Rich Geldreich

@richgel999

Entrepreneur at , Khronos member working on Basis. Ex-Valve/MS,Age1DE,3/Halo Wars/Portal2/DotA2/CS:GO.

Joined March 2011

Media

  1. Jul 9

    Rendez-vous in Montreal (1987) - one of the first examples of skeletal animation:

  2. Jun 19

    Played some Age of Empires Definitive Edition multiplayer for fun. MP is still pretty active and I got into a game quickly.

  3. Jun 19
    Replying to

    Andrew sent me his improved PVRTC encoding. So good I did a doubletake.

  4. Jun 18

    Here's a real-time encode to PVRTC (27.16 dB) and high quality BC1 (44.15).

    Show this thread
  5. Jun 18

    This artificial test image (from the Dava engine) is tough for PVRTC. Left is original, middle is my encoding (44.3 Y dB), right is PVRTexTool (37.89). Those artifacts on the 1 and 0 are so annoying.

    Show this thread
  6. Jun 13

    If you ever were sitting up late at night, wondering if a grayscale H ramp looked better in BC1 vs. PVRTC, well here you go. BC1=51 Y dB (left/first image), PVRTC=60.2 Y dB (clamp addressing).

    Show this thread
  7. Jun 13

    Torture testing PVRTC. First image is PVRTC and the rest are visualizations (low/high endpoints and 2-bit pixel modulations). I'm beating PVRTexTool for PSNR/SSIM but visually my results look worse to the eye.

  8. Jun 12

    If you're curious how this was encoded into PVRTC, here are visualizations of the low endpoints, high endpoints, 0-3 modulations, and whether each block used punchthrough (3 color) alpha or not. (I processed all these images using )

    Show this thread
  9. Jun 12

    Just achieved a 60.5 dB [0,255] H ramp with my experimental PVRTC encoder. Left is PVRTexTool best quality (46.8 Y dB), right is the new lookup table based algorithm (60.5 Y dB). This is the best H ramp I've ever seen in PVRTC 4bpp from any encoder.

    Show this thread
  10. Jun 1

    Was watching the old 90's sci-fi series "Earth: Final Conflict". It takes place in the 2020's, but it's loaded to the gills with bulky CRT monitors. You can tell the age of movie/TV productions by paying attention to the monitors used. Except for "2001" - Kubrick got it right.

  11. May 31
  12. May 31

    Real-time PVRTC1 encoding is an interesting problem. The first pass computes endpoints by computing per-block RGB bounds, the second computes modulation. I really wish colorB's blue channel had 5 bits (vs. 4) - 4 bits for blue leads to slightly blue tinged ramps and more banding.

    Show this thread
  13. May 30

    I added cascaded shadow mapping to CS:GO (which was a ridiculously hard thing to do in Source1). Players have found the best "Shadow Spots" on various maps, which gives them slight advantages. I think this is awesome:

  14. May 23
    Replying to

    I'm such a sucker for retro graphics!

  15. May 22

    Been watching the original Star Trek TV series remastered. It's surprisingly good although the writing quality varies a lot. The new FX are good and don't stand out too much. The FX are what could have been done in the late 80's.

  16. May 12

    Non-RDO BC7 encoding with random codec params, with more samples vs. my last tweet. The Pareto frontier is starting to become visible now. The test texture contains random blocks picked from thousands of textures in my corpus.

  17. May 12

    We're now feeding random codec configuration parameters into our non-RDO BC7 codec and plotting the results. Partially for testing, but also to find good settings to use for the various speed vs. quality profiles.

  18. May 12

    Our non-RDO BC7 encoder in Basis is now faster and higher quality than all other CPU BC7 encoders that I've tried (NVTT, Intel's ispc_texcomp, MS). I just put in several mode-6 only optimizations, so we're now 19% faster than ispc_texcomp's ultrafast profile at higher quality.

  19. May 12

    I'm optimizing mode 6 encoding in our vectorized BC7 encoder. Trying to optimize the output block packing function - clearly there's a problem.

    Show this thread
  20. May 11
    Show this thread

Loading seems to be taking a while.

Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.

    You may also like

    ·