So, I read this answer, and while it seems to work, it dramatically lowers my FPS, and generally my program can only run for 10 seconds before WebGL just completely crashes.
Basically, I'm trying to implement lighting in my 2D top-down game. I'm doing this by having two WebGL canvases. In the first one, each pixel ranges from pure black to pure white, and each pixel represents the intensity of the light at that location. The second WebGL canvas is just my regular game canvas where everything in the game is rendered. Then, during runtime, these two canvases are blended together every frame, to simulate lighting.
I'm not just using a single canvas because I'm not sure how to do it with just one. I cannot simply darken my game canvas and then selectively brighten lit areas, because the original pixel colors are lost from the darkening shader. So instead I'm trying to use two canvases. The first one accumulates all the lighting values, and then when all the lighting values are determined, there is a single blending operation that blends the two canvases.
Again, this is working, but it's very slow and prone to crashes. I think the issue is that I am creating two textures every frame (one of the lighting canvas, and one of the game canvas) and then uploading all the texture data to each texture, and then blending.
Is there a more efficient way to perform this operation? It seems uploading all this texture data every frame is just completely killing performance.
For the game I'm writing, I'm adding 100,000 trees, each of which is a merged geometry. When I add these from a cloned model using tree.clone(), I save tons of memory by doing this, but the game runs at 3 FPS because of the 100k geometries.
In order to get the game up to 60 FPS, I'd need to merge these trees together into a few geometries total. However, when I do this, chrome crashes due to using too much memory.
What's the reason for the extreme memory usage when merging these trees? Is it because I'm undoing the positives of using the .clone() function?
You need to look into instancing, your use case is exactly what the thing is made for.
Alternatively, using BufferGeometry instead of regular geometry, should be less memory intensive.
edit
What's actually eating your memory is the overhead of the merge operation when it operates on THREE.Geometry. The reason for this is that you have to allocate a ton of JS objects, like Vector3,Vector2, Face3, etc. Which then are discarded since the original geometry doesn't exist any more. This stresses everything, even if you didn't experience a crash, you'd experience slow down from garbage collection. The reason buffer geometry works better is because it's using typed arrays. For starters all your floats are floats not doubles, only primitives are being copied around, no object allocation etc etc.
You are eating more memory on the gpu though, since instead of storing one instance of geometry and referring to it in multiple draw calls, you now hold N instances within the same buffer (same data just repeated, and pre-transformed). This is where instancing would help. In summary you have:
Mesh (Node, Object)
Describes a 3d object within a "scene graph". Is it a child of some other node, does it have children, where its positioned (Translation), how its rotated, and is it scaled. This is the THREE.Object3D class. Extending from that is THREE.Mesh that references a geometry and a material, along with some other properties.
Geometry
Holds the geometry data, (which is actually referred to as "mesh" in modeling programs), file formats etc. hence the ambiguity. In your example that would be a single "tree":
describing how far away a leaf is from the root or how segmented is the trunk or branches (vertices)
where the leaves or trunk look up textures (UVS),
how it reacts to light (explicit normals , optional, but there are actually techniques for rendering foliage with overriden/modified/non-regular normals)
What's important to understand is that this stuff exists in "object (model) space". Let's say that a modeler modeled this, this would mean that he designated the object as "vertical" (the trunk goes up say Z axis, the ground is considered XY) thus giving it the initial Rotation, he put the root nicely at 0,0,0 thus giving it the initial Translation, by default we can assume that the Scale part is 1,1,1.
This tree can now be scattered around a 3d scene, be it in a modeling program or three.js. Let's say we import it into something like Blender. It would come in at the center of the world, at 0,0,0, rotated by 0,0,0, and at its own scale of 1,1,1. We realize that the modeler worked in inches, and our world in meters, so we scale it in all three directions by some constant. Now we realize that the tree is under ground, so we move it up by X units, until it sits on the terrain. But now its going through a house, so we move it sideways, or perhaps in all three directions because the house is on a hill and now it sits where we want it. We now observe that the silhouette is not aesthetically pleasing so we rotate it around the "vertical" axis by N degrees.
Let's observe now what happened. We haven't made a single clone of the tree, we scaled it, moved it around, and rotated it.We haven't modified the geometry in any way (adding leaves, branches, deleting something, changing uvs), it's still the same tree. If we say that the geometry is it's own entity, we have a scene graph node that has its TRS set. In context of three.js this is THREE.Mesh (inherits from Object3D), which has .rotation,.scale, position(translation) and finally .geometry (in your case a "tree").
Now lets say that we need a new tree for the forest. This tree is actually going to be the exact copy of the previous tree, but residing at a different place (T), rotated along the Z axis (R), and scaled non-uniformly (S). We only need a new node that has a different TRS, lets call it tree_02, it's using the same geometry, lets call it treeGeometryOption_1. Since tree geometry one has a defined set of UVS, it also has a corresponding texture. A texture goes into a material, and the material describes the properties, how shiny is the leaf, how dull is the trunk, is it using a normal map, does it have a color overlay, etc.
This means that you can have some sort of TreeMasterMaterial that sets these properties, and then have a treeOptionX_material corresponding to a geoemetry. Ie. if a leaf looks up the uvs in some range, the texture there should be green, and more shiny, then the range that trunk looks up.
Now lets reiterate the entire process. We imported the initial tree model, and gave it some scale, rotation and position. This is a node, with a geometry attached to it. We then made multiple copies of that node, which all refers to the same geometry TreeOption1. Since the geometry is the same, all of these clones can have the same material treeOption1_material which has it's own set of textures.
This was all a super lengthy explanation of why the clone code looks like this:
return
new this.constructor( //a new instance of Mesh
this.geometry , //with the sources geometry (reference)
this.material //with the sources material (reference)
).copy(this) //utility to copy other properties per class (Points, Mesh...)
The other answer is misleading and makes it sound like:
return
new this.constructor( //a new instance of Mesh
this.geometry.clone() , //this would be devastating for memory
this.material.clone() //this is actually sort of common to be done, but it would be done after the clone operation
).copy(this)
Let's say we want to tint the trees to have different colors, for example 3.
var materialOptions = [];
colorOptions.forEach( color =>{
var mOption = masterTreeMaterial.clone()
//var mOption = myImportedTreeMesh.material.clone(); //lets say youve loaded the mesh with a material
mOption.color.copy( color ); //generate three copies of the material with different color tints
materialOptions.push( mOption );
});
scatterTrees( myImportedTreeMesh , materialOptions );
//where you would have something like
var newTree = myImportedTreeMesh.clone(); //newTree has the same geometry, same material - the master one
newTree.material = someMaterialOption; //assign it a different material
//set the node TRS
newTree.position.copy( somePosition );
newTree.rotation.copy( someRotation );
newTree.scale.copy( someScale );
Now what happens is that this generates many many draw calls. For each tree to be drawn, a set of low level instructions need to be sent to set up the uniforms (matrices for TRS ), textures (when trees with different materials are drawn) and this produces overhead. If these are combined and the draw call number is reduced, the overhead is reduced, and webgl can handle transforming many vertices, so a low poly object can be drawn at 60fps thousands of times, but not with thousands of draw calls.
This is the cause of your 3 fps result.
Fancy optimizations aside, the brute force way would be to merge multiple trees into a single object. If we combine multiple THREE.Mesh nodes into one, we only have room for one TRS. What do we do with the thousands of individual trees scattered over the terrain, what happened to their TRSs? They get baked into geometry.
First of all, each node now requires a clone of the geometry, since the geometry is going to be modified. The first step is multiplying every vertex by the TRS matrix of that node. This is now a tree that does not sit at 0,0,0 and is no longer in inches, but sits somewhere at XYZ relative to the terrain, and is in meters.
After doing this a thousand times, these thousand individual tree geometries need to be merged into one. Easy peasy, it's just making a new geometry, and filling it's vertices, faces, and uvs with these thousand new geometries. You can imagine the overhead thats involved when these numbers are high, JS has its limitations, GC can be slow, and this is a lot of data, because it's 3d.
This should answer the question from the title. If we go in reverse, we can say that you had a game that was consuming lots of memory ( by having a model - geometry, of a "forest") but was running at 60fps. You used the clone method to save memory, by breaking the forest apart into individual trees, extracting the TRS at every root, and by using a single tree reference for each node.
Now the gpu holds only a low poly model of a tree, instead of holding a giant model of a forest. Memory saved. Draw call abundance.
FPS RIP.
How do both, reduce the number of draw calls, while rendering a model of a tree, not a forest?
By using a feature called instancing. Webgl allows for a special draw call to be issued. It uses an attribute buffer to set up multiple TRS information at the same time. 10000 nodes would have a buffer of 10000 TRS matrices, this is 16 floats, describing a single triangle with no uvs or normals takes 9, so you can see where this is going. You hold one instance of the tree geometry on the gpu, and instead of setting up thousands of draw calls, you set one with the data for all of them. If the objects are static, the overhead is minimal, since you'd set both buffers at once.
Three.js abstracts this quite nicely with THREE.InstancedBufferGeometry (or something like that).
One tree, many nodes:
T memory
N draw calls
One forest, single node:
T * N memory
1 draw call
One tree, instanced many times:
T + N memory (kind of, but it's mostly just T)
1 draw call
/edit
100k is quite a lot, three is probably better at handling this than before, but it used to tank your fps whenever you had more than 65536 vertices. I'm not sure off the top of my head, but it's now addressed by either breaking it up into multiple drawcalls internally, or the fact that webgl can address more than 2^16 vertices.
I save tons of memory by doing this, but the game runs at 3 FPS because of the 100k geometries.
You still have one geometry, you have 100k "nodes" that all point to the same instance of geometry. It's the overhead of the 100k draw calls that is slowing this down.
There are many ways to skin this cat.
The problem of memory has nothing to do with the usage or non-usage of the clone() function, which simply makes a separate copy of your object geometry and material.
The memory problem is due to the fact that the program is reaching the memory limit while merging.
The mergin process is an expensive process during setup (loop through all vertices, faces of both objects and create a third combined object). Plus, the merged geometry you are building is getting too large at a certain point to be processed and causes the browser to crash.
Have you tried to split the merged geometry into smaller chunks of 100 objects instead of using 1 merged object to store the 100K trees?
To optimize the code, I would also look at the tree mesh itself and see if you can reduce the number of vertices and faces.
For the usage of that many trees, you might want to consider using tricks and have only a few real mesh for the trees that are close to the camera and for the ones further away use a "waterdown" version (2-3 planes mesh that would need to be merged as well).
I have a demo where i used hundreds of cubes that have exactly same geometry and texture, example:
texture = THREE.ImageUtils.loadTexture ...
material = new THREE.MeshLambertMaterial( map: texture )
geometry = new THREE.BoxGeometry( 1, 1, 1 )
cubes = []
for i in [0..1000]
cubes.push new THREE.Mesh geometry, material
... on every frame
for cube in cubes
// do something with each cube
Once all the cubes are created i start moving them on the screen.
All of them have the same texture, same size, they just change position and rotation. The problem here is that when i start using many hundreds of cubes the computer starts to suffer to render it.
Is there any way i could tell Three.js / WebGL that all those objects are the same object, they are identical copies just in different positions ?
I ready something about BufferGeometry and Geometry2 being able to boost performance for this sort of situation but i'm not exactly sure what would be the best in this case.
Thank you
Is there any way i could tell Three.js / WebGL that all those objects are the
same object, they are identical copies just in different positions ?
Unfortunately there's nothing that can automatically determine and optimize rendercalls in that regard. That would be pretty awesome.
I read something about BufferGeometry and Geometry2 being able to boost performance for this sort of situation but i'm not exactly sure what would be the best in this case.
So, the details here is this: the normal THREE.Geometry-class three.js provides is built for developer-convenience, but is a bit far from how data is handled by WebGL. This is what the DirectGeometry (earlier called Geometry2) and BufferGeometry are for. A BufferGeometry is a representation of how WebGL expects data for drawcalls to be held: it contains a typed-array for every attribute of the geometry. The conversion from Geometry to BufferGeometry happens automatically every time geometry.verticesNeedsUpdate is set to true.
If you don't change any of the attributes, this conversion will happen once per geometry (of which you have 1) so this is completely ok and moving to a buffer-geometry won't help (simply because you are already using it).
The main problem you face with several hundred geometries is the number of drawcalls required to render the scene. Generally speaking, every instance of THREE.Mesh represents a single drawcall. And those drawcalls are expensive: A single drawcall that outputs hundred thousands of triangles is no problem at all, but a thousands of drawcalls with 100 triangles each will very quickly become a serious performance problem.
Now, there are different ways how the number of drawcalls can be reduced using three.js. The first is (as already mentioned in comments) to combine multiple meshes/geometries into a single one (in the end, meshes are just a collection of triangles, so there's no requirement that they form a single "body" or something like that). This isn't too practical in your case as this would involve applying the position and rotation of each of your cubes via JS and update the vertex-arrays accordingly on each frame.
What you are really looking for is a WebGL-feature called geometry instancing.
This is not as easy to use as regular meshes and geometries, but not too complicated either.
With instancing, you can create a huge amount of objects in a single drawcall. All of the rendered objects will share a single geometry (your cube-geometry with its vertices, normals and uv-coordinates). The instancing happens when you add special attributes named InstancedBufferAttribute that can contain independent values for each of the instances. So you could add two per-instance attributes for position and rotation (or a single instance transformation-matrix if you like).
These examples should pretty much be what you are looking for:
http://threejs.org/examples/?q=instancing
The only difficulty with instancing as of now is the material: you will need to provide a custom vertex-shader that knows how to apply your per-instance-attributes to the vertex-positions from the original geometry (this can also be seen in the code of the examples).
You have a webgl tag so Im going to give a non three js answer.
The best way to handle this is to allocate a float texture array made of model transform matrix data (or just vec3 positions if thats all you need). Then you allocate a mesh chunk containing all your cube data. You need to add an additional attribute which I refer to as modelTransform index. For each "cube instance" in the mesh chunk, write the correct modelTransform index value corresponding to the correct offset in the model transform data texture.
On each frame, you calculate the correct model transform data for all the cubes and write to the model transform data texture with correct offsets and such. Upload the texture to GPU on each frame.
In the vertex shader, access the model transform data from the modelTransform index attribute and the float texture. Rest is the same.
This is what I am using in my engine and it works well for smallish objects such as cubes. Note however, updating 150000 cubes on 60 FPS will likely take most of your CPU resources from JS. This is unavoidable regardless of which instancing scheme you take.
If the motion/animation of each cube is fixed, then a even better way to do it is to upload a velocity attribute and initial creation time stamp attribute for each cube instance. On each frame, send the current time as uniform and calculate the position as "pos += attr_velocity * getDeltaTime(attr_initTime, unif_currentTime);". This skips work on CPU all together and allows you to render a much higher number of cubes.
I'm trying to project massive, 32k-resolution equirectangular maps on spheres.
Since a 32k texture is hardly accessible for older graphics cards supporting 1k-2k sized textures, and scaling a 32k image to 1k loses a tremendous amount of detail, I've resolved to splitting each source map by projecting each into 6 cube faces to make up a cubemap, so that more detail can be displayed on older cards.
However, I'm having trouble actually displaying these cubemapped spheres with three.js. I can set the MeshPhongMaterial.envMap to my cubemap, but of course this makes the sphere mesh reflect the texture instead.
An acceptable result can be produced by using ShaderMaterial along with ShaderLib['cube'] to "fake" a skysphere of some sort. But this drops all ability for lighting, normal mapping and all the other handy (and pretty) things possible with MeshPhongMaterial.
If at all possible, I'd like to not have to write an entire shader from scratch for such a simple tweak (switching one texture2D call to textureCube). Is there a way to coerce three.js to read the diffuse term from a cubemap instead of a 2D texture, or a simple way to edit the shader three.js uses internally?
I have a world full of voxels, lets say that my world is 320*320*96 voxels. My idea is to load that entire world into the ram of my videocard so that no peformance is lost in transferring new "chunks" to the GPU. The amount of faces generated to display that voxelworld should easyly fit into the memory of modern graphic cards.
However, the problem I am facing now it how to not display parts of that world, I want to limit the view of this world to (for example) 128*128*96 and shift the world or the camera around to show different parts.
To demonstrate my problem, have a look at a (simple) scene consisting of a ground with in white the "viewable" area, I am looking for the right WebGL/three.js functions to restrict the view to just the white part.
You could remove voxels you don't want to display from the scene.
scene.remove( mesh )
And add them to the scene when you do want to display them.
scene.add( mesh )
I recommend splitting your voxel world into chunks (like Minecraft) and making those chunks into meshes individually. Add the chunk meshes that you want visible to the scene and remove them when you want to hide them.