Page 1 of 1
Please confirm voidground failure
Posted: 25 Feb 2015, 14:48
by enetheru
Hi,
I run intel(i know i know) and i wanted to confirm that anyone else running intel or whatever gets no voidground with the core platform map
I can only assume that its because of the recompress to ETC1 that happens
I just wanted to get another persons input before i chuck it in the wiki
it may not be obvious but the ramps are half seethrough
http://springfiles.com/spring/spring-maps/core-platform
Re: Please confirm voidground failure
Posted: 25 Feb 2015, 15:07
by Kloot
Shaders are disabled on intel hardware and the voidground effect relies on fragment alpha values being written by the map shader.
Re: Please confirm voidground failure
Posted: 25 Feb 2015, 15:12
by enetheru
Kloot wrote:Shaders are disabled on intel hardware and the voidground effect relies on fragment alpha values being written by the map shader.
that's a little weird because my Intel card supports DXT1 so I commented out the recompress code and it works fine.
Re: Please confirm voidground failure
Posted: 28 Feb 2015, 08:12
by jK
At least it doesn't report it does, cause spring checks for GL_EXT_texture_compression_s3tc and only when not available it recompress.
Also the used etc1 compressor doesn't support an alpha channel, neither does many drivers support etc1/2 in hw. So it's a pure _fallback_ solution to not show a white/black ground, when DXT isn't available.
Re: Please confirm voidground failure
Posted: 28 Feb 2015, 10:29
by Anarchid
I think it tries to recompress on a
different criterion:
Code: Select all
#if defined(USE_LIBSQUISH) && !defined(HEADLESS) && defined(GLEW_ARB_ES3_compatibility)
if (RecompressTiles(!GLEW_EXT_texture_compression_s3tc && GLEW_ARB_ES3_compatibility)) {
So the deciding thing is GLEW_ARB_ES3_compatibility. Idk what that is but it's different from GL_EXT_texture_compression_s3tc
Re: Please confirm voidground failure
Posted: 28 Feb 2015, 10:55
by jK
recheck your boolean understanding, it does what I said
and obv. it needs to check if ETC is supported, too (it's part of OES3)
Re: Please confirm voidground failure
Posted: 28 Feb 2015, 12:40
by enetheru
ok i understand now thanks :D
here's what my glxinfo reports
Code: Select all
enetheru@legit ~ $ glxinfo | grep dxt
GL_ANGLE_texture_compression_dxt3, GL_ANGLE_texture_compression_dxt5,
GL_EXT_texture_compression_dxt1, GL_EXT_texture_compression_rgtc,
GL_AMD_shader_trinary_minmax, GL_ANGLE_texture_compression_dxt3,
GL_ANGLE_texture_compression_dxt5, GL_APPLE_object_purgeable,
GL_EXT_texture_array, GL_EXT_texture_compression_dxt1,
enetheru@legit ~ $ glxinfo | grep s3tc
GL_OES_EGL_image, GL_OES_read_format, GL_S3_s3tc
GL_OES_read_format, GL_S3_s3tc, GL_SGIS_generate_mipmap,
and hence why it attempts to recompress to ext even when it doesnt need to.
stupid GL extensions are stupid
https://www.opengl.org/registry/specs/S3/s3tc.txt
https://www.opengl.org/registry/specs/E ... n_s3tc.txt
https://www.opengl.org/registry/specs/E ... n_dxt1.txt