@Juliean

I’m passing a set of points into a geometry shader, which produces triangles by using Marching Cubes.

My problem is that I can’t figure out how to do feedback to copy the triangle data to a CPU buffer. Do you know how to do this?

My prototype code actually only generates 2 triangles from an input buffer consisting of 1 point: https://github.com/sjhalayka/opengl_gs_transform_feedback​

Do you know how to fix this?

P.S. This is for an open source project, where the code is in the public domain: https://github.com/sjhalayka/julia4d3​

I am surprised that no one is talking about vcl-style ui, basically he idea is that you have a layer where you put a set of controls that are rendered and you can interact with them. This is how wyswig editors (visual editors where you put lets say an image on a form)

Basically you can implement something like this

ill make a simplified pseudo code.

Base Control class - BControl

{

vec2 position, vec2 size;

bool OnMouseDown(x,y) { if lets say in its rect return true }

Base Layer cass CLayer : public BControl { where it holds an array of BControl }

UIManager where you handle drawing and touch/mouse input

{

CLayer * active_layer;

draw() { through all activelayer cotrols }

void handle_input { through all activelayer controls test if i hit it(x,y) }

then in main game loop you just call Draw() and handle_input() which does pretty everything.

taby said:
My problem is that I can’t figure out how to do feedback to copy the triangle data to a CPU buffer. Do you know how to do this?

No, I don't think I know how do that specifically in OpenGL. I haven't had to do draw-feedback for readbacks on CPU so far, and I only know the very basic process in DirectX, but not at all in GL. Sorry

_WeirdCat_ said:
I am surprised that no one is talking about vcl-style ui,

Because thats not what was asked? From OPs question it seems he wanted to have some technical details, not how to organise the UI.

What would be the best way to implement these windows

Now someone smart explain me, what line above means, i fell that stupid that no one can measure now.

how is that you know what he meant and i don't, thats not a sarcasm etc. i'm just curious

We in our AAA toolbox use batching all the time regardless if it is just a button or a full blown game UI. The reason is as I wrote in your other quaestion, that GPU is faster in parallelizing a single draw call to a huge buffer than a lot of smaller draws. So what we have is that every part of the UI that is unlikely to change will be put into the same vertex buffers, while anything that could change is placed in it's own buffer. This solution is optimized, so if text is supposed to have a size limit, we just change the texture to something that is alpha blended to invisible instead of generating a new buffer all the time, this is much more efficient.

Anyways, it get's tricky if something causes a relayout of the UI, then you have to redraw everything.

Another interesting technique I recently use for drawing free text on the screen is that I have a font texture of equal sized letters which I attach to a shader. Another texture is attached to the sprite I finally want to render and a third texture defines my text buffer. This works because our Font is limited to 256 characters and so a character occupies a single byte in every cell of the texture. The remaining bytes are used to store information like position and color of the sentense I want to draw.

We use this to have a performant way to display rapid changing debug text in our engine