feler404 / ocvl-addon Goto Github PK

I tested the Video Sample with latest addon update, it is working fine everything is good. But if close Blender and reopen the BLEND, seems like the node tree is no longer streaming the video sample like before. Re-adding the node does not help.

Workaround for now is to delete the Video Sample node. Save the Blend. Reopen and re-add the Video Sample node to work...

While trying to activate the addon it gives an error: no module named 'ocvl_addon'
is this project abandonned? i would really love to see someone forking it and continue supporting it.

Hi Dawid,
thanks for your efforts. This tool could be very useful to me but, I couln't make it work.
First of all the Linux binary executable is missing:file (Blender package left, ocvl binary right):

I just tried to copy it into opencvl/bin. Here is the result:

It seems there is more missing.

When using an image sample node or video node (or any node with a preview) the preview image will appear next to it rather inside the node.
Here are three images taken in succession as I move the node from left to right in the node editor.

This preview moves as the nodes moves but not by the same amount so there is always a misalignment. I have a dual monitor setup so maybe it is reading the window size incorrectly?

I want to try Object Detect but seems like none of the node working, or am I missing something not sure if that keypoints etc is producing anything.

For example Simple Blob Detector node, I am not sure how this can actually work or should work but just it does not work on my computer?

1. Flags to show preview: NODE_PREVIEW, and optional NODE_CLASS_INPUT
   cmp_node_type_base(&ntype, CMP_NODE_R_LAYERS, "Render Layers", NODE_CLASS_INPUT, NODE_PREVIEW);
   https://github.com/behollister/blender2.8/blob/blender2.8/source/blender/nodes/composite/nodes/node_composite_image.c

2. Static type: bl_static_type
   TypeError: bpy_struct: item.attr = val: enum "23" not found in ('CUSTOM', 'UNDEFINED', 'FRAME', 'GROUP', 'GROUP_INPUT', 'GROUP_OUTPUT', 'REROUTE', 'VIEWER', 'RGB', 'VALUE', 'MIX_RGB', 'VALTORGB', 'RGBTOBW', 'NORMAL', 'CURVE_VEC', 'CURVE_RGB', 'ALPHAOVER', 'BLUR', )

Hi,
thanks for your great tool.
I tried it today for the first time but unfortunatly i wasnt able open the realsense depth camera in the windows beta (blender 2.92) version only the rgb camera is available.
If there would be a python node and a raw data stream i could extract the data by myself (its only a numpy array).
Maybe you could help me.
At the moment i use opencv and pyrealsense2 in pycharm to create a tracker with ids to send moments of each tracked contour as a dictionary via udp stream to blender.

Your tool is killer for realtime interaction like touchdesigner.

Traceback (most recent call last):
File "D:\blender\2.92\2.92\scripts\modules\addon_utils.py", line 382, in enable
mod.register()
File "C:\Users\VPS\AppData\Roaming\Blender Foundation\Blender\2.92\scripts\addons\ocvl-addon-master_init_.py", line 49, in register
from ocvl_addon import logger_conf
ModuleNotFoundError: No module named 'ocvl_addon'

I can see video streaming image coming from node, but it seems static unless I keep clicking manually on the node. Is this a normal behaviour?

I wanted to try using my own custom Python code and searching for Custom Node, but could not find it. Is it hidden somewhere?

Basically wanted to try something like OpenPose to make pose estimation for realtime motion capture.

hello,I meet a problems when install addon
location: :-1
Traceback (most recent call last):
File "/root/Blender/blender-280/2.80/scripts/startup/bl_ui/space_userpref.py", line 1809, in draw
split.label(text=info["author"], translate=False)
TypeError: UILayout.label(): error with keyword argument "text" - Function.text expected a string type, not tuple

and to

"author": (
    "Dawid Aniol, ",
    "OCVL team, ",
    "Teredo team",
),

I try to change as

"author": "Dawid Aniol,OCVL team,Teredo team",

then solve

I keep getting error when trying to process a simple contour output:

2019-08-09 11:19:54,737 [WARNING] ocvl.core.node_base: CV process problem: fn=<built-in function findContours>, kwargs={'image': array([[[ 71,  71,  71, 255],
        [ 71,  71,  71, 255],
        [ 72,  72,  72, 255],
        ...,
        [ 70,  70,  70, 255],
        [ 71,  71,  71, 255],
        [ 72,  72,  72, 255]],

       [[ 71,  71,  71, 255],
        [ 71,  71,  71, 255],
        [ 71,  71,  71, 255],
        ...,
        [ 71,  71,  71, 255],
        [ 70,  70,  70, 255],
        [ 72,  72,  72, 255]],

       [[ 71,  71,  71, 255],
        [ 71,  71,  71, 255],
        [ 71,  71,  71, 255],
        ...,
        [ 71,  71,  71, 255],
        [ 71,  71,  71, 255],
        [ 70,  70,  70, 255]],

       ...,

       [[ 71,  71,  71, 255],
        [ 71,  71,  71, 255],
        [ 71,  71,  71, 255],
        ...,
        [ 71,  71,  71, 255],
        [ 71,  71,  71, 255],
        [ 71,  71,  71, 255]],

       [[ 71,  71,  71, 255],
        [ 71,  71,  71, 255],
        [ 71,  71,  71, 255],
        ...,
        [ 71,  71,  71, 255],
        [ 71,  71,  71, 255],
        [ 71,  71,  71, 255]],

       [[ 70,  70,  70, 255],
        [ 71,  71,  71, 255],
        [ 71,  71,  71, 255],
        ...,
        [ 70,  70,  70, 255],
        [ 71,  71,  71, 255],
        [ 71,  71,  71, 255]]], dtype=uint8), 'mode': 3, 'method': 2, 'offset': (0, 0)}, self=<bpy_struct, OCVLfindContoursNode("findContours")>, exception=OpenCV(4.1.0) /Users/travis/build/skvark/opencv-python/opencv/modules/imgproc/src/contours.cpp:197: error: (-210:Unsupported format or combination of formats) [Start]FindContours supports only CV_8UC1 images when mode != CV_RETR_FLOODFILL otherwise supports CV_32SC1 images only in function 'cvStartFindContours_Impl'

Not sure how to explain this, but OCVL addon seems to be affecting Sverchok nodes not in the good way... it clashes and making some nodes not working.

I watched your video:
https://www.youtube.com/watch?v=DIlc5mhjYZE&t=50s

But the Cascade Classifier node is nowhere to be found in the latest version of the addon.

It doesn't matter what ksize_in you choose the out results is always the same: