Information & Tips for making your Background Image
Kindly researched and explained by "Cato"
At http://www.nerd-out.com/darrenk/Backgrounds/Background.htm are the instructions for creating Hiteker background images.
The purpose of this exercise was to explore further the nature of those images to determine the ideal method of creating them.
Summary Conclusion:
The Apex/Hiteker background image is essentially a jpg image. It's specific format is an MPEG2 Still Frame which is pretty much the same thing.
When creating the initial bitmap, you need to have an image resolution of 704x480 pixels.
Note however that on an Australian PAL output to a Sony TV, the top and bottom 40 pixels are not visible, neither are the left and right 30 pixels. So work inside that space, and make that border area a single flat colour.
General Tips
Black and White images do not provide much benefit over colour images.
"Blurring" is much more friendly to JPG/MPG compression than lots of straight edges, so photos are good images to use.
If you do need to use straight edges, continuous lengths of horizontal colour are better than vertical colour because one aspect of the compression is "Run Length Encoding".
Make a border around your image 40 pixels high on the top and bottom edges, and 30 pixels wide on the left and right edges. Fill this border with a single colour. This will reduce your file size, and this area of the image is not visible (at least on a PAL output :-)
Only put as much on your image as you need to display - simplify the image as much as possible. e.g. instead of showing a whole photograph, show the bits you need and surround the rest of it with whitespace.
To view your final output file rename your .bin file to a .mpg file and execute it. In Windows, Media Player will start and show you the file. If you have problems with Windows Media player, you can also try opening it with IrfanView.
Stop here if you don't want to read about the actual experiments.
Experiments and Tests
I created 9 images.
The first was a complex image with many coloured elements, a black and white photo, and a complex patterned background.
Each subsequent image was less and less complex, with the final image being a black and white photo, and a blue horizontal bar.
Bit Depth
Part 1 - 24 bits
----------------
The first set of images were full 24 bit colour images, saved as 24 bit Windows Bitmaps.
From each image I generated 3 .bin files using compression levels of 29, 30, and 31 with mklogo.
Generally, the least complex images produced smaller .bin files than the more complex images.
However, in several cases, the reverse was true. This is a side effect of JPG/MPG compression.
Part 2 - 8 bits
---------------
Using Paint Shop Pro, I reduced each image to an 8 bit colour image and resaved it. This reduced the base file size of each bitmap from 991k to 332k.
I ran the file creation batch again to produce 3 compression levels of each output file.
Each .bin file produced from a 256 colour (8 bit) bitmpap was AT LEAST 100 bytes larger than the file produced from its corresponding 24 bit bitmap.
Conclusion: Reducing the number of colours in the source bitmap, prior to saving it, does not reduce the final .bin size. It consistently appears to increase the final file size.
BLUR
On advice from Darren King, I then blurred each 24 bit image using the standard "blur" function in Paint Shop Pro.
I then ran the batch routine again. Each .bin file was consistently smaller than its unblurred counterpart, in one case a blurred image was 78% of the size of the matching unblurred file.
This may appear unusual because blurring an image often increases the complexity of the image, however jpeg compression treats soft edges a lot better than hard edges.
I then blurred each of the 8 bit images. The resulting files were smaller than their unblurred 8 bit counterparts.
However the blurred 8 bit files were all at least 100 bytes larger than their blurred 24 bit counterparts.
BLUR vs Gaussian Blur
OK, so blurring appears to be a good thing. So I now blurred each image using a Gaussian blur of 0.8 and compared the images. A Gaussian blur of 0.8 is less of a blur than the standard Paint Shop Pro "blur".
The Gaussian blurred pictures were smaller than the unblurred pictures. The fully "blurred" pictures were smaller again.
JPG Comparison
To compare, I took the unblurred, gaussian, and blurred version of my first bitmap, and converted them to jpeg's using a Paint Shop Pro compression of 70.
The resulting files sizes of the jpg files were consistently smaller from unblurred, to gaussian, to blurred.
Conclusion:
From previous information we know that the program mklogo produces an MPEG2 Still Frame. To people who know nothing about MPEG, imagine it as a type of intermediary jpg file. The program c2bin converts the intermediate file into a final jpg file of a format that the Hiteker is looking for (i.e. mpeg2)
You may CLOSE this window now.