48-bit tifs not visible

ueanyueany Posts: 0
edited December 1969 in Bryce Discussion

Hi there,

I've rendered a 48-bit TIF to disk. Viewing it in Photoshop CC 2014 or IrfanView or any other picture viewer gives me images that consist more or less only of a grid of points. This happens every time with 48-bit TIFs. 48-bit dithering is switched off. And no problems with 24-bit.

Any help is welcome!

pic.png
229 x 202 - 23K

Comments

  • ueanyueany Posts: 0
    edited December 1969

    Similar problem with 48-bit PNG, 48-bit EXR, and 48-bit HDR: They appear ALWAYS black in every pic viewer I have.

  • HoroHoro Posts: 10,637
    edited August 2014

    @ueany - 48-bit TIFF are integer values some programs can display, most cannot. 96-bit TIFF have floating point values. There is no such thing like an 48-bit HDR, only 96-bit. IrfanView cannot display 96-bit image data. You need Photoshop >= CS2 or HDRShop, or any high dynamic range image compatible software. 48-bit Dithering off is good. Also possible is that your images are corrupt because you render to disk. If rendering normal, you have to export TIFF/48/96 HDR-96, etc. before you do anything else once the render is finished. I never tested it with rendering to disk because I find it unreliable.
    EDIT: the free Picturenaut can also display 96-bit Files.

    Post edited by Horo on
  • ueanyueany Posts: 0
    edited December 1969

    For HDR Bryce at least gives you the possibility to save in 48-bit!

    XN View can also display 96-bit files, tried at least with TIF.

    Anyways, I found out that it is indeed a problem with Render to Disk, a BUG to say so.
    No problems with Export Image. :)

    Thank you

    PS: With what precision renders Bryce internally, 48 or 96 bit?

  • HoroHoro Posts: 10,637
    edited December 1969

    @ueany - that darn render to disk then. Good to know. Thanks for reporting back. I heard that Bryce renders 48 bit internally. This would mean that signed shorts are used (speaking C++). It is possible (though not confirmed) that floats are used which add up to 96 bit. With math co-processors, floats are not slower than integers.

Sign In or Register to comment.