New HDR demo

Discussion in 'Architecture and Products' started by Humus, Sep 6, 2005.

  1. Tim

    Tim
    Regular

    Joined:
    Mar 28, 2003
    Messages:
    875
    Likes Received:
    5
    Location:
    Denmark
    Strange it works in FireFox for me.
     
  2. Guden Oden

    Guden Oden Senior Member
    Legend

    Joined:
    Dec 20, 2003
    Messages:
    6,201
    Likes Received:
    91
    Ok, I updated drivers to the latest ones on NV's site, now instead of throwing up an empty window and complaining about no cable TV available ( ;) ), it simply doesn't do anything other than make a button on the taskbar and occupy one of the hardware threads of my CPU to 100%. No window, and NO GRAPHICS.

    I don't want to ridicule people of l33ter sk1llz than myself, but if windows games were as dodgy as Humus' demo coding, everybody would be up in arms! ;) It's becoming sort of a tradition of the new great demo that never runs properly on NV hardware... :lol: Of course, it might all be NV's fault though, I've no real clue.
     
  3. Skrying

    Skrying S K R Y I N G
    Veteran

    Joined:
    Jul 8, 2005
    Messages:
    4,815
    Likes Received:
    61
    Sounds like your own issue though. Every other 6x00 card seems to be running it now, but with that weird pixilation thing now. Though when that's not happening it does look pretty damn great.
     
  4. Humus

    Humus Crazy coder
    Veteran

    Joined:
    Feb 6, 2002
    Messages:
    3,217
    Likes Received:
    77
    Location:
    Stockholm, Sweden
    Well, that's life if you insist on using the latest and greatest features and only have one platform to test on. It's bound to cause issues. It's a good way to make people update their drivers though. :p
     
  5. Humus

    Humus Crazy coder
    Veteran

    Joined:
    Feb 6, 2002
    Messages:
    3,217
    Likes Received:
    77
    Location:
    Stockholm, Sweden
    AF is enabled by the demo itself, so there's no need. AA on the other hand is not available due to the fact that I'm using a floating point render target.
     
  6. Humus

    Humus Crazy coder
    Veteran

    Joined:
    Feb 6, 2002
    Messages:
    3,217
    Likes Received:
    77
    Location:
    Stockholm, Sweden
    Interestingly, the "pixels" are 4x4 tiles. Not sure what that tells me though, but it makes me think it's more likely to be a driver bug than an app problem.
     
  7. Moloch

    Moloch God of Wicked Games
    Veteran

    Joined:
    Jun 20, 2002
    Messages:
    2,981
    Likes Received:
    72
    Im just unsure why forced breaks it since I've never had that problem with far cry or the new splinter cell.
    I used to assume AF worked in every app untill I played richard burns really (not really related to your demo since you enable it in the app tho).
    Now I'm unsure :???:
     
  8. skirst

    Newcomer

    Joined:
    Sep 19, 2003
    Messages:
    5
    Likes Received:
    0
    Location:
    Cincinnati, Ohio USA
    For what it's worth, I face the same problem Humus does, but on the other end of the spectrum. I primarily develop and test on Nvidia hardware. It seems more often than not, when I implement new effects and features using the latest and greatest today's hardware has to offer, It'll "just work" on my Nvidia systems but fall flat on it's face on my only ATI system. Humus, for the love of god tell me ATI has fixed the glGenerateMipmapEXT bug and it'll be in the coming driver releases :)
     
  9. atlantis

    Newcomer

    Joined:
    Aug 13, 2004
    Messages:
    43
    Likes Received:
    1
    Location:
    Milano, Italy
    Works fine for me.
    9700pro + cat 5.8

    Thanks for the demo Humus.
    :)
     
  10. Guden Oden

    Guden Oden Senior Member
    Legend

    Joined:
    Dec 20, 2003
    Messages:
    6,201
    Likes Received:
    91
    Shouldn't a man of your salary be able to buy a second PC with a NV card in it to test things on? ;)

    Anyway, isn't it WEIRD that things aren't just automagically working in the 3D realm if you follow the rules and do stuff the way you're supposed to? Standard windows API function calls don't suddenly start to freak out if they run on AMD or VIA or Transmeta CPUs, but hop from ATi to NV and DX goes absolutely bananas...

    Hey... Now that you mention it, that IS true! Last two times I updated my drivers was because I was trying one of your new demos. ;)
     
  11. Jallen

    Newcomer

    Joined:
    Jun 11, 2002
    Messages:
    25
    Likes Received:
    0
    The blur render targets are getting converted to 8 bit (see RGBA16 in http://developer.nvidia.com/object/nv_ogl_texture_formats.html), try using RGBA16F.
     
  12. Humus

    Humus Crazy coder
    Veteran

    Joined:
    Feb 6, 2002
    Messages:
    3,217
    Likes Received:
    77
    Location:
    Stockholm, Sweden
    Alright, try again now and see if it's any better.
     
  13. Skrying

    Skrying S K R Y I N G
    Veteran

    Joined:
    Jul 8, 2005
    Messages:
    4,815
    Likes Received:
    61
    Perfect on my 6800nu now Humus, very beautiful!
     
  14. wireframe

    Veteran

    Joined:
    Jul 14, 2004
    Messages:
    1,347
    Likes Received:
    33
    For the most part it looks good. Perhaps not everyone will notice this is in the auto fly mode, but there are brief instances when rendering still fails. It can be easily spotted when you go into "free fly" mode. Picture included (6800 Ultra FW 77.77).

    [​IMG]

    Furthermore, is there a good reason why it is so slow on the 6800 Ultra? I see people using X800XT getting 120+ fps at 1280*960 with 4x AA and I am getting ~50fps at 1280x960 with no AA. Surely the 6800 isn't that inferior.
     
  15. DarN

    Regular

    Joined:
    Jan 19, 2004
    Messages:
    406
    Likes Received:
    1
    Location:
    Norway
    I think you can disregard the AA part, as it's not enabled, just selectable...
    Yes it's that inferior. :twisted:
     
  16. bloodbob

    bloodbob Trollipop
    Veteran

    Joined:
    May 23, 2003
    Messages:
    1,630
    Likes Received:
    27
    Location:
    Australia
    If humus is using FP16 textures for the blur on the 6800 ( which he is since the NV was converting INT16 requests to a INT8 texture ) then the reason this is most likely happening is in that case you have a infinity value in a texture. Generally +/-Inf is rather annoying for simple graphical rendering. We really need some special clamp options to clamp the textures at Max rather then infinite.

    Any idea whats causing the problems with ATI users? with the graphics corruption? Since I get it with a fresh latest driver install.
     
    #56 bloodbob, Sep 9, 2005
    Last edited by a moderator: Sep 9, 2005
  17. ChronoReverse

    Newcomer

    Joined:
    Apr 14, 2004
    Messages:
    245
    Likes Received:
    1
    [​IMG]


    Yeah, getting the same artifact here on my 6800LE.


    40ish FPS at 1280x960 =(
     
  18. JaylumX

    Newcomer

    Joined:
    Apr 24, 2005
    Messages:
    94
    Likes Received:
    1
    Location:
    Somewhere Else
    Well the HDR seems to be fixed now. Before when it shifted to high exposures it looked like it was polsterised. Now it okay. Nice one Humus.
     
  19. KimB

    Legend

    Joined:
    May 28, 2002
    Messages:
    12,928
    Likes Received:
    230
    Location:
    Seattle, WA
    The black screen is fixed by disabling FSAA/anisotropic (didn't check which, just disabled both...probably anisotropic).

    On my 6600 GT it usually looks good, but sometimes the high exposure scenes show some pretty bad banding around the sun. Might be nice to produce a FP16 version for comparison.
     
    bloodbob likes this.
  20. Humus

    Humus Crazy coder
    Veteran

    Joined:
    Feb 6, 2002
    Messages:
    3,217
    Likes Received:
    77
    Location:
    Stockholm, Sweden
    Good point, that's probably it. The skybox has a maximum value of 78600, while MAX_FLOAT16 is 65504 I think. Try changing the last line in skybox.shd to this and see if it goes away:

    gl_FragColor.rgb = min(sky.rgb, vec3(60000.0));

    I was just able to reproduce it myself. It seems to be resolution dependent, but couldn't figure out the exact pattern. Also, it seems to be fixed in recent driver builds, but not yet gone public.
     
Loading...

Share This Page

  • About Us

    Beyond3D has been around for over a decade and prides itself on being the best place on the web for in-depth, technically-driven discussion and analysis of 3D graphics hardware. If you love pixels and transistors, you've come to the right place!

    Beyond3D is proudly published by GPU Tools Ltd.
Loading...