viziblr

Even though I typically avoid building interactive diagrams, sometimes they are the only way to convey the point. This blog post by Amit is a great example. You can interactively explore the relationship between the driver’s head position, the three car mirrors, and the resulting blind spots.

http://amitp.blogspot.com/2011/08/interactive-diagrams.html

Here is a screenshot of the app, click to go directly to the interactive diagram

“Inkling bridges the gap between traditional, freehand sketching and digital development by capturing a digital likeness of a pen-on-paper sketch.”

http://www.wacom.com/en/Products/Inkling.aspx

I love my Wacom tablet and all but … meh. I can’t get excited about this. Maybe I’m missing something amazing but from what I read and see in the video below, I don’t see this going anywhere.. I’d much rather they focus on doing something interesting with their Cintiq pen display tablets.

Video below:

This blog post picks up from Part 1.

BACKGROUND

Previously we saw that the PNG encoder in .NET achieved on average a 1.3x increase in the size of a PNG file compared to the original Hypersnap PNG version. It’s not entirely clear though where the Hypersnap itself stands. To get a better sense of it, we’ll investigate using the PNGOUT tool – and assume that PNGOUT represents the “best case”,

I ran PNGOUT on the same collection of 11206 input files using this script:

PS> get-childitem . -include *.png -recurse | foreach ($_) { pngout "$_"}

First realization at the maximum optimization level (which is the PNGOUT default) each PNG file takes a very long time to analyze and encode. It takes about 1 MINUTE per PNG. So instead of converting all 11206 I let the script run overnight and only compared the files that were converted. So this accounts for 1450 files. So the sample set is smaller, but I think it contains a reasonable variety of screen captures.

HYPERSNAP ENCODER VS. PNGOUT ENCODER

Quick summary: ALL the files were smaller after PNGOUT was finished. On average PNGOUT encoded files were 85.7% of the Hypersnap-encoded files. The bar chart below shows the number of files that fell within a range of compression,. As you can see most the compression results do cluster in the 85% - 90% band.

OBSERVATIONS

When taken together with the data from the previous blog post we see the following

• PNGOUT can make a decent difference compared to Hypersnap – they are about 14% smaller after PNGOUT encoding
• On-demand application of the extreme PNGOUT compression on every save doesn’t seem practical – with an an encoding time of one minute per file -  it’s not a reasonable thing to do on every save.
• .NET Framework PNG files are about 1.25x bigger than Hypersnap-encoded version.
• .NET Framework PNG files are about 1.47x bigger than PNGOUT-encoded version
• Because we see that PNGOUT is so slow it seems clear that that Hypersnap PNG encoding is a very reasonable midway point. Whatever Hypersnap is doing is fast (there’ no evident slowdown when saving PNGs) and does provides a significant improvement over .NET
• Keep in mind that if you are transmitting a lot of PNG files over a network – the encoding will incur a size cost. If you are paying for bandwidth this might even be significant.

RAW DATA

USEFUL SOURCE CODE

The code below shows how to find how big a .NET Framework-encoded PNG will be – given a starting image. Instead of writing to the disk, an memory stream is used – an re-used. This keeps the memory usage small and avoid the I/O overhead of writing to the disk.

        public static long GetDotNetPNGSize(string filename, System.IO.MemoryStream memstream)         {             // resets the memory stream             // encodes as PNG             // and returns the number of bytes in the stream             using (var bmp0 = new System.Drawing.Bitmap(filename))             {                 memstream.SetLength(0);                 bmp0.Save(memstream, System.Drawing.Imaging.ImageFormat.Png);                 long memfilesize = memstream.Length;                 return memfilesize;             }         }

A video from the Blender Foundation – highlights from various artists

“Displayhack is a site for everyone participating in or enjoying real-time graphics, visual arts, motion graphics and the demoscene. Often adduced as the reason or motivation for expending effort toward a seemingly useless goal, the point being that the accomplished goal is a hack. As Louis Armstrong once said when asked to explain jazz: “Man, if you gotta ask you’ll never know.”

If you are into demoscene or just want to see cool videos keep track of this site.

http://www.displayhack.org/