Loading ...
Sorry, an error occurred while loading the content.

Re: Background Stars in Maya

Expand Messages
  • Drew Gilmore
    It occurs to me that there are producers like you guys upthread who know what you re talking about regarding generating starfields, and then a goodly number of
    Message 1 of 14 , Mar 21, 2011
    • 0 Attachment
      It occurs to me that there are producers like you guys upthread who know
      what you're talking about regarding generating starfields, and then a goodly
      number of smaller producers who don't, but still need good looking stars. To
      me this sounds like an opportunity for someone.

      This has been a subtle hint. We now return you to your thread. :p

      Drew
      --
      Drew Gilmore
      Asst. Director
      Sudekum Planetarium
      www.SudekumPlanetarium.com

      Just a guy made of dots and lines.
    • Mark Aldred
      I also have a scripted database import into 3dsMax. I position the stars on a sphere and use apparent magnitude to adjust size. Then I output six ninety degree
      Message 2 of 14 , Mar 21, 2011
      • 0 Attachment
        I also have a scripted database import into 3dsMax. I position the stars
        on a sphere and use apparent magnitude to adjust size. Then I output
        six ninety degree images and make a cube like Tom. I'm not familiar with Maya, but you'll also be able to make a random alien star map cube using this method. Scatter objects across the surface of a sphere, or use a particle system for your stars.

        The problem with anti-aliasing doesn't really arise when you output
        the six maps - you can hike up the AA min/max settings as high as you
        want since it's just six images you need. But when you have your cube
        and try to render an animation, that's when render time becomes
        unacceptable if you choose settings to prevent the smallest stars
        flickering. My solution was this - to render at 4k. Anti-aliasing works
        on absolute pixel values. The bigger the image, the more pixels
        represent each star. You can have a star that's 1 pixel wide at 2k that
        flickers horribly when animated at low AA settings. But the same star at
        4k will be represented by more pixels and is more likely to be picked
        up by the renderer even at low AA settings. And the best thing is,
        rendering a 4k star map at low settings takes much less time that a 2k
        image at high settings. Much much less time in fact. Once you get the
        balance between the AA settings and smallest star size sorted, there's
        no more flicker to
        worry about.

        Mark

        www.starlightstudio.jp

        ===

        To: fulldome@yahoogroups.com
        From: nuclearsugar@...
        Date: Wed, 16 Mar 2011 21:09:38 +0000
        Subject: [fulldome] Background Stars in Maya

        We have been experimenting with new techniques for our star backgrounds.

        Our typical setup includes a camera with a fisheye lens shader. We have a sphere with stars upon the inner side. This star-sphere is point constrained to the camera; so it translates with the camera but does not rotate. The star-sphere is always the largest thing because it needs to be in the background of everything.

        We take care of flying through stars by creating sprites.

        We were able to get over the hurdle of the poles pinching by creating custom UV maps. But we are not satisfied with our file texture itself. But we cannot find another software that will output fisheye or exact hemi-cube renders of stars. Plus we need alien constellations star setups. So far, Digital Sky has been our only option.

        We are experimenting with Maas Digital StarPro, a Lightwave plugin, but we are not excited about adding another step and toolset in our pipeline. Plus, exporting camera paths may prove to be tricky.

        I am hoping to hear some of the techniques others are using in 3D toolsets such as Maya. Do you have experience with this problem?
      • Tom Casey
        I just love hints. I ll tell you what... I will meet you halfway and do the harder part to get you started. I will provide a pre-formatted text file with a
        Message 3 of 14 , Mar 21, 2011
        • 0 Attachment
          I just love hints. I'll tell you what... I will meet you halfway and
          do the harder part to get you started.

          I will provide a pre-formatted text file with a numbered array of
          ~46,000 star positions, ra and dec already translated into decimal
          degree rotational values and simple decimal magnitudes. Depending on
          what 3D software application you use, you will be able to use this
          data and position stars using xy or z rotation values (depends on
          your programs orientation) and then create some method to vary the
          size/brightness of each star to get a full sphere of accurate stars.
          Then you can render textures for a cubic approach or any other
          approach you can come up with.

          And then, no matter which 3D software application you use, it might
          be nice if you made your solution available to the Fulldome group. I
          would love to see a more current way of doing this than the way we
          solved it years ago. Here's where you can find the data file...

          http://www.homerunpictures.com/fulldome/


          On Mar 21, 2011, at 10:33 AM, Drew Gilmore wrote:

          This has been a subtle hint.


          ************************************************
          H o m e R u n P i c t u r e s

          Tom Casey
          President & Creative Director

          100 First Avenue - Suite 450
          Pittsburgh, PA 15222
          412-391-8200
          mailto:tom@...
          http://www.homerunpictures.com
        • Drew Gilmore
          Many thanks Tom, I look forward to giving it a shot... I have much to learn, but If I get to a usable solution I d happily make it available for all - sharing
          Message 4 of 14 , Mar 21, 2011
          • 0 Attachment
            Many thanks Tom, I look forward to giving it a shot... I have much to learn,
            but If I get to a usable solution I'd happily make it available for all -
            sharing in this community should always be encouraged.

            (On the other hand though, it is possible to create an entire fulldome show
            while staying within an After Effects workflow... except for the stars. I
            haven't done any sort of survey but I wouldn't be surprised if some folks
            would appreciate the ability to purchase just the final product, in the form
            of a cubic projection at several popular resolutions. That is more where my
            hinting was directed.)

            Drew

            On Mon, Mar 21, 2011 at 3:38 PM, Tom Casey <tom@...> wrote:

            > I just love hints. I'll tell you what... I will meet you halfway and
            > do the harder part to get you started.

            > I will provide a pre-formatted text file with a numbered array of
            > ~46,000 star positions, ra and dec already translated into decimal
            > degree rotational values and simple decimal magnitudes. Depending on
            > what 3D software application you use, you will be able to use this
            > data and position stars using xy or z rotation values (depends on
            > your programs orientation) and then create some method to vary the
            > size/brightness of each star to get a full sphere of accurate stars.
            > Then you can render textures for a cubic approach or any other
            > approach you can come up with.

            > And then, no matter which 3D software application you use, it might
            > be nice if you made your solution available to the Fulldome group. I
            > would love to see a more current way of doing this than the way we
            > solved it years ago. Here's where you can find the data file...

            > http://www.homerunpictures.com/fulldome/

            > On Mar 21, 2011, at 10:33 AM, Drew Gilmore wrote:

            > This has been a subtle hint.
          • Jason Fletcher
            Tom, Wow. Thanks for taking time to describe your process. It is quite refreshing. In the past, we had considered creating a MEL script to use a star dataset
            Message 5 of 14 , Mar 21, 2011
            • 0 Attachment
              Tom,
              Wow. Thanks for taking time to describe your process. It is quite refreshing. In the past, we had considered creating a MEL script to use a star dataset and create nurb-circles upon a parent-sphere. But your technique much more elegant than we had envisioned. It is incredible of you to share the star database! Hot damn. If we get something working, I'll be sure to share our MEL gui since you have made your database public.

              When you use a star-textured-cube enclosing the scene, doesn't the perspective of the camera cause some stars to look distorted? Such as the stars in the corner? Perhaps I'm missing something.

              On a different note, I thought it would be helpful to share how I fix pinched poles for spherical mapping. It was a long road to figure this out, so I hope you find it helpful.
              http://home.jasonfletcher.info/all/planetarium/spherical_mapping.png


              Rob,
              Indeed we have had similar issues where either MR misses or flickers subpixel sized objects. Raising the base sampling has proven to be the only fix. It increases render time exponentially. But as Mark suggested, rendering twice the resolution and downsizing is just as good of an option. Interesting idea to render straight into the render buffer. But I agree with Tom that slicing and encoding to MPEG is detrimental to the background stars. Perhaps when SkySkan converts over to the H264 encoder... There has been talk of this happening someday soon.


              --
              Jason Fletcher
              Charles Hayden Planetarium
            • Andy Dolph
              it is also possible (particularly in an AE workflow) to produce the starfields with Nightshade, create a script and output a frame sequence of that and then
              Message 6 of 14 , Mar 22, 2011
              • 0 Attachment
                it is also possible (particularly in an AE workflow) to produce the
                starfields with Nightshade, create a script and output a frame sequence of
                that and then bring it in to AE depending on what motions you need and
                such...

                Andy

                On Mon, Mar 21, 2011 at 6:41 PM, Drew Gilmore <dotsandlines@...>wrote:

                > (On the other hand though, it is possible to create an entire fulldome show
                > while staying within an After Effects workflow... except for the stars. I
                > haven't done any sort of survey but I wouldn't be surprised if some folks
                > would appreciate the ability to purchase just the final product, in the
                > form
                > of a cubic projection at several popular resolutions. That is more where my
                > hinting was directed.)
              • Drew Gilmore
                If you have the E&S Dome View Assembler plugin for After Effects, you can put together the six cubic sides of the starfield, and then animate the basic motions
                Message 7 of 14 , Mar 22, 2011
                • 0 Attachment
                  If you have the E&S Dome View Assembler plugin for After Effects, you can
                  put together the six cubic sides of the starfield, and then animate the
                  basic motions just with AE keyframes. I've done this a few times before with
                  random starfields, procedurally generated in 3D Max.

                  Drew

                  On Tue, Mar 22, 2011 at 8:11 AM, Andy Dolph <acdolph@...> wrote:

                  > it is also possible (particularly in an AE workflow) to produce the
                  > starfields with Nightshade, create a script and output a frame sequence of
                  > that and then bring it in to AE depending on what motions you need and
                  > such...
                • Tom Casey
                  Since you originally rendered the square textures with a 90 degree field of view camera, the texture actually has distortion build in to the image, the
                  Message 8 of 14 , Mar 22, 2011
                  • 0 Attachment
                    Since you originally rendered the square textures with a 90 degree
                    field of view camera, the texture actually has distortion build in to
                    the image, the distortion you would typically get as your view
                    changes towards the corners/edges of your cube. So the view from the
                    camera of the stars near the edges is appropriate... just make sure
                    you keep the camera view point near the center of the cube. If you
                    don't move the camera very much or make the cube extremely large, you
                    don't usually need to be concerned about constraining the cube to the
                    camera. But remember that the larger the cube is the more voxels (3D
                    space) your raytracer will need to contend with, which may slow the
                    render or create approximation errors and other render artifacts.

                    When using the star data base, the best way to create "lessor"
                    starfields is to have your "for each loop" ignore stars of lower
                    magnitude and thus render less stars. The 46,000 array is a little
                    much at times, but does begin to give you a slight Milky Way band.
                    Adding in our galaxy has always been a challenge, we usually just do
                    it in AE assuming the camera is not moving much.

                    Tom


                    On Mar 21, 2011, at 11:38 PM, Jason Fletcher wrote:

                    When you use a star-textured-cube enclosing the scene, doesn't the
                    perspective of the camera cause some stars to look distorted? Such as
                    the stars in the corner? Perhaps I'm missing something.

                    ************************************************
                    H o m e R u n P i c t u r e s

                    Tom Casey
                    President & Creative Director

                    100 First Avenue - Suite 450
                    Pittsburgh, PA 15222
                    412-391-8200
                    mailto:tom@...
                    http://www.homerunpictures.com
                  • Jason Fletcher
                    Ah fascinating! Thanks for the clarification. I will have to try that; creating fisheye renders is so tricky sometimes. I will begin experimenting with the
                    Message 9 of 14 , Mar 22, 2011
                    • 0 Attachment
                      Ah fascinating! Thanks for the clarification. I will have to try that;
                      creating fisheye renders is so tricky sometimes.

                      I will begin experimenting with the star database soon. Indeed the milky way
                      band may be tricky... I was thinking of overlaying a digital sky milky way
                      ontop of the final render. Just as you mentioned.

                      --
                      Jason Fletcher
                      www.jasonfletcher.info
                      512.695.7750


                      On Tue, Mar 22, 2011 at 12:18 PM, Tom Casey <tom@...> wrote:

                      > Since you originally rendered the square textures with a 90 degree
                      > field of view camera, the texture actually has distortion build in to
                      > the image, the distortion you would typically get as your view
                      > changes towards the corners/edges of your cube. So the view from the
                      > camera of the stars near the edges is appropriate... just make sure
                      > you keep the camera view point near the center of the cube. If you
                      > don't move the camera very much or make the cube extremely large, you
                      > don't usually need to be concerned about constraining the cube to the
                      > camera. But remember that the larger the cube is the more voxels (3D
                      > space) your raytracer will need to contend with, which may slow the
                      > render or create approximation errors and other render artifacts.

                      > When using the star data base, the best way to create "lessor"
                      > starfields is to have your "for each loop" ignore stars of lower
                      > magnitude and thus render less stars. The 46,000 array is a little
                      > much at times, but does begin to give you a slight Milky Way band.
                      > Adding in our galaxy has always been a challenge, we usually just do
                      > it in AE assuming the camera is not moving much.

                      > On Mar 21, 2011, at 11:38 PM, Jason Fletcher wrote:

                      > When you use a star-textured-cube enclosing the scene, doesn't the
                      > perspective of the camera cause some stars to look distorted? Such as
                      > the stars in the corner? Perhaps I'm missing something.
                    • Mark Aldred
                      The Milky Way is easy to include in the initial six bitmaps you output. If your star sphere is 100 units in diameter, make a cylinder 100.1 units wide, delete
                      Message 10 of 14 , Mar 22, 2011
                      • 0 Attachment
                        The Milky Way is easy to include in the initial six bitmaps you output. If your star sphere is 100 units in diameter, make a cylinder 100.1 units wide, delete the top and bottom, flip the normals, add the Milky Way texture and adjust height and orientation.

                        Mark
                        www.starlightstudio.jp

                        To: fulldome@yahoogroups.com
                        From: nuclearsugar@...
                        Date: Tue, 22 Mar 2011 12:43:03 -0400
                        Subject: [fulldome] Re: Background Stars in Maya




























                        Ah fascinating! Thanks for the clarification. I will have to try that;

                        creating fisheye renders is so tricky sometimes.



                        I will begin experimenting with the star database soon. Indeed the milky way

                        band may be tricky... I was thinking of overlaying a digital sky milky way

                        ontop of the final render. Just as you mentioned.



                        --

                        Jason Fletcher

                        www.jasonfletcher.info

                        512.695.7750



                        On Tue, Mar 22, 2011 at 12:18 PM, Tom Casey <tom@...> wrote:



                        > Since you originally rendered the square textures with a 90 degree

                        > field of view camera, the texture actually has distortion build in to

                        > the image, the distortion you would typically get as your view

                        > changes towards the corners/edges of your cube. So the view from the

                        > camera of the stars near the edges is appropriate... just make sure

                        > you keep the camera view point near the center of the cube. If you

                        > don't move the camera very much or make the cube extremely large, you

                        > don't usually need to be concerned about constraining the cube to the

                        > camera. But remember that the larger the cube is the more voxels (3D

                        > space) your raytracer will need to contend with, which may slow the

                        > render or create approximation errors and other render artifacts.



                        > When using the star data base, the best way to create "lessor"

                        > starfields is to have your "for each loop" ignore stars of lower

                        > magnitude and thus render less stars. The 46,000 array is a little

                        > much at times, but does begin to give you a slight Milky Way band.

                        > Adding in our galaxy has always been a challenge, we usually just do

                        > it in AE assuming the camera is not moving much.



                        > On Mar 21, 2011, at 11:38 PM, Jason Fletcher wrote:



                        > When you use a star-textured-cube enclosing the scene, doesn't the

                        > perspective of the camera cause some stars to look distorted? Such as

                        > the stars in the corner? Perhaps I'm missing something.
















                        [Non-text portions of this message have been removed]
                      Your message has been successfully submitted and would be delivered to recipients shortly.