what is the best image format which keeps the most color and lowest size? I normally use png but ive been told that bitmaps are much better but their size is ...

Best format for development is probably TGA. You get an independent alpha channel for transparencies.

Deployment format is usually the DDS formats (DXTn). Many card suport them natively. PNG is also a good choice.

Luck!
Guimo

Hey,

I'd have to put my vote to PNG/MNG. Or at least PNG. PNG pwns other static image formats IMO.

The only problem with MNG is that it's not particularly well supported [and difficult to extract individual sub images from, because of the flexibility inside the format].

PNGs will usually [almost always] beat equivelent .gif files on size, and will always [except in special, devised cases] beat equivelent bitmaps. They can be lossless [like bitmaps] (or not lossless), and may contain an alpha channel. Similarly, AFAIK, you can use almost any number of bits for colour channels [like 16 bit alpha, w00t], contain colour display information [like gamma settings, w00t], maintain layers [like psd] and easily compress the format.

In addition, they can be progressive across 2 dimensions [so unlike gifs, you can extract basic information more easily].

The additional plusses that MNG would get in my eyes is that a PNG is a one frame MNG animation, you can define complex animation [ie scrolling, movement, etc] and put JNG [basically jpeg compressed images?] subpictures into the animation stream.

Edit: Forgot to mention, if you're using Photoshop, keep in mind that the default exporter is horrible, size/qualitywise. But fortunately there are compression tools out there to reduce the size of your PNGs without changing anything [or anything significant]...

--CJM

Quote:Original post by nes8bit
I'm a little curious about this one...is DDS portable?

Effectively, yes. The compression algorithms might not be supported by a given OS, but they're certainly supported by the video card. Using a DDS file as a texture in, say, OpenGL is pretty much just a matter of reading in information from the header, then shoving large chunks of the remainder of the file into glCompressedTexImageARB.

Now, that's not to say that Microsoft won't do something crazy with the DDS format with future versions of DirectX. It's just like them to pull something like that. But current DDS files work jes' fine.

To answer the original question: I use PSD for working versions of all image assets, and export to DDS for use in the engine. IMHO, it's a bad idea to use anything but DDS for final data, since you don't have full control over the compression, and it's a bad idea to use DDS for working versions since you will experience progressive quality degradation.