View Full Version : GL_COMPRESSED_LUMINANCE
04-18-2002, 12:38 AM
What is this Token ? Is there compressed luminance process ?
THX for help
04-18-2002, 04:22 AM
Originally posted by ToF_Fou:
What is this Token ?0x84EA?
Is there compressed luminance process ?This is just a forward looking thing. One day, there might be a suitable compressed format, in the meantime this probably maps to COMPRESSED_RGB_S3TC_DXT1.
The texture compression extension (the general one, not the ones describing a specific compression method) does not define any compression methods, and the driver is free to choose the one it feels is best. Since there are no suitable compressed formats for luminace texture today, you will get an uncompressed format. COMPRESSED_RGB_S3TC_DXT1 is for RGB textures, as the name says.
04-18-2002, 10:06 AM
A Luminance texture is just an RGB texture where R==G==B. So, it is perfectly reasonable to S3TC a luminance texture.
04-18-2002, 01:57 PM
DXT1 quantizes each component to 5 bits (6 for green), and then uses 4 bits per pixel to store the actual texture data, leading to possibly more quantization based on the block. Not really that great. And the extra bit of green means you don't get absolute gray, so I wouldn't say it's all that reasonable to use S3TC for luminance, compared to getting full quality at 8 bits per pixel.
Now, if you could separate out the alpha compression of DTX5 and use for alpha-only or luminance-only textures, that would probably rock!
I know luminance is R=G=B (and A=1), but then you need to convert your image to RGB format. What I know of, and what I meant to say, is that there is no format to compress luminace texture data, as a single channel.
I'm not sure if I tried it or not, but my GF2MX couldn't give me a compressed format if I tried to upload a luminance texture, with internal format GL_COMPRESSED_LUMINANCE.
04-18-2002, 11:48 PM
Originally posted by jwatte:
DXT1 quantizes each component to 5 bits (6 for green)Wrong. It doesn't. Read the specs. DXT1 can reach better than 565 resolution inside of blocks and it usually does so where it counts, in fine gradients.
[This message has been edited by zeckensack (edited 04-19-2002).]
Powered by vBulletin® Version 4.2.2 Copyright © 2014 vBulletin Solutions, Inc. All rights reserved.