James has a new post called http://prog21.dadgum.com/206.html in which he wonders if lots of the idioms like const, and static or sealed classes are them to allow developers to protect themselves from themselves. I would have commented on his blog, but he doesn’t like comments, which is a completely sane stance to have, heck the only comments I get here are spam.

So I had always thought those idioms where there to allow complier optimizations because there was extra information that was not tracked on older compilers, or to make the code cleaner.

const variables are to a avoid macro’s “true evil” and to keep the type information.
static allowed to keep the symbol table size down
sealed allowed real optimization, as functions will never to replaced.

And these were needed, so the new language could perform “better” at some key benchmark used to once again prove that poor “real world” problems in assembly, C are faster than in new language X. Thus all problems should use C/asm.

Well with the news Google is closing down Google Code, I’ve moved the two project I am responsible for to GitHub.

so https://code.google.com/p/coab/ is now at https://github.com/simeonpilgrim/coab

and https://code.google.com/p/nikon-firmware-tools/ is now at https://github.com/simeonpilgrim/nikon-firmware-tools

I’ve swapped over to the git tools, which I dislike compared to the hg tools, but I couldn’t get hg-git to play nice with github, so have moved on.

The other day on Google+ I read a post from Ana Andres talking about coding Tiny Planets, and how it was running really slow in Matlab somewhere in the pixel drawing code.

That got me thinking about how I would code it in C#, and Ana had a good blog post showing one way of thinking about how the transformation was happening , one of her goals was to make an animated GIF so people could see the wrap bend happening.

Now the this was plenty fast enough, I was getting 2-3 fps on my laptop using only one core, so from a speed perspective moved on.

My first plan was to interpolate between the points (all though that would really slow, and look bad at the outside), but when I showed my co-worker my method, he said I was silly and should just traverse the output space, and find the input pixel that maps, so I did that and got:

I then had to put the bend code back in, as that was a special aspect needed for the animation and got:

I then started playing with sub-sampling to make the output less jaggie (F), and then I decided blend in YUV (or YCrCb) colour space, and finally settled on five point YUV average where you have a center of the pixel and the four 1/4 towards the pixel diagonal corners. Giving:

For now the GIF’s at coverted to videos, but if they are to slow I will have to go back to like links

Now improvement on the process would be to allow scaling and translation of the origin on the polar focus, this might be useful in an interactive UI to allow exploring the space.

Here’s my source code for Step H, it was fun to learn to make GIFs in C#.

The full source is part of my Old School Graphics repo on GitHub

using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Drawing;
using System.Diagnostics;
using System.Windows.Media.Imaging;
using System.Windows;
using System.IO;

namespace TinyPlanet
{
    class Program
    {
        static void Main(string[] args)
        {
            Bitmap src = new Bitmap(@"C:\temp\tinyplanet-1.png");

            GifH(src);
        }

        static void GifH(Bitmap src)
        {
            GifBitmapEncoder gEnc = new GifBitmapEncoder();

            for (int i = 0; i <= 100; i++)
            {
                Bitmap dst = Bend_H(src, i);
                Debug.WriteLine(string.Format("frame {0}", i));

                var frame = System.Windows.Interop.Imaging.CreateBitmapSourceFromHBitmap(
                        dst.GetHbitmap(),
                        IntPtr.Zero,
                        Int32Rect.Empty,
                        BitmapSizeOptions.FromEmptyOptions());
                gEnc.Frames.Add(BitmapFrame.Create(frame));
            }

            gEnc.Save(new FileStream(@"C:\temp\tinyplanet-out_H.gif", FileMode.Create));
        }

        static Bitmap Bend_H(Bitmap src, int bend_i)
        {
            int src_half_width = src.Width / 2;
            int dst_width = (src.Height * 2) + src.Width;
            int dst_height = src.Height * 2;
            int dst_origin_x = dst_width / 2;
            int dst_origin_y = dst_height / 2;

            double bend = bend_i / 100.0; // turn to percentage;


            int bend_x = (int)(src_half_width * (1.0 - bend));
            double bent_pixels = src_half_width - bend_x;
            double final_ang_d = 180.0 * bend;
            double final_ang = final_ang_d * (Math.PI / 180.0);


            Bitmap dst = new Bitmap(dst_width, dst_height);
            Graphics g = Graphics.FromImage(dst);
            g.Clear(Color.White);
            g.DrawImage(dst, 0, 0, dst.Width, dst.Height);

            int bend_start_x = dst_origin_x - bend_x;
            int bend_end_x = dst_origin_x + bend_x;

            for (int x = 0; x < dst_width; x++)
            {
                int fix_x = (x < dst_origin_x ? bend_x : -bend_x);
                int ox = (x - bend_start_x) + (src_half_width - bend_x);

                for (int y = 0; y < dst_height; y++)
                {
                    if (x > bend_start_x &&
                        x < bend_end_x)
                    {
                        // rectilinear
                        if (y < src.Height)
                        {
                            dst.SetPixel(x, y, src.GetPixel(ox, y));
                        }
                    }
                    else
                    {
                        // map from output to input
                        int dx = x - dst_origin_x + fix_x;
                        int dy = y - dst_origin_y;

                        int p0 = ARGBAtPoint(dx - 0.25, dy - 0.25, final_ang, ref src, bent_pixels, fix_x);
                        int p1 = ARGBAtPoint(dx - 0.25, dy + 0.25, final_ang, ref src, bent_pixels, fix_x);
                        int p2 = ARGBAtPoint(dx + 0.25, dy - 0.25, final_ang, ref src, bent_pixels, fix_x);
                        int p3 = ARGBAtPoint(dx + 0.25, dy + 0.25, final_ang, ref src, bent_pixels, fix_x);
                        int p4 = ARGBAtPoint(dx, dy, final_ang, ref src, bent_pixels, fix_x);

                        int yy, cr, cb;
                        int yt = 0, crt = 0, cbt = 0;
                        YCbCrFromRGB(p0, out yy, out cr, out cb);
                        yt += yy; crt += cr; cbt += cb;
                        YCbCrFromRGB(p1, out yy, out cr, out cb);
                        yt += yy; crt += cr; cbt += cb;
                        YCbCrFromRGB(p2, out yy, out cr, out cb);
                        yt += yy; crt += cr; cbt += cb;
                        YCbCrFromRGB(p3, out yy, out cr, out cb);
                        yt += yy; crt += cr; cbt += cb;
                        YCbCrFromRGB(p4, out yy, out cr, out cb);
                        yt += yy; crt += cr; cbt += cb;

                        dst.SetPixel(x, y, Color.FromArgb(ARGBFromYCbCr(yt / 5, cbt / 5, crt / 5)));
                    }
                }
            }

            return dst;
        }

        static void YCbCrFromRGB(int RGB, out int Y, out int Cb, out int Cr)
        {
            int r = (RGB >> 16) & 0xFF;
            int g = (RGB >> 8) & 0xFF;
            int b = RGB & 0xFF;
            Y = (int)(0.299 * r + 0.587 * g + 0.114 * b);
            Cb = (int)(128 - 0.169 * r - 0.331 * g + 0.500 * b);
            Cr = (int)(128 + 0.500 * r - 0.419 * g - 0.081 * b);
        }

        static int ARGBFromYCbCr(int Y, int Cb, int Cr)
        {
            int r = Y + (int)((1.4f * (Cb - 128))+0.5);
            int g = Y + (int)((-0.343f * (Cr - 128) - 0.711f * (Cb - 128))+0.5);
            int b = Y + (int)((1.765f * (Cr - 128)) + 0.5);
            return (int)(0xFF000000 + (r << 16) + (g << 8) + (b));
        }


        static int ARGBAtPoint(double dx,
                double dy,
                double final_ang,
                ref Bitmap src,
                double bent_pixels,
                int fix_x)
        {
            double r = Math.Sqrt((dx * dx) + (dy * dy));
            double q = Math.Atan2(dy, dx);

            double pic_ang = q + (Math.PI / 2.0);
            double mod_ang = ((pic_ang + Math.PI) % (Math.PI * 2.0)) - Math.PI;

            if (Math.Abs(mod_ang) <= final_ang)
            {
                int dev_x = (int)((mod_ang / final_ang) * bent_pixels) - fix_x + (src.Width/2);
                int dev_y = (int)r;

                if (dev_x < src.Width && dev_x >= 0 &&
                    dev_y < src.Height)
                {
                    return src.GetPixel(dev_x, src.Height - dev_y - 1).ToArgb();
                }
            }

            return Color.White.ToArgb();
        }
    }

}

We are porting our MSVC/Win32 applications to Clang/GCC/Linux and have just spent the morning tracing why our unit tests fail.

void BadExampleCode()
{
    double a = NAN;
    ASSERT(!isnan(a));
}

Under MSVC and Clang all good, GCC asserts. We added printf’s and looked at the assembly and the code was hard coded to 0.

Some googling found 2006 posts stating GCC -ffast-math did odd things with isnan, and it’s still a presently reported issue

This came about because we are using Premake, and had the FloatFast flag set, because that’s how our MSVC projects were set, and we don’t want to change those builds, so for now we have tweaked the Premake code for this flag under GCC, as it doesn’t make sense that you would ever want isnan to be hard coded to zero, that’s really not fast maths at all.

We also put a #ifdef for ffast-math to break if this flag is present in the future.

And by we I mean Dave, and I just sat and talking it though with him.

Moons/Planets instead of countries, quite funny to see this morning:

After a couple of weeks of hard code tracing Coderat nailed the where/how of the bit rates for Nikon DSLR movies recording are set. The sad thing is the source of the bit rates is just a simple function that had been documented for a couple of years, the completeness of it’s effect was just not understood. Sigh.

But now we have patched in higher bit rates. To that end 34mbps and 54mbps have been tested on a D5100, and I’ve tested 34mbps, 49mbps and 64mbps on my D7000.

There are some D7000 video’s that can be reviewed if that’s your thing 1080 24fps 18mbps (default) 29MB 1080 24fps 30mbps (default) 47MB 1080 24fps 45mbps (default) 60MB

The online patch tool now lets to install these rates, on D5100 and D7000, and only at 1080p, but I’m sure more models/modes will follow as requested.

I had no record problems with 64mbps (really 60mbps as I had no sound being recorded), using both my slow Transcend Class 10 card, or my fast SanDisk Extreme Pro, which I purchased for these very tests back in 2012.

So if you want to test or talk about this come on over to the Nikon Hacker forums.

I’ve become very frustrated at my teenage children who spend way too much time watching video’s and using our monthly bandwidth all up. I’ve also got sick of turning the ADSL router off at the wall to signal annoyance.

So I found the great resource NPR resource for the Tompson TG582n, which is the same device as the Technicolor, it just name different.

That first page explains the basics of configuring the device, but to setup Web Access Control Schedule (TOD), I had to do a number of things so it worked.

I had to set the Router time, 8:07pm NZDT (UTC+13:00), as the system was thinking it was 2000

:system settime date=17/02/2014 time=20:07:00 rtc=enabled timezone\

What I really keen to workout is how to do total data usage, so I can cap each user. But the above gives me enough to be able to turn Internet on/off person, and I can tweak it to have permanent outage. And they know this, so we’ll see how behavior changes.

[2021-07-17 Another never posted post, did some quick digging on the web.arcine.org to find the relevant content]

I recently purchased the Rebecca & Fiona - Luminary Ones (Nause Remix) off iTunes Store as I really like the song, and that mix in particular. The song is a 7m 47s DJ friendly mix, thus has 2 minutes of intro and 2 minutes of outro, that are honestly boring.

So I have been day dreaming of making a new cut of the track with those minutes trimmed, so I jumped into iTunes to see the file type to work out what needs to be done to trim the song, and just for the laughs I look at the Options tab of the track, and low and behold, a ‘Start Time’ and ‘Stop Time’.

And the settings worked on my PC, so I synced my iPhone, and they work there also. Sorted!

Now I would like a fade out option, as the ending is abrupt, but there is not a nice breakdown to end on, and maybe I will trim the song externally, but 95% of my problem is solved, so most likely I won’t.

Just changed the layout a touch of the web tool trying to emphasize you don’t have to pay for the patch tool. To help with this added a Big ugly link to the Help and a FAQ.

So if you have Patching related questions, feel free to ask them here or on the FAQ page.

If there are any web designers that want to make the patch tool less ugly, or simpler etc, I’m very open to less ugly. Also open to better work flow etc.

Wow, I just read the Galaga No-Fire Cheat write-up, and that is one of the coolest things I read in ages. I love the fact it’s possible via Mame to dig though the code. I love that people care enough to do it, and I love the cracking open of an old school game cheat, to solve the how/why it occurred.

hat tip to Raymond for this gem.