Skip to content

Instantly share code, notes, and snippets.

View mrbid's full-sized avatar

James William Fletcher mrbid

View GitHub Profile
@mrbid
mrbid / esLuma.h
Created June 21, 2024 17:39
A simple helper library for OpenGL ES / WebGL applications that utilise Vertex Colors.
/*
--------------------------------------------------
James William Fletcher (github.com/mrbid)
June 2024 - esLuma.h v1.0
--------------------------------------------------
A pretty good color converter: https://www.easyrgb.com/en/convert.php
Lambertian fragment shaders make a difference, but only if you normalise the
vertNorm in the fragment shader. Most of the time you won't notice the difference.
@mrbid
mrbid / matvec.h
Created June 1, 2024 16:47
Portable Matrix & Vector Library.
#ifndef MATVEC_H
#define MATVEC_H
/*
James William Fletcher (github.com/mrbid)
September 2021 - February 2023
Portable floating-point Vec4 lib.
(Don't rely on this, copy and paste functions from this or write your own
@mrbid
mrbid / esAux7.h
Last active May 24, 2024 23:27
A simple helper library for OpenGL ES / WebGL applications.
/*
--------------------------------------------------
James William Fletcher (github.com/mrbid)
January 2024 - esAux7.h v7.0
--------------------------------------------------
A pretty good color converter: https://www.easyrgb.com/en/convert.php
Lambertian fragment shaders make a difference, but only if you normalise the
vertNorm in the fragment shader. Most of the time you won't notice the difference.
@mrbid
mrbid / fprop.c
Last active April 17, 2024 06:52
A custom brute force alternative to backpropagation.
/*
Test_User (notabug.org/Test_User)
April 2024
A custom brute force alternative to backpropagation for training of FFN/MLP forward-pass process.
Best generated weights from this process are available to download here:
https://raw.githubusercontent.com/mrbid/mrbid.github.io/main/fprop_best_weights_save
The network trains on the Zodiac dataset:
@mrbid
mrbid / facegen.py
Last active April 9, 2024 12:46
ThisPersonDoesNotExist.com Face Generator using TripoSR
# This code needs to be executed in root dir of TripoSR: https://github.com/VAST-AI-Research/TripoSR
# Run this using the modified marching cubes from github.com/thatname for 1.3-1.5x speed improvement
# https://github.com/VAST-AI-Research/TripoSR/issues/22#issuecomment-2010318709
# https://github.com/thatname/TripoSR/tree/main
# James William Fletcher (github.com/mrbid) (April 2024)
import os
import secrets
import requests
import numpy as np
import rembg
@mrbid
mrbid / rprop.c
Last active April 2, 2024 02:48
RPROP in C
/*
James William Fletcher (github.com/mrbid) & Test_User (notabug.org/Test_User)
April 2024
A sort of RPROP in C
Train network to output TRAIN_MAX when 333 is input and TRAIN_MIN otherwise
Check the last sign of the unit, if both signs are the same move in direction
of last sign if signs are different don't update weight just update sign.
@mrbid
mrbid / ptnpc.md
Last active May 6, 2024 11:05
Pre-Trained Neural Point Clouds (PT-NePC)

Pre-Trained Neural Point Clouds (PT-NePC)

Recalculate all objects origin, scale to unit sphere, project to Vertex Colors

  • bpy.ops.object.origin_set(type='GEOMETRY_ORIGIN', center='BOUNDS')
  • Normalise the scale in blender (ref).
  • The margin for a unit cube to fit in a unit sphere is to scale each component by 0.55 after -1 to +1 normalisation.
  • Convert to PLY with textures projected to Vertex Colors.

Possible filtering of meshes

@mrbid
mrbid / ffmpeg_screenshare.md
Last active April 5, 2024 00:57
FFmpeg Direct IP to IP screen sharing

Cast Screen

ffmpeg -f x11grab -i :0.0 -f mpegts -preset veryslow -r 60 udp://remote-host.com:1337
ffmpeg -f x11grab -i :0.0 -f mpegts -preset ultrafast -r 60 udp://remote-host.com:1337

ffmpeg -framerate 60 -f x11grab -i :0.0 -f mpegts -preset veryslow udp://remote-host.com:1337
ffmpeg -framerate 60 -f x11grab -i :0.0 -f mpegts -preset ultrafast udp://remote-host.com:1337

Receive Screen

@mrbid
mrbid / mrbid-itch-mirror.md
Last active April 8, 2024 00:30
[Itch.io Mirror] Share your favourite sources of free 3D content for games.
@mrbid
mrbid / text-to-3d_manual_setup.txt
Last active March 13, 2024 03:09
manually configuring a text-to-3d pipeline
Zero123plus (https://github.com/SUDO-AI-3D/zero123plus) output views are a fixed set of camera poses:
Azimuth (relative to input view): 30, 90, 150, 210, 270, 330.
Elevation (absolute): 30, -20, 30, -20, 30, -20.
To generate the images from Zero123++ it's easier to just use:
https://huggingface.co/spaces/sudo-ai/zero123plus-demo-space
and enable both background removals
StableSAM is usually used to remove backgrounds: