Loading...
> >

Creating HDRI from a scene - how?

2»

Comments

  • second_technician_rimmer_9571136c47second_technician_rimmer_9571136c47 Posts: 481

    I tried rendering my canvas, making the render a 8000X4000; no matter what I do, when I reuse tyhat as an envioropnment map, when projected it remains blurry. Also, even befor that, the initial 360 render stops way before it's fin,ished and I get an awful lot if "unfinished dots" noise. Why does it stop so early?

  • CWR63CWR63 Posts: 19
    second_technician_rimmer_9571136c47 said:

    Sorry to necro this, but there's just one tiny detail I would need to know: how high on the Y axis should the camera be to match the standard level provided in commercial products?

    Just my 2 cents, but I usually try to match the spherical camera's position in space with the presumed render camera's xyz position, seeing as the only thing you can change as far as the environment is the rotation - I've been using them as backgrounds, and have forgotten this a couple of times to my frustration; especially since I favor such extreme angles in my overly dramatic compositions...

    Commercial photographers generally prefer chest height for interiors, so 4-5 feet?

  • brainmuffinbrainmuffin Posts: 1,204

    Maybe you will have better luck with this tutorial than I did:

     

  • DiscipleDisciple Posts: 95

    Anybody seen this happen, bright regions chopped to black in their EXRs?

    I just found this thread and followed directions, outputting a beauty canvas of something simple.  That's G8M on a shiny plane under the default environment dome, image level reduced to 1.0, spherical lens.  The top portion shows the 8-bit image from the render window.  Below is the 32-bit beauty EXR canvas downconverted in Photoshop.  Can it be that the renderer can fit the brights into an 8-bit image, but not into a 32-bit image?  That's like saying I can't fit the elephant into the barn, but I can fit him into my car.  Illogical.

    I was so hopeful to read that Studio could write an EXR, now I'm deflated.  What am I doing wrong?  Thanks for reading.

    Hallelujah!

    Disciple

    BadIK0001.jpg
    360 x 360 - 9K
  • AbnerKAbnerK Posts: 718
    Disciple said:

    Anybody seen this happen, bright regions chopped to black in their EXRs?

    I just found this thread and followed directions, outputting a beauty canvas of something simple.  That's G8M on a shiny plane under the default environment dome, image level reduced to 1.0, spherical lens.  The top portion shows the 8-bit image from the render window.  Below is the 32-bit beauty EXR canvas downconverted in Photoshop.  Can it be that the renderer can fit the brights into an 8-bit image, but not into a 32-bit image?  That's like saying I can't fit the elephant into the barn, but I can fit him into my car.  Illogical.

    I was so hopeful to read that Studio could write an EXR, now I'm deflated.  What am I doing wrong?  Thanks for reading.

    Hallelujah!

    Disciple

    That's weird, I haven't seen that. Have you repeated the step? I would try a 400x200px render and test it. Restart and test. Just to make sure it wasn't a glitch? I haven't done a lot of exr renders but, I haven't had any problems like that. That looks like when you tell Photoshop to show what's clipping. 

  • DiscipleDisciple Posts: 95
    edited November 2020

    I have now tried another render, Apartment Kitchen out-of-the-box.  The brights act the same in my Adobe tools PS and AE.  The first JPG shows the AE info regarding floating point pixel values as they range on and off the scale.  I haven't googled their meaning yet, but "overflow" seems a pretty safe bet.  The surviving pixels are heavily overdriven.  Scaling them down to 2% yields picture 2.  I output all the other types of canvas as well ("multipass rendering" in my education... thanks Leeloo), and every canvas with bright emitters behaved the same way.  Not cricket.

    My Lightwave opened the EXR and didn't choke on the brights.  This was strange, since the pixels came in at even more overdriven levels than Adobe's.  Some measured at 990 million percent of full white! surprise Scaled down by 0.0001, the image could produce the environment sphere reflected in the mirror ball image, some color tweaking required.  I also tried dropping the exposure down so low that even the lights looked black in the render window.  It didn't make a lick of difference to the levels on the canvas, still astronomical, still overflowing.

    Maybe I can come up with a way to tame and use this insanely inflated pixel data, but there's obviously something wrong with this picture.  I hope somebody can tell me what it is.

    Hallelujah!

    Disciple

    EXRoverflows01.jpg
    210 x 270 - 17K
    KitchenSphere02-Canvas1-Beauty (0;00;00;00).jpg
    800 x 400 - 122K
    KitchenSphere02.jpg
    640 x 480 - 159K
    Post edited by Disciple on
  • AbnerKAbnerK Posts: 718
    Disciple said:

    I have now tried another render, Apartment Kitchen out-of-the-box.  The brights act the same in my Adobe tools PS and AE.  The first JPG shows the AE info regarding floating point pixel values as they range on and off the scale.  I haven't googled their meaning yet, but "overflow" seems a pretty safe bet.  The surviving pixels are heavily overdriven.  Scaling them down to 2% yields picture 2.  I output all the other types of canvas as well ("multipass rendering" in my education... thanks Leeloo), and every canvas with bright emitters behaved the same way.  Not cricket.

    My Lightwave opened the EXR and didn't choke on the brights.  This was strange, since the pixels came in at even more overdriven levels than Adobe's.  Some measured at 990 million percent of full white! surprise Scaled down by 0.0001, the image could produce the environment sphere reflected in the mirror ball image, some color tweaking required.  I also tried dropping the exposure down so low that even the lights looked black in the render window.  It didn't make a lick of difference to the levels on the canvas, still astronomical, still overflowing.

    Maybe I can come up with a way to tame and use this insanely inflated pixel data, but there's obviously something wrong with this picture.  I hope somebody can tell me what it is.

    Hallelujah!

    Disciple

     

    Would you take a screen shot of your Tone Mapping settings in the Render Tab?

     

  • markusmaternmarkusmatern Posts: 559
    Disciple said:

    I have now tried another render, Apartment Kitchen out-of-the-box.  The brights act the same in my Adobe tools PS and AE.  The first JPG shows the AE info regarding floating point pixel values as they range on and off the scale.  I haven't googled their meaning yet, but "overflow" seems a pretty safe bet.  The surviving pixels are heavily overdriven.  Scaling them down to 2% yields picture 2.  I output all the other types of canvas as well ("multipass rendering" in my education... thanks Leeloo), and every canvas with bright emitters behaved the same way.  Not cricket.

    My Lightwave opened the EXR and didn't choke on the brights.  This was strange, since the pixels came in at even more overdriven levels than Adobe's.  Some measured at 990 million percent of full white! surprise Scaled down by 0.0001, the image could produce the environment sphere reflected in the mirror ball image, some color tweaking required.  I also tried dropping the exposure down so low that even the lights looked black in the render window.  It didn't make a lick of difference to the levels on the canvas, still astronomical, still overflowing.

    Maybe I can come up with a way to tame and use this insanely inflated pixel data, but there's obviously something wrong with this picture.  I hope somebody can tell me what it is.

    Hallelujah!

    Disciple

    Canvas renderings using IRAY in DAZ Studio are much too bright by default. I use Affinity Photo for post processing and usually lower the brightness of a canvas by 13 stops. I use Affinity Photo because IMHO the support for 32 Bit images is much better (all available tools can be used).

    You may also take a look in this thread about canvas rendering: https://www.daz3d.com/forums/discussion/comment/6217741/

  • DiscipleDisciple Posts: 95
    edited November 2020
    AbnerK said:

    Would you take a screen shot of your Tone Mapping settings in the Render Tab?

     

    Gladly. It's attached. These seem to have no effect on the canvas outputs, only upon the interactive and the final, which does make sense.  So, my nVIDIA and my NewTek can map the cosmic pixels down to something recognizable, but my (old) Adobes can not.  I wonder if I should be asking at nVIDIA instead of at DAZ?

    @markusmatern: Thank you.  I looked in and saw more good-sounding tips.  I'll try 'em Monday.

    Hallelujah!

    Disciple

    ToneMap002.jpg
    1242 x 679 - 145K
    Post edited by Disciple on
  • AbnerKAbnerK Posts: 718
    edited November 2020

    ...

     

    Post edited by AbnerK on
  • DiscipleDisciple Posts: 95
    edited November 2020

    Here in America I'm giving thanks that a tip from the Canvas Frustration thread is producing good results. smiley It involved finding every light emitter and scaling down all luminances by 0.0001, then opening up the exposure numbers for my render window's sake.  The pixel values in the EXRs eased down to reasonable numbers and my Adobe-ware no longer chokes.  This means that the crazy numbers have now migrated into the tonemapping settings, like 4 second shutter at f/2.0, but it works.

    Many thanks to all who replied and blessed holidays.

    Hallelujah!

    Disciple

    Post edited by Disciple on
  • MouserMouser Posts: 675

    You can export as .hdr format in gimp v2.10.

    Also there is an exposure blend script at Git hub (cant remember the link.)

  • EndlessdescentEndlessdescent Posts: 105

    Trying this today and not getting proper light sources. Dome shows up fine but light isn't emitted from the lights in the HDR scene?

Sign In or Register to comment.