d9/light-types). >> I, personally, find it more intuitive that these are the "radius angles", >> so half these angles, as used in the spotlight factor formula. Do you think >> I can change this or do you prefer the current definition of the angles? >> >> - [ ] The current implementation uses an ad-hoc direction property (using a >> `Point3D`). It crossed my mind that we could use the rotation transforms of >> the node to control the direction instead, just like we use the >> translation/layout of the node to get the position (there is an internal >> Affine3D transform for lights, not sure why `AmbientLight` needs it). >> Wouldn't that make more sense? When I rotate the light I would expect to see >> a change in direction. >> >> ### Implementation discussion points >> >> - [ ] I've gotten advice from a graphics engineer to treat point lights as >> spot lights with a 360 degrees coverage, which simplifies a few places. We >> can still try to optimize for a point light by looking at the light >> parameters: `falloff = 0` and `outerAngle = 180`. These possible >> optimization exist in `ES2PhongShader.java` and `D3DMeshView.cc`, and in the >> pixel/fragment shaders in the form of 3 different ways to compute the >> spotlight factor (the `computeLightN` methods). We need to check which of >> these give the best results. >> >> ## Performance >> >> Testing 3 point lights and comparing this branch with `master` using a 1000 >> division sphere, 200 meshes, and 5000 meshes. >> Using an AMD RX 470 4GB GPU. >> >> In this branch, there is a possible CPU optimization for checking the light >> type and using precalculated values (in `D3DMeshView.cc` for d3d and >> `ES2PhongShader.java` for opengl). On the GPU, I tried 3 ways of computing >> the spotlight factor contributions (`computeSpotlightFactor`, >> `computeSpotlightFactor2` and `computeSpotlightFactor3`) trying out >> different branching and shortcuts. >> >> ### Results >> The CPU "optimizations" made no difference, which is understandable >> considering it will not be the bottleneck. We can remove these if we want to >> simplify, though maybe if we allow a large number of lights it could make a >> difference (I doubt it). I don't have a strong preference either way. >> >> The sphere 1000 tests always gave max fps (120 on Win and 60 on Ubuntu). >> >> **Win 10** >> Compared with the `master` branch, this patch shows 5-10 fps drop in the >> mesh 200 test and ~5 in the mesh 5000 test. I repeated the tests on several >> occasions and got different results in terms of absolute numbers, but the >> relative performance difference remained more or less the same. Out of the 3 >> `computeSpotlightFactor` methods, `computeSpotlightFactor3`, which has no >> "optimizations", gives slightly better performance. >> >> **Ubuntu 18** >> The mesh 200 test always gave 60 fps because it is locked to this fps, so we >> can't measure the real GPU performance change. >> The mesh 5000 test shows 2-6 fps drop from master, with >> `computeSpotlightFactor` > `computeSpotlightFactor2` > >> `computeSpotlightFactor3` at in terms of performance (~2 fps difference >> each). >> >> **Conclusion**: we can expect a 5 fps drop more or less with 3 point lights. >> `computeSpotlightFactor3` on d3d and `computeSpotlightFactor` on opengl gave >> the best performances. > > Nir Lisker has updated the pull request incrementally with one additional > commit since the last revision: > > Combined rotation and direction I think the API as currently defined is what we want. Having an explicit direction vector, and then transforming that by the transform of the light is quite flexible in a natural and expected way. I took the existing [`LightMotion`](https://github.com/openjdk/jfx/blob/master/apps/toys/FX8-3DFeatures/src/fx83dfeatures/LightMotion.java) example program, and created a new `SpotLightMotion` program (attached). It's easy to programmatically point the light to the object you wish to illuminate (the sphere or cylinder, in this example). And once you do, rotation works naturally. Here is the code that computes the spot light direction:
Re: RFR: 8234920: Add SpotLight to the selection of 3D light types [v14]
On Thu, 15 Apr 2021 02:21:50 GMT, Nir Lisker wrote: >> Added a SpotLight only to the D3D pipeline currently. >> >> ### API discussion points >> >> - [X] Added `SpotLight` as a subclass of `LightBase`. However, it could >> also be a subclass of `PointLight` as it's a point light with direction and >> extra factors. I saw that `scenario.effect.light.SpotLight` extends its >> respective `PointLight`, but it's not a perfect analogy. In the end, I think >> it's a questions of whether `PointLight` will be expanded in a way which >> doesn't not suit `SpotLight`, and I tend to think that the answer is no. >> >> - [X] The inner and outer angles are the "diameter angles" as shown >> [here](https://docs.microsoft.com/en-us/windows/win32/direct3d9/light-typeshttps://docs.microsoft.com/en-us/windows/win32/direct3d9/light-types). >> I, personally, find it more intuitive that these are the "radius angles", >> so half these angles, as used in the spotlight factor formula. Do you think >> I can change this or do you prefer the current definition of the angles? >> >> - [ ] The current implementation uses an ad-hoc direction property (using a >> `Point3D`). It crossed my mind that we could use the rotation transforms of >> the node to control the direction instead, just like we use the >> translation/layout of the node to get the position (there is an internal >> Affine3D transform for lights, not sure why `AmbientLight` needs it). >> Wouldn't that make more sense? When I rotate the light I would expect to see >> a change in direction. >> >> ### Implementation discussion points >> >> - [ ] I've gotten advice from a graphics engineer to treat point lights as >> spot lights with a 360 degrees coverage, which simplifies a few places. We >> can still try to optimize for a point light by looking at the light >> parameters: `falloff = 0` and `outerAngle = 180`. These possible >> optimization exist in `ES2PhongShader.java` and `D3DMeshView.cc`, and in the >> pixel/fragment shaders in the form of 3 different ways to compute the >> spotlight factor (the `computeLightN` methods). We need to check which of >> these give the best results. >> >> ## Performance >> >> Testing 3 point lights and comparing this branch with `master` using a 1000 >> division sphere, 200 meshes, and 5000 meshes. >> Using an AMD RX 470 4GB GPU. >> >> In this branch, there is a possible CPU optimization for checking the light >> type and using precalculated values (in `D3DMeshView.cc` for d3d and >> `ES2PhongShader.java` for opengl). On the GPU, I tried 3 ways of computing >> the spotlight factor contributions (`computeSpotlightFactor`, >> `computeSpotlightFactor2` and `computeSpotlightFactor3`) trying out >> different branching and shortcuts. >> >> ### Results >> The CPU "optimizations" made no difference, which is understandable >> considering it will not be the bottleneck. We can remove these if we want to >> simplify, though maybe if we allow a large number of lights it could make a >> difference (I doubt it). I don't have a strong preference either way. >> >> The sphere 1000 tests always gave max fps (120 on Win and 60 on Ubuntu). >> >> **Win 10** >> Compared with the `master` branch, this patch shows 5-10 fps drop in the >> mesh 200 test and ~5 in the mesh 5000 test. I repeated the tests on several >> occasions and got different results in terms of absolute numbers, but the >> relative performance difference remained more or less the same. Out of the 3 >> `computeSpotlightFactor` methods, `computeSpotlightFactor3`, which has no >> "optimizations", gives slightly better performance. >> >> **Ubuntu 18** >> The mesh 200 test always gave 60 fps because it is locked to this fps, so we >> can't measure the real GPU performance change. >> The mesh 5000 test shows 2-6 fps drop from master, with >> `computeSpotlightFactor` > `computeSpotlightFactor2` > >> `computeSpotlightFactor3` at in terms of performance (~2 fps difference >> each). >> >> **Conclusion**: we can expect a 5 fps drop more or less with 3 point lights. >> `computeSpotlightFactor3` on d3d and `computeSpotlightFactor` on opengl gave >> the best performances. > > Nir Lisker has updated the pull request incrementally with one additional > commit since the last revision: > > Combined rotation and direction You may be right about the scale problem being preexisting. I'll double check when I test on Mac next week (both Retina and external screen). - PR: https://git.openjdk.java.net/jfx/pull/334
Re: RFR: 8234920: Add SpotLight to the selection of 3D light types [v14]
On Fri, 14 May 2021 21:28:15 GMT, Kevin Rushforth wrote: > There is also the outstanding issue with a retina display (scale=2) on macOS. Yes, it has to be looked at, but I think it is unrelated to this particular patch. If this is the case, we can file a different issue and continue with this one. - PR: https://git.openjdk.java.net/jfx/pull/334
Re: RFR: 8234920: Add SpotLight to the selection of 3D light types [v14]
On Thu, 15 Apr 2021 02:21:50 GMT, Nir Lisker wrote: >> Added a SpotLight only to the D3D pipeline currently. >> >> ### API discussion points >> >> - [X] Added `SpotLight` as a subclass of `LightBase`. However, it could >> also be a subclass of `PointLight` as it's a point light with direction and >> extra factors. I saw that `scenario.effect.light.SpotLight` extends its >> respective `PointLight`, but it's not a perfect analogy. In the end, I think >> it's a questions of whether `PointLight` will be expanded in a way which >> doesn't not suit `SpotLight`, and I tend to think that the answer is no. >> >> - [X] The inner and outer angles are the "diameter angles" as shown >> [here](https://docs.microsoft.com/en-us/windows/win32/direct3d9/light-typeshttps://docs.microsoft.com/en-us/windows/win32/direct3d9/light-types). >> I, personally, find it more intuitive that these are the "radius angles", >> so half these angles, as used in the spotlight factor formula. Do you think >> I can change this or do you prefer the current definition of the angles? >> >> - [ ] The current implementation uses an ad-hoc direction property (using a >> `Point3D`). It crossed my mind that we could use the rotation transforms of >> the node to control the direction instead, just like we use the >> translation/layout of the node to get the position (there is an internal >> Affine3D transform for lights, not sure why `AmbientLight` needs it). >> Wouldn't that make more sense? When I rotate the light I would expect to see >> a change in direction. >> >> ### Implementation discussion points >> >> - [ ] I've gotten advice from a graphics engineer to treat point lights as >> spot lights with a 360 degrees coverage, which simplifies a few places. We >> can still try to optimize for a point light by looking at the light >> parameters: `falloff = 0` and `outerAngle = 180`. These possible >> optimization exist in `ES2PhongShader.java` and `D3DMeshView.cc`, and in the >> pixel/fragment shaders in the form of 3 different ways to compute the >> spotlight factor (the `computeLightN` methods). We need to check which of >> these give the best results. >> >> ## Performance >> >> Testing 3 point lights and comparing this branch with `master` using a 1000 >> division sphere, 200 meshes, and 5000 meshes. >> Using an AMD RX 470 4GB GPU. >> >> In this branch, there is a possible CPU optimization for checking the light >> type and using precalculated values (in `D3DMeshView.cc` for d3d and >> `ES2PhongShader.java` for opengl). On the GPU, I tried 3 ways of computing >> the spotlight factor contributions (`computeSpotlightFactor`, >> `computeSpotlightFactor2` and `computeSpotlightFactor3`) trying out >> different branching and shortcuts. >> >> ### Results >> The CPU "optimizations" made no difference, which is understandable >> considering it will not be the bottleneck. We can remove these if we want to >> simplify, though maybe if we allow a large number of lights it could make a >> difference (I doubt it). I don't have a strong preference either way. >> >> The sphere 1000 tests always gave max fps (120 on Win and 60 on Ubuntu). >> >> **Win 10** >> Compared with the `master` branch, this patch shows 5-10 fps drop in the >> mesh 200 test and ~5 in the mesh 5000 test. I repeated the tests on several >> occasions and got different results in terms of absolute numbers, but the >> relative performance difference remained more or less the same. Out of the 3 >> `computeSpotlightFactor` methods, `computeSpotlightFactor3`, which has no >> "optimizations", gives slightly better performance. >> >> **Ubuntu 18** >> The mesh 200 test always gave 60 fps because it is locked to this fps, so we >> can't measure the real GPU performance change. >> The mesh 5000 test shows 2-6 fps drop from master, with >> `computeSpotlightFactor` > `computeSpotlightFactor2` > >> `computeSpotlightFactor3` at in terms of performance (~2 fps difference >> each). >> >> **Conclusion**: we can expect a 5 fps drop more or less with 3 point lights. >> `computeSpotlightFactor3` on d3d and `computeSpotlightFactor` on opengl gave >> the best performances. > > Nir Lisker has updated the pull request incrementally with one additional > commit since the last revision: > > Combined rotation and direction I started to look at this today, and will continue next week. There is also the outstanding issue with a retina display (scale=2) on macOS. - PR: https://git.openjdk.java.net/jfx/pull/334
Re: RFR: 8234920: Add SpotLight to the selection of 3D light types [v14]
On Thu, 15 Apr 2021 02:21:50 GMT, Nir Lisker wrote: >> Added a SpotLight only to the D3D pipeline currently. >> >> ### API discussion points >> >> - [X] Added `SpotLight` as a subclass of `LightBase`. However, it could >> also be a subclass of `PointLight` as it's a point light with direction and >> extra factors. I saw that `scenario.effect.light.SpotLight` extends its >> respective `PointLight`, but it's not a perfect analogy. In the end, I think >> it's a questions of whether `PointLight` will be expanded in a way which >> doesn't not suit `SpotLight`, and I tend to think that the answer is no. >> >> - [X] The inner and outer angles are the "diameter angles" as shown >> [here](https://docs.microsoft.com/en-us/windows/win32/direct3d9/light-typeshttps://docs.microsoft.com/en-us/windows/win32/direct3d9/light-types). >> I, personally, find it more intuitive that these are the "radius angles", >> so half these angles, as used in the spotlight factor formula. Do you think >> I can change this or do you prefer the current definition of the angles? >> >> - [ ] The current implementation uses an ad-hoc direction property (using a >> `Point3D`). It crossed my mind that we could use the rotation transforms of >> the node to control the direction instead, just like we use the >> translation/layout of the node to get the position (there is an internal >> Affine3D transform for lights, not sure why `AmbientLight` needs it). >> Wouldn't that make more sense? When I rotate the light I would expect to see >> a change in direction. >> >> ### Implementation discussion points >> >> - [ ] I've gotten advice from a graphics engineer to treat point lights as >> spot lights with a 360 degrees coverage, which simplifies a few places. We >> can still try to optimize for a point light by looking at the light >> parameters: `falloff = 0` and `outerAngle = 180`. These possible >> optimization exist in `ES2PhongShader.java` and `D3DMeshView.cc`, and in the >> pixel/fragment shaders in the form of 3 different ways to compute the >> spotlight factor (the `computeLightN` methods). We need to check which of >> these give the best results. >> >> ## Performance >> >> Testing 3 point lights and comparing this branch with `master` using a 1000 >> division sphere, 200 meshes, and 5000 meshes. >> Using an AMD RX 470 4GB GPU. >> >> In this branch, there is a possible CPU optimization for checking the light >> type and using precalculated values (in `D3DMeshView.cc` for d3d and >> `ES2PhongShader.java` for opengl). On the GPU, I tried 3 ways of computing >> the spotlight factor contributions (`computeSpotlightFactor`, >> `computeSpotlightFactor2` and `computeSpotlightFactor3`) trying out >> different branching and shortcuts. >> >> ### Results >> The CPU "optimizations" made no difference, which is understandable >> considering it will not be the bottleneck. We can remove these if we want to >> simplify, though maybe if we allow a large number of lights it could make a >> difference (I doubt it). I don't have a strong preference either way. >> >> The sphere 1000 tests always gave max fps (120 on Win and 60 on Ubuntu). >> >> **Win 10** >> Compared with the `master` branch, this patch shows 5-10 fps drop in the >> mesh 200 test and ~5 in the mesh 5000 test. I repeated the tests on several >> occasions and got different results in terms of absolute numbers, but the >> relative performance difference remained more or less the same. Out of the 3 >> `computeSpotlightFactor` methods, `computeSpotlightFactor3`, which has no >> "optimizations", gives slightly better performance. >> >> **Ubuntu 18** >> The mesh 200 test always gave 60 fps because it is locked to this fps, so we >> can't measure the real GPU performance change. >> The mesh 5000 test shows 2-6 fps drop from master, with >> `computeSpotlightFactor` > `computeSpotlightFactor2` > >> `computeSpotlightFactor3` at in terms of performance (~2 fps difference >> each). >> >> **Conclusion**: we can expect a 5 fps drop more or less with 3 point lights. >> `computeSpotlightFactor3` on d3d and `computeSpotlightFactor` on opengl gave >> the best performances. > > Nir Lisker has updated the pull request incrementally with one additional > commit since the last revision: > > Combined rotation and direction Can we continue with the review? We need to decide about the rotation behavior and the performance options need to be tested. - PR: https://git.openjdk.java.net/jfx/pull/334
Re: RFR: 8234920: Add SpotLight to the selection of 3D light types [v14]
On Thu, 15 Apr 2021 02:21:50 GMT, Nir Lisker wrote: >> Added a SpotLight only to the D3D pipeline currently. >> >> ### API discussion points >> >> - [X] Added `SpotLight` as a subclass of `LightBase`. However, it could >> also be a subclass of `PointLight` as it's a point light with direction and >> extra factors. I saw that `scenario.effect.light.SpotLight` extends its >> respective `PointLight`, but it's not a perfect analogy. In the end, I think >> it's a questions of whether `PointLight` will be expanded in a way which >> doesn't not suit `SpotLight`, and I tend to think that the answer is no. >> >> - [X] The inner and outer angles are the "diameter angles" as shown >> [here](https://docs.microsoft.com/en-us/windows/win32/direct3d9/light-typeshttps://docs.microsoft.com/en-us/windows/win32/direct3d9/light-types). >> I, personally, find it more intuitive that these are the "radius angles", >> so half these angles, as used in the spotlight factor formula. Do you think >> I can change this or do you prefer the current definition of the angles? >> >> - [ ] The current implementation uses an ad-hoc direction property (using a >> `Point3D`). It crossed my mind that we could use the rotation transforms of >> the node to control the direction instead, just like we use the >> translation/layout of the node to get the position (there is an internal >> Affine3D transform for lights, not sure why `AmbientLight` needs it). >> Wouldn't that make more sense? When I rotate the light I would expect to see >> a change in direction. >> >> ### Implementation discussion points >> >> - [ ] I've gotten advice from a graphics engineer to treat point lights as >> spot lights with a 360 degrees coverage, which simplifies a few places. We >> can still try to optimize for a point light by looking at the light >> parameters: `falloff = 0` and `outerAngle = 180`. These possible >> optimization exist in `ES2PhongShader.java` and `D3DMeshView.cc`, and in the >> pixel/fragment shaders in the form of 3 different ways to compute the >> spotlight factor (the `computeLightN` methods). We need to check which of >> these give the best results. >> >> ## Performance >> >> Testing 3 point lights and comparing this branch with `master` using a 1000 >> division sphere, 200 meshes, and 5000 meshes. >> Using an AMD RX 470 4GB GPU. >> >> In this branch, there is a possible CPU optimization for checking the light >> type and using precalculated values (in `D3DMeshView.cc` for d3d and >> `ES2PhongShader.java` for opengl). On the GPU, I tried 3 ways of computing >> the spotlight factor contributions (`computeSpotlightFactor`, >> `computeSpotlightFactor2` and `computeSpotlightFactor3`) trying out >> different branching and shortcuts. >> >> ### Results >> The CPU "optimizations" made no difference, which is understandable >> considering it will not be the bottleneck. We can remove these if we want to >> simplify, though maybe if we allow a large number of lights it could make a >> difference (I doubt it). I don't have a strong preference either way. >> >> The sphere 1000 tests always gave max fps (120 on Win and 60 on Ubuntu). >> >> **Win 10** >> Compared with the `master` branch, this patch shows 5-10 fps drop in the >> mesh 200 test and ~5 in the mesh 5000 test. I repeated the tests on several >> occasions and got different results in terms of absolute numbers, but the >> relative performance difference remained more or less the same. Out of the 3 >> `computeSpotlightFactor` methods, `computeSpotlightFactor3`, which has no >> "optimizations", gives slightly better performance. >> >> **Ubuntu 18** >> The mesh 200 test always gave 60 fps because it is locked to this fps, so we >> can't measure the real GPU performance change. >> The mesh 5000 test shows 2-6 fps drop from master, with >> `computeSpotlightFactor` > `computeSpotlightFactor2` > >> `computeSpotlightFactor3` at in terms of performance (~2 fps difference >> each). >> >> **Conclusion**: we can expect a 5 fps drop more or less with 3 point lights. >> `computeSpotlightFactor3` on d3d and `computeSpotlightFactor` on opengl gave >> the best performances. > > Nir Lisker has updated the pull request incrementally with one additional > commit since the last revision: > > Combined rotation and direction I asked someone with a Mac Book Pro with a retina screen to run JavaFX 16, and I saw an issue with lighting positioning: the source of the light was not the position of the `PointLight` node. I couldn't debug it, but it seems that this issue is not related to this patch, and possibly not even to the range property specifically. - PR: https://git.openjdk.java.net/jfx/pull/334
Re: RFR: 8234920: Add SpotLight to the selection of 3D light types [v14]
On Thu, 15 Apr 2021 02:21:50 GMT, Nir Lisker wrote: >> Added a SpotLight only to the D3D pipeline currently. >> >> ### API discussion points >> >> - [X] Added `SpotLight` as a subclass of `LightBase`. However, it could >> also be a subclass of `PointLight` as it's a point light with direction and >> extra factors. I saw that `scenario.effect.light.SpotLight` extends its >> respective `PointLight`, but it's not a perfect analogy. In the end, I think >> it's a questions of whether `PointLight` will be expanded in a way which >> doesn't not suit `SpotLight`, and I tend to think that the answer is no. >> >> - [X] The inner and outer angles are the "diameter angles" as shown >> [here](https://docs.microsoft.com/en-us/windows/win32/direct3d9/light-typeshttps://docs.microsoft.com/en-us/windows/win32/direct3d9/light-types). >> I, personally, find it more intuitive that these are the "radius angles", >> so half these angles, as used in the spotlight factor formula. Do you think >> I can change this or do you prefer the current definition of the angles? >> >> - [ ] The current implementation uses an ad-hoc direction property (using a >> `Point3D`). It crossed my mind that we could use the rotation transforms of >> the node to control the direction instead, just like we use the >> translation/layout of the node to get the position (there is an internal >> Affine3D transform for lights, not sure why `AmbientLight` needs it). >> Wouldn't that make more sense? When I rotate the light I would expect to see >> a change in direction. >> >> ### Implementation discussion points >> >> - [ ] I've gotten advice from a graphics engineer to treat point lights as >> spot lights with a 360 degrees coverage, which simplifies a few places. We >> can still try to optimize for a point light by looking at the light >> parameters: `falloff = 0` and `outerAngle = 180`. These possible >> optimization exist in `ES2PhongShader.java` and `D3DMeshView.cc`, and in the >> pixel/fragment shaders in the form of 3 different ways to compute the >> spotlight factor (the `computeLightN` methods). We need to check which of >> these give the best results. >> >> ## Performance >> >> Testing 3 point lights and comparing this branch with `master` using a 1000 >> division sphere, 200 meshes, and 5000 meshes. >> Using an AMD RX 470 4GB GPU. >> >> In this branch, there is a possible CPU optimization for checking the light >> type and using precalculated values (in `D3DMeshView.cc` for d3d and >> `ES2PhongShader.java` for opengl). On the GPU, I tried 3 ways of computing >> the spotlight factor contributions (`computeSpotlightFactor`, >> `computeSpotlightFactor2` and `computeSpotlightFactor3`) trying out >> different branching and shortcuts. >> >> ### Results >> The CPU "optimizations" made no difference, which is understandable >> considering it will not be the bottleneck. We can remove these if we want to >> simplify, though maybe if we allow a large number of lights it could make a >> difference (I doubt it). I don't have a strong preference either way. >> >> The sphere 1000 tests always gave max fps (120 on Win and 60 on Ubuntu). >> >> **Win 10** >> Compared with the `master` branch, this patch shows 5-10 fps drop in the >> mesh 200 test and ~5 in the mesh 5000 test. I repeated the tests on several >> occasions and got different results in terms of absolute numbers, but the >> relative performance difference remained more or less the same. Out of the 3 >> `computeSpotlightFactor` methods, `computeSpotlightFactor3`, which has no >> "optimizations", gives slightly better performance. >> >> **Ubuntu 18** >> The mesh 200 test always gave 60 fps because it is locked to this fps, so we >> can't measure the real GPU performance change. >> The mesh 5000 test shows 2-6 fps drop from master, with >> `computeSpotlightFactor` > `computeSpotlightFactor2` > >> `computeSpotlightFactor3` at in terms of performance (~2 fps difference >> each). >> >> **Conclusion**: we can expect a 5 fps drop more or less with 3 point lights. >> `computeSpotlightFactor3` on d3d and `computeSpotlightFactor` on opengl gave >> the best performances. > > Nir Lisker has updated the pull request incrementally with one additional > commit since the last revision: > > Combined rotation and direction Yep, there's definitely a scale problem going on. It looks much different if I run it on a non-Retina display. And knowing that, I was able to reproduce it on Windows as well by comparing what it looks like with `-Dglass.win.uiScale=1` and `-Dglass.win.uiScale=2`. - PR: https://git.openjdk.java.net/jfx/pull/334
Re: RFR: 8234920: Add SpotLight to the selection of 3D light types [v14]
On Thu, 15 Apr 2021 02:21:50 GMT, Nir Lisker wrote: >> Added a SpotLight only to the D3D pipeline currently. >> >> ### API discussion points >> >> - [X] Added `SpotLight` as a subclass of `LightBase`. However, it could >> also be a subclass of `PointLight` as it's a point light with direction and >> extra factors. I saw that `scenario.effect.light.SpotLight` extends its >> respective `PointLight`, but it's not a perfect analogy. In the end, I think >> it's a questions of whether `PointLight` will be expanded in a way which >> doesn't not suit `SpotLight`, and I tend to think that the answer is no. >> >> - [X] The inner and outer angles are the "diameter angles" as shown >> [here](https://docs.microsoft.com/en-us/windows/win32/direct3d9/light-typeshttps://docs.microsoft.com/en-us/windows/win32/direct3d9/light-types). >> I, personally, find it more intuitive that these are the "radius angles", >> so half these angles, as used in the spotlight factor formula. Do you think >> I can change this or do you prefer the current definition of the angles? >> >> - [ ] The current implementation uses an ad-hoc direction property (using a >> `Point3D`). It crossed my mind that we could use the rotation transforms of >> the node to control the direction instead, just like we use the >> translation/layout of the node to get the position (there is an internal >> Affine3D transform for lights, not sure why `AmbientLight` needs it). >> Wouldn't that make more sense? When I rotate the light I would expect to see >> a change in direction. >> >> ### Implementation discussion points >> >> - [ ] I've gotten advice from a graphics engineer to treat point lights as >> spot lights with a 360 degrees coverage, which simplifies a few places. We >> can still try to optimize for a point light by looking at the light >> parameters: `falloff = 0` and `outerAngle = 180`. These possible >> optimization exist in `ES2PhongShader.java` and `D3DMeshView.cc`, and in the >> pixel/fragment shaders in the form of 3 different ways to compute the >> spotlight factor (the `computeLightN` methods). We need to check which of >> these give the best results. >> >> ## Performance >> >> Testing 3 point lights and comparing this branch with `master` using a 1000 >> division sphere, 200 meshes, and 5000 meshes. >> Using an AMD RX 470 4GB GPU. >> >> In this branch, there is a possible CPU optimization for checking the light >> type and using precalculated values (in `D3DMeshView.cc` for d3d and >> `ES2PhongShader.java` for opengl). On the GPU, I tried 3 ways of computing >> the spotlight factor contributions (`computeSpotlightFactor`, >> `computeSpotlightFactor2` and `computeSpotlightFactor3`) trying out >> different branching and shortcuts. >> >> ### Results >> The CPU "optimizations" made no difference, which is understandable >> considering it will not be the bottleneck. We can remove these if we want to >> simplify, though maybe if we allow a large number of lights it could make a >> difference (I doubt it). I don't have a strong preference either way. >> >> The sphere 1000 tests always gave max fps (120 on Win and 60 on Ubuntu). >> >> **Win 10** >> Compared with the `master` branch, this patch shows 5-10 fps drop in the >> mesh 200 test and ~5 in the mesh 5000 test. I repeated the tests on several >> occasions and got different results in terms of absolute numbers, but the >> relative performance difference remained more or less the same. Out of the 3 >> `computeSpotlightFactor` methods, `computeSpotlightFactor3`, which has no >> "optimizations", gives slightly better performance. >> >> **Ubuntu 18** >> The mesh 200 test always gave 60 fps because it is locked to this fps, so we >> can't measure the real GPU performance change. >> The mesh 5000 test shows 2-6 fps drop from master, with >> `computeSpotlightFactor` > `computeSpotlightFactor2` > >> `computeSpotlightFactor3` at in terms of performance (~2 fps difference >> each). >> >> **Conclusion**: we can expect a 5 fps drop more or less with 3 point lights. >> `computeSpotlightFactor3` on d3d and `computeSpotlightFactor` on opengl gave >> the best performances. > > Nir Lisker has updated the pull request incrementally with one additional > commit since the last revision: > > Combined rotation and direction I'll try it on macOS with a non-retina display and see if that makes a difference (if so, it could point to a transform issue). I'll also try with the Boxes as you suggest. - PR: https://git.openjdk.java.net/jfx/pull/334
Re: RFR: 8234920: Add SpotLight to the selection of 3D light types [v14]
On Thu, 15 Apr 2021 02:21:50 GMT, Nir Lisker wrote: >> Added a SpotLight only to the D3D pipeline currently. >> >> ### API discussion points >> >> - [X] Added `SpotLight` as a subclass of `LightBase`. However, it could >> also be a subclass of `PointLight` as it's a point light with direction and >> extra factors. I saw that `scenario.effect.light.SpotLight` extends its >> respective `PointLight`, but it's not a perfect analogy. In the end, I think >> it's a questions of whether `PointLight` will be expanded in a way which >> doesn't not suit `SpotLight`, and I tend to think that the answer is no. >> >> - [X] The inner and outer angles are the "diameter angles" as shown >> [here](https://docs.microsoft.com/en-us/windows/win32/direct3d9/light-typeshttps://docs.microsoft.com/en-us/windows/win32/direct3d9/light-types). >> I, personally, find it more intuitive that these are the "radius angles", >> so half these angles, as used in the spotlight factor formula. Do you think >> I can change this or do you prefer the current definition of the angles? >> >> - [ ] The current implementation uses an ad-hoc direction property (using a >> `Point3D`). It crossed my mind that we could use the rotation transforms of >> the node to control the direction instead, just like we use the >> translation/layout of the node to get the position (there is an internal >> Affine3D transform for lights, not sure why `AmbientLight` needs it). >> Wouldn't that make more sense? When I rotate the light I would expect to see >> a change in direction. >> >> ### Implementation discussion points >> >> - [ ] I've gotten advice from a graphics engineer to treat point lights as >> spot lights with a 360 degrees coverage, which simplifies a few places. We >> can still try to optimize for a point light by looking at the light >> parameters: `falloff = 0` and `outerAngle = 180`. These possible >> optimization exist in `ES2PhongShader.java` and `D3DMeshView.cc`, and in the >> pixel/fragment shaders in the form of 3 different ways to compute the >> spotlight factor (the `computeLightN` methods). We need to check which of >> these give the best results. >> >> ## Performance >> >> Testing 3 point lights and comparing this branch with `master` using a 1000 >> division sphere, 200 meshes, and 5000 meshes. >> Using an AMD RX 470 4GB GPU. >> >> In this branch, there is a possible CPU optimization for checking the light >> type and using precalculated values (in `D3DMeshView.cc` for d3d and >> `ES2PhongShader.java` for opengl). On the GPU, I tried 3 ways of computing >> the spotlight factor contributions (`computeSpotlightFactor`, >> `computeSpotlightFactor2` and `computeSpotlightFactor3`) trying out >> different branching and shortcuts. >> >> ### Results >> The CPU "optimizations" made no difference, which is understandable >> considering it will not be the bottleneck. We can remove these if we want to >> simplify, though maybe if we allow a large number of lights it could make a >> difference (I doubt it). I don't have a strong preference either way. >> >> The sphere 1000 tests always gave max fps (120 on Win and 60 on Ubuntu). >> >> **Win 10** >> Compared with the `master` branch, this patch shows 5-10 fps drop in the >> mesh 200 test and ~5 in the mesh 5000 test. I repeated the tests on several >> occasions and got different results in terms of absolute numbers, but the >> relative performance difference remained more or less the same. Out of the 3 >> `computeSpotlightFactor` methods, `computeSpotlightFactor3`, which has no >> "optimizations", gives slightly better performance. >> >> **Ubuntu 18** >> The mesh 200 test always gave 60 fps because it is locked to this fps, so we >> can't measure the real GPU performance change. >> The mesh 5000 test shows 2-6 fps drop from master, with >> `computeSpotlightFactor` > `computeSpotlightFactor2` > >> `computeSpotlightFactor3` at in terms of performance (~2 fps difference >> each). >> >> **Conclusion**: we can expect a 5 fps drop more or less with 3 point lights. >> `computeSpotlightFactor3` on d3d and `computeSpotlightFactor` on opengl gave >> the best performances. > > Nir Lisker has updated the pull request incrementally with one additional > commit since the last revision: > > Combined rotation and direction I'll fix the documentation once we establish how the light is rotated. I will need to add notes about the rotation transforms too. - PR: https://git.openjdk.java.net/jfx/pull/334
Re: RFR: 8234920: Add SpotLight to the selection of 3D light types [v14]
On Thu, 15 Apr 2021 02:21:50 GMT, Nir Lisker wrote: >> Added a SpotLight only to the D3D pipeline currently. >> >> ### API discussion points >> >> - [X] Added `SpotLight` as a subclass of `LightBase`. However, it could >> also be a subclass of `PointLight` as it's a point light with direction and >> extra factors. I saw that `scenario.effect.light.SpotLight` extends its >> respective `PointLight`, but it's not a perfect analogy. In the end, I think >> it's a questions of whether `PointLight` will be expanded in a way which >> doesn't not suit `SpotLight`, and I tend to think that the answer is no. >> >> - [X] The inner and outer angles are the "diameter angles" as shown >> [here](https://docs.microsoft.com/en-us/windows/win32/direct3d9/light-typeshttps://docs.microsoft.com/en-us/windows/win32/direct3d9/light-types). >> I, personally, find it more intuitive that these are the "radius angles", >> so half these angles, as used in the spotlight factor formula. Do you think >> I can change this or do you prefer the current definition of the angles? >> >> - [ ] The current implementation uses an ad-hoc direction property (using a >> `Point3D`). It crossed my mind that we could use the rotation transforms of >> the node to control the direction instead, just like we use the >> translation/layout of the node to get the position (there is an internal >> Affine3D transform for lights, not sure why `AmbientLight` needs it). >> Wouldn't that make more sense? When I rotate the light I would expect to see >> a change in direction. >> >> ### Implementation discussion points >> >> - [ ] I've gotten advice from a graphics engineer to treat point lights as >> spot lights with a 360 degrees coverage, which simplifies a few places. We >> can still try to optimize for a point light by looking at the light >> parameters: `falloff = 0` and `outerAngle = 180`. These possible >> optimization exist in `ES2PhongShader.java` and `D3DMeshView.cc`, and in the >> pixel/fragment shaders in the form of 3 different ways to compute the >> spotlight factor (the `computeLightN` methods). We need to check which of >> these give the best results. >> >> ## Performance >> >> Testing 3 point lights and comparing this branch with `master` using a 1000 >> division sphere, 200 meshes, and 5000 meshes. >> Using an AMD RX 470 4GB GPU. >> >> In this branch, there is a possible CPU optimization for checking the light >> type and using precalculated values (in `D3DMeshView.cc` for d3d and >> `ES2PhongShader.java` for opengl). On the GPU, I tried 3 ways of computing >> the spotlight factor contributions (`computeSpotlightFactor`, >> `computeSpotlightFactor2` and `computeSpotlightFactor3`) trying out >> different branching and shortcuts. >> >> ### Results >> The CPU "optimizations" made no difference, which is understandable >> considering it will not be the bottleneck. We can remove these if we want to >> simplify, though maybe if we allow a large number of lights it could make a >> difference (I doubt it). I don't have a strong preference either way. >> >> The sphere 1000 tests always gave max fps (120 on Win and 60 on Ubuntu). >> >> **Win 10** >> Compared with the `master` branch, this patch shows 5-10 fps drop in the >> mesh 200 test and ~5 in the mesh 5000 test. I repeated the tests on several >> occasions and got different results in terms of absolute numbers, but the >> relative performance difference remained more or less the same. Out of the 3 >> `computeSpotlightFactor` methods, `computeSpotlightFactor3`, which has no >> "optimizations", gives slightly better performance. >> >> **Ubuntu 18** >> The mesh 200 test always gave 60 fps because it is locked to this fps, so we >> can't measure the real GPU performance change. >> The mesh 5000 test shows 2-6 fps drop from master, with >> `computeSpotlightFactor` > `computeSpotlightFactor2` > >> `computeSpotlightFactor3` at in terms of performance (~2 fps difference >> each). >> >> **Conclusion**: we can expect a 5 fps drop more or less with 3 point lights. >> `computeSpotlightFactor3` on d3d and `computeSpotlightFactor` on opengl gave >> the best performances. > > Nir Lisker has updated the pull request incrementally with one additional > commit since the last revision: > > Combined rotation and direction Provided few comments on documentation. Have to review and test the code. modules/javafx.graphics/src/main/java/javafx/scene/SpotLight.java line 46: > 44: * > 45: * The light cone is defined by 3 factors: an {@link > #innerAngleProperty() inner angle}, an {@link #outerAngleProperty() > 46: * outer angle}, and a {@link #falloffProperty() falloff} factor. For a > point whose angle to the light is {@code a}, if -> A {@code SpotLight} is a {@code PointLight} that radiates a cone of light in a specific direction. < p > The direction is defined by the {@link #directionProperty() direction} property. The light cone is defined by 3 prope
Re: RFR: 8234920: Add SpotLight to the selection of 3D light types [v14]
On Fri, 16 Apr 2021 10:35:29 GMT, Ambarish Rapte wrote: >> Nir Lisker has updated the pull request incrementally with one additional >> commit since the last revision: >> >> Combined rotation and direction > > modules/javafx.graphics/src/main/java/javafx/scene/SpotLight.java line 46: > >> 44: * >> 45: * The light cone is defined by 3 factors: an {@link >> #innerAngleProperty() inner angle}, an {@link #outerAngleProperty() >> 46: * outer angle}, and a {@link #falloffProperty() falloff} factor. For a >> point whose angle to the light is {@code a}, if > > -> > A {@code SpotLight} is a {@code PointLight} that radiates a cone of light in > a specific direction. > < p > > The direction is defined by the {@link #directionProperty() direction} > property. > The light cone is defined by 3 properties: an {@link #innerAngleProperty() > inner angle}, an {@link #outerAngleProperty() outer angle}, and a {@link > #falloffProperty() falloff} property. Also Will it be illustrative to mention these three properties using li tags with a statement that briefs the use of property. Something like, - {@link #innerAngleProperty() inner angle}: Angle of the inner cone where light intensity is always maximum. - {@link #outerAngleProperty() outer angle}: Angle of the outer cone where light intensity attenuates based on fallof. - {@link #falloffProperty() falloff}: The attenuation factor that controls the attenuation of light intensity from edge of inner cone to the edge of outer cone. - PR: https://git.openjdk.java.net/jfx/pull/334
Re: RFR: 8234920: Add SpotLight to the selection of 3D light types [v14]
On Thu, 15 Apr 2021 02:21:50 GMT, Nir Lisker wrote: >> Added a SpotLight only to the D3D pipeline currently. >> >> ### API discussion points >> >> - [X] Added `SpotLight` as a subclass of `LightBase`. However, it could >> also be a subclass of `PointLight` as it's a point light with direction and >> extra factors. I saw that `scenario.effect.light.SpotLight` extends its >> respective `PointLight`, but it's not a perfect analogy. In the end, I think >> it's a questions of whether `PointLight` will be expanded in a way which >> doesn't not suit `SpotLight`, and I tend to think that the answer is no. >> >> - [X] The inner and outer angles are the "diameter angles" as shown >> [here](https://docs.microsoft.com/en-us/windows/win32/direct3d9/light-typeshttps://docs.microsoft.com/en-us/windows/win32/direct3d9/light-types). >> I, personally, find it more intuitive that these are the "radius angles", >> so half these angles, as used in the spotlight factor formula. Do you think >> I can change this or do you prefer the current definition of the angles? >> >> - [ ] The current implementation uses an ad-hoc direction property (using a >> `Point3D`). It crossed my mind that we could use the rotation transforms of >> the node to control the direction instead, just like we use the >> translation/layout of the node to get the position (there is an internal >> Affine3D transform for lights, not sure why `AmbientLight` needs it). >> Wouldn't that make more sense? When I rotate the light I would expect to see >> a change in direction. >> >> ### Implementation discussion points >> >> - [ ] I've gotten advice from a graphics engineer to treat point lights as >> spot lights with a 360 degrees coverage, which simplifies a few places. We >> can still try to optimize for a point light by looking at the light >> parameters: `falloff = 0` and `outerAngle = 180`. These possible >> optimization exist in `ES2PhongShader.java` and `D3DMeshView.cc`, and in the >> pixel/fragment shaders in the form of 3 different ways to compute the >> spotlight factor (the `computeLightN` methods). We need to check which of >> these give the best results. >> >> ## Performance >> >> Testing 3 point lights and comparing this branch with `master` using a 1000 >> division sphere, 200 meshes, and 5000 meshes. >> Using an AMD RX 470 4GB GPU. >> >> In this branch, there is a possible CPU optimization for checking the light >> type and using precalculated values (in `D3DMeshView.cc` for d3d and >> `ES2PhongShader.java` for opengl). On the GPU, I tried 3 ways of computing >> the spotlight factor contributions (`computeSpotlightFactor`, >> `computeSpotlightFactor2` and `computeSpotlightFactor3`) trying out >> different branching and shortcuts. >> >> ### Results >> The CPU "optimizations" made no difference, which is understandable >> considering it will not be the bottleneck. We can remove these if we want to >> simplify, though maybe if we allow a large number of lights it could make a >> difference (I doubt it). I don't have a strong preference either way. >> >> The sphere 1000 tests always gave max fps (120 on Win and 60 on Ubuntu). >> >> **Win 10** >> Compared with the `master` branch, this patch shows 5-10 fps drop in the >> mesh 200 test and ~5 in the mesh 5000 test. I repeated the tests on several >> occasions and got different results in terms of absolute numbers, but the >> relative performance difference remained more or less the same. Out of the 3 >> `computeSpotlightFactor` methods, `computeSpotlightFactor3`, which has no >> "optimizations", gives slightly better performance. >> >> **Ubuntu 18** >> The mesh 200 test always gave 60 fps because it is locked to this fps, so we >> can't measure the real GPU performance change. >> The mesh 5000 test shows 2-6 fps drop from master, with >> `computeSpotlightFactor` > `computeSpotlightFactor2` > >> `computeSpotlightFactor3` at in terms of performance (~2 fps difference >> each). >> >> **Conclusion**: we can expect a 5 fps drop more or less with 3 point lights. >> `computeSpotlightFactor3` on d3d and `computeSpotlightFactor` on opengl gave >> the best performances. > > Nir Lisker has updated the pull request incrementally with one additional > commit since the last revision: > > Combined rotation and direction I update the patch to include rotation transforms applying on the direction vector. When testing, I suggest using the "Boxes" model. You will need to rotate (right/middle mouse buttons) to see model properly as the back wall obscures the camera's initial position. - PR: https://git.openjdk.java.net/jfx/pull/334
Re: RFR: 8234920: Add SpotLight to the selection of 3D light types [v14]
> Added a SpotLight only to the D3D pipeline currently. > > ### API discussion points > > - [X] Added `SpotLight` as a subclass of `LightBase`. However, it could also > be a subclass of `PointLight` as it's a point light with direction and extra > factors. I saw that `scenario.effect.light.SpotLight` extends its respective > `PointLight`, but it's not a perfect analogy. In the end, I think it's a > questions of whether `PointLight` will be expanded in a way which doesn't not > suit `SpotLight`, and I tend to think that the answer is no. > > - [X] The inner and outer angles are the "diameter angles" as shown > [here](https://docs.microsoft.com/en-us/windows/win32/direct3d9/light-typeshttps://docs.microsoft.com/en-us/windows/win32/direct3d9/light-types). > I, personally, find it more intuitive that these are the "radius angles", > so half these angles, as used in the spotlight factor formula. Do you think I > can change this or do you prefer the current definition of the angles? > > - [ ] The current implementation uses an ad-hoc direction property (using a > `Point3D`). It crossed my mind that we could use the rotation transforms of > the node to control the direction instead, just like we use the > translation/layout of the node to get the position (there is an internal > Affine3D transform for lights, not sure why `AmbientLight` needs it). > Wouldn't that make more sense? When I rotate the light I would expect to see > a change in direction. > > ### Implementation discussion points > > - [ ] I've gotten advice from a graphics engineer to treat point lights as > spot lights with a 360 degrees coverage, which simplifies a few places. We > can still try to optimize for a point light by looking at the light > parameters: `falloff = 0` and `outerAngle = 180`. These possible optimization > exist in `ES2PhongShader.java` and `D3DMeshView.cc`, and in the > pixel/fragment shaders in the form of 3 different ways to compute the > spotlight factor (the `computeLightN` methods). We need to check which of > these give the best results. > > ## Performance > > Testing 3 point lights and comparing this branch with `master` using a 1000 > division sphere, 200 meshes, and 5000 meshes. > Using an AMD RX 470 4GB GPU. > > In this branch, there is a possible CPU optimization for checking the light > type and using precalculated values (in `D3DMeshView.cc` for d3d and > `ES2PhongShader.java` for opengl). On the GPU, I tried 3 ways of computing > the spotlight factor contributions (`computeSpotlightFactor`, > `computeSpotlightFactor2` and `computeSpotlightFactor3`) trying out different > branching and shortcuts. > > ### Results > The CPU "optimizations" made no difference, which is understandable > considering it will not be the bottleneck. We can remove these if we want to > simplify, though maybe if we allow a large number of lights it could make a > difference (I doubt it). I don't have a strong preference either way. > > The sphere 1000 tests always gave max fps (120 on Win and 60 on Ubuntu). > > **Win 10** > Compared with the `master` branch, this patch shows 5-10 fps drop in the mesh > 200 test and ~5 in the mesh 5000 test. I repeated the tests on several > occasions and got different results in terms of absolute numbers, but the > relative performance difference remained more or less the same. Out of the 3 > `computeSpotlightFactor` methods, `computeSpotlightFactor3`, which has no > "optimizations", gives slightly better performance. > > **Ubuntu 18** > The mesh 200 test always gave 60 fps because it is locked to this fps, so we > can't measure the real GPU performance change. > The mesh 5000 test shows 2-6 fps drop from master, with > `computeSpotlightFactor` > `computeSpotlightFactor2` > > `computeSpotlightFactor3` at in terms of performance (~2 fps difference each). > > **Conclusion**: we can expect a 5 fps drop more or less with 3 point lights. > `computeSpotlightFactor3` on d3d and `computeSpotlightFactor` on opengl gave > the best performances. Nir Lisker has updated the pull request incrementally with one additional commit since the last revision: Combined rotation and direction - Changes: - all: https://git.openjdk.java.net/jfx/pull/334/files - new: https://git.openjdk.java.net/jfx/pull/334/files/7c36639e..3d6cdb19 Webrevs: - full: https://webrevs.openjdk.java.net/?repo=jfx&pr=334&range=13 - incr: https://webrevs.openjdk.java.net/?repo=jfx&pr=334&range=12-13 Stats: 27 lines in 3 files changed: 17 ins; 3 del; 7 mod Patch: https://git.openjdk.java.net/jfx/pull/334.diff Fetch: git fetch https://git.openjdk.java.net/jfx pull/334/head:pull/334 PR: https://git.openjdk.java.net/jfx/pull/334