glumpy/chatroom

iPsych

@iPsych

view.camera.transform.matrix

[[   1.         0.         0.         0.     ]
 [   0.         1.         0.         0.     ]
 [   0.        -0.         1.         0.     ]
 [-474.40224 -308.97636  923.01759    1.     ]]

view.camera._projection.matrix

[[ 0.00011  0.       0.       0.     ]
 [ 0.       0.0001   0.       0.     ]
 [ 0.       0.      -0.       0.     ]
 [-0.      -0.      -0.       1.     ]]

Nicolas P. Rougier

@rougier

Ok, now I see. I guess you'll to dive into vispy code to find how the camera transform is computed. Generally, I use a MVP (Model , View, Projection) matrix but I'm not sure how Luke implemented it in the end. Maybe you n open an issue on vispy to ask fro some help.

Andrew Chalmers

@chalmersgit

Is this expected behaviour?
print(self.program['u_Texture']) # print out values
self.program['u_Texture'].delete()
print(self.program['u_Texture']) # still prints out values, i would expect something else (e.g., zeros, None type, or better yet that the handle doesn't exist anymore)

Nicolas P. Rougier

@rougier

@chalmersgit The deletion occurs on the GPU, no need to delete the CPU copy.

Andrew Chalmers

@chalmersgit

I guess the issue is that if later on in the application, if I want to load a new texture with different dimensions, it's remembered the old dimensions and won't let me do it.

(new texture with same variable name)

@rougier

Nicolas P. Rougier

@rougier

Ok, can you open an issue on Gtihub with a short example showign the problem?

David Pineda

@pineiden_gitlab

Hi! i found this module, but i think i need understand more how to create vertex and fragment shaders code, whre can i find more books or tutorials about?

Nicolas P. Rougier

@rougier

Here: https://www.labri.fr/perso/nrougier/python-opengl/

David Pineda

@pineiden_gitlab

Thanks Nicolás!

Si a beautifull book!

HukamiEstia

@HukamiEstia

Hi, I want to draw multiple 3D boxes with different translation and rotation, to do that I started from the outlined cube example and I am trying to use transforms to specify the translation and rotation, I was able to use the Rotate transform but not the Translate. Also is it possible to combine multiple transform (rotate the box then translate it) ?

here is the main from my vertex shader :

gl_Position = u_projection * u_view * u_model * vec4(<transform(position)>, 1.0);

and here is how I define the transform:

cube["transform"] = Translate(translate=(0,0,0))

which doesn't work, but if I use :

cube["transform"] = Rotate(angle=30)

it works... I must be missing something but I'm still a beginner concerning OpenGL stuffs

HukamiEstia

@HukamiEstia

Ok so it was because of a syntax error in translate.glsl which was corrected 25days ago in commit 580a1e8 but isn't in the pip install yet. Now it's working.

Also, I managed to combine multiple transforms:

vertex = """
attribute vec3 position;
vec4 rotation;
vec4 translation;
void main()
{
rotation = vec4( <rotate(position)>, 1.0);
translation = <translate>;
gl_Position = <transform>;
}
"""

cube["rotate"] = Rotate(angle=30)
cube["translate"] = Translate("rotation", translate=(0.5,0,0))
cube["transform"] = PVMProjection("translation")

HukamiEstia

@HukamiEstia

If there is a better way please let me know

jcyc

@jcyc

Hi,I use ten data csv file(each 3 col*2,000,000 row) with glumpy path or polygons function to drawing ,but I find out path or polygons function append one data array need more 10 secs ,I had trans csv to numpy array before append data, is possible any way to speed up 'append ' process ?ex: make file before then load ? or other ways, pls give me some direction, many thanks.

Lucas Menicucci

@lsmenicucci

Hey guys, could you give me a little help? I'm leaning the basics of OpenGL for making animated quiver plots and I can't understand very well the vertex and fragment shaders of this example https://www.labri.fr/perso/nrougier/python-opengl/#id37. In that example only 4 vertexes can create a bunch of arrows, I can't understand how that's possible. Furthermore, wouldn't be necessary at least 4N floats for drawing N 2D arrows?

Julio Vera

@nekonekno

Hey everybody! i'm trying to run the examples of the main page, without success. I have an Arch Linux installation with python 3.10 and Jupyter, and the output send me the following error:
object of type 'NoneType' has no len()
Any suggestions? Thank you!

QiuJiangkun

@qiujiangkun

Is there any follow up with newer version of numpy?

I have fixed some minor errors manually

Benoit Tachet

@benoittachet

Capture d’écran de 2020-12-02 09-22-43.png

Hi,

I'm trying to use glumpy to display some figures using transparency, and I am wondering about one thing that I discovered using one of the examples :
I'm currently using the cube example (solid-cube.py), and I wanted to change the alpha of its color. However, doing so makes the faces sometimes look different according to the rotation angle, like on this image :

Would anyone have an idea about why this does happen ?

Le minaw

@LeMinaw

Would anyone have an idea about why this does happen ?

Hard to say without manipulating the actual program
Note you have some exemples of OIT (order independant transparency) implementation there: https://github.com/glumpy/glumpy/tree/master/examples/transparency

Peter Vince

@pvince99_twitter

I'm trying to draw several objects, all of the same type, each with rotation and translation and specified colour. Should I use Collections as I can't seem to get it working otherwise? Are there examples to look at?

glumpyfan101

@glumpyfan101

Hi! First of all: fantastic project and excellent book/tutorial. I have tried to use this module for some time, my end goal is to use it for visualizing CFD-calculations. Anyways: Right now I am trying to learn about buffering data to the GPU. I find the documentation a little sparse. The goal with the above code is to write a program that gives the GPU some data which the GPU then performs some sort of calculation on, after which I reach into the buffered memory and get the calculated data out. It runs without crashing, but does not seem to return correct values. Anyone able to give me a hint as to what is wrong? I am obviously an amature at this, and there is probably something conceptually really wrong here.

Nicolas P. Rougier

@rougier

To read data from the GPU, you'll need to use a framebuffer. There are some examples in the source directory.

Alex Sizov

@SizovAlex_gitlab

Hello
I need to visualization a 3D numpy array with a size of 900x900x100. The value of the array elements is 0 ... 255. Can someone share an example for this type of visualization?

venkatbalaji221

@venkatbalaji221

I am trying to create a basic "Hello world Triangle" using glumpy. I am getting a triangle when vertices are only among four corner points of the frame and in other cases, the entire screen is getting cleared(Insted of getting a small triangle inside the frame). what changes needed to be done? (Even not working for quad drawing given in documentation)

PauliX51

@PauliX51

Hey, I am trying to create a visualization where I would also like to add some labels using texts and fonts as described here: https://glumpy.readthedocs.io/en/latest/examples-text.html. The problem that I have is, that the text is only displayed once after I opened a new terminal. So what actually happens is that everything works fine the first time I run the program but if I close the window and re-run the program in the same terminal, the labels won't be displayed anymore. What I have to do is opening a new terminal and run the program there so that the labels are displayed again. Is there any way to avoid this? Thanks in advance!

pumplerod

@pumplerod

Hello, I came across glumpy because I'm trying to convert samples from a data set into images. I have been using matplotlib in my test however it is far too slow to perform on millions of samples. Glumpy seems promising. I'm never actually viewing the images created, I'm feeding them into a neural network for feature extraction with convolution, so what I really need to do is get the image data from the framebuffer to pass into the neural net. I can't find, in the documentation, a method for extracting the framebuffer data. Am I incorrect in thinking glumpy would be suitable for this purpose?

Hiro Protagonist

@hiroprotagonist84:matrix.org

[m]

I'm looking for an example where one inherits from app.Window instead of using the @window.* decorators.

I tried this, and my on_draw is never called:

class TestWindow(app.Window):
    def __init__(self,  *args, **kwargs):
        print(args, kwargs)
        super().__init__(*args, **kwargs)
        print(self)
        console = app.Console(rows=10, cols=80, scale=4, color = (1.0,1.0,1.0,1) )
        console._default_foreground = 1,0,1,1
        console._default_background = 0,1,0,1
        self.console = console
        self.attach(console)

    def on_draw(self, dt):
        print("on_draw")
        try:
            window = self
            window.clear()
            console = self.console
            console.clear()
            console.write("hello world")
            console.draw()
            print("here")
        except:
            logger.exception("draw")

Dhawal S

@dhawal1939

Hey, I am trying to write the contents of the CUDA buffer onto the torch tensor based on the example given here.
The example is very insightful but does the task in reverse. I have altered the
Functions set_src_device -> set_src_array and set_dst_array -> set_dst_array
Unfortunately, I am unable to retrieve the data correctly. I have torch tensor of type 'f4' and texture also created with type 'f4' instead of 'uint8' as given in the example.
https://gist.github.com/victor-shepardson/5b3d3087dc2b4817b9bffdb8e87a57c4

Can anyone let me know why?!

Nicolas P. Rougier

@rougier

Best would be top open an isue on GitHub with a precse decription of the problem/output

Dhawal S

@dhawal1939

Where communities thrive

People

Repo info

Activity