Vista - 3d Reflection effects

Hello,

I'm trying to achieve some mirror effects like those seen here:
http://www.microsoft.com/max/

From Adam Smith's post a while back, he suggested a couple ways to go about
this:
1) Add each element twice to your viewport. For copied items, flip em, and
apply a gradient, blending the original to an opaque color.
2) Add a second viewport which holds the same items the first viewport, but
flip the camera.

I'm a 3D newb, so please do correct me if I'm wrong, but it seems as though
the (2) approach would not work to achieve reflections like those in Max as
items are place at several different z values.

So I'm trying approach (1) right now, but I'm unsure of which transforms I
would need to apply for the reflection. I'm currently adding these
transformations to my cloned Model3DGroup in the OnLoaded()

a) rotation 180 around z axis
b) translate it down

But how do I obtain the inverted image? Do I need to apply a
ScaleTransform3D with negative values? Or is there a much simpler way to go
about this? I'm currently using Nov '5 CTP bits.

Thanks in advanced.

CK

Use a negate scale on the image itself, then apply the imagebrush to the plane.

"Craig Kelly" wrote:

> Hello,
>
> I'm trying to achieve some mirror effects like those seen here:
> http://www.microsoft.com/max/
>
> From Adam Smith's post a while back, he suggested a couple ways to go about
> this:
> 1) Add each element twice to your viewport. For copied items, flip em, and
> apply a gradient, blending the original to an opaque color.
> 2) Add a second viewport which holds the same items the first viewport, but
> flip the camera.
>
> I'm a 3D newb, so please do correct me if I'm wrong, but it seems as though
> the (2) approach would not work to achieve reflections like those in Max as
> items are place at several different z values.
>
> So I'm trying approach (1) right now, but I'm unsure of which transforms I
> would need to apply for the reflection. I'm currently adding these
> transformations to my cloned Model3DGroup in the OnLoaded()
>
> a) rotation 180 around z axis
> b) translate it down
>
> But how do I obtain the inverted image? Do I need to apply a
> ScaleTransform3D with negative values? Or is there a much simpler way to go
> about this? I'm currently using Nov '5 CTP bits.
>
> Thanks in advanced.
>
> CK

That works to mirror the contents, but if the mesh itself isn't symmetric
about the Y-Z plane (or the plane perpendicular to the "mirror" - I'm
mentally picturing a "reflective table" mirror for this example), I think
this will still look incorrect. Consider a mesh with 1 equilateral triangle
in the X-Y plane, point to the +X direction (so there's a flat side in
the -X direction). Rotating this 180 about Z and translating -Y will end up
with a triangle with a point in the -X direction.

My suggestion was indeed to use a -1 scale in the camera (in whatever
direction you want as the normal for the mirror). You may also need a
center, if you don't want to mirror about the origin. You should then be
able to re-use (share) all of the Model3Ds in your scene, which'll reduce
the working set (duplicating meshes can be expensive, if they're large).

-Adam Smith [MS]

"TheRHogue" <TheRHogue@discussions.microsoft.com> wrote in message
news:41EC828F-BEF9-4A05-BCF1-F6DEF9673ACB@microsoft.com...
> Use a negate scale on the image itself, then apply the imagebrush to the
> plane.
>
> "Craig Kelly" wrote:
>
>> Hello,
>>
>> I'm trying to achieve some mirror effects like those seen here:
>> http://www.microsoft.com/max/
>>
>> From Adam Smith's post a while back, he suggested a couple ways to go
>> about
>> this:
>> 1) Add each element twice to your viewport. For copied items, flip em,
>> and
>> apply a gradient, blending the original to an opaque color.
>> 2) Add a second viewport which holds the same items the first viewport,
>> but
>> flip the camera.
>>
>> I'm a 3D newb, so please do correct me if I'm wrong, but it seems as
>> though
>> the (2) approach would not work to achieve reflections like those in Max
>> as
>> items are place at several different z values.
>>
>> So I'm trying approach (1) right now, but I'm unsure of which transforms
>> I
>> would need to apply for the reflection. I'm currently adding these
>> transformations to my cloned Model3DGroup in the OnLoaded()
>>
>> a) rotation 180 around z axis
>> b) translate it down
>>
>> But how do I obtain the inverted image? Do I need to apply a
>> ScaleTransform3D with negative values? Or is there a much simpler way to
>> go
>> about this? I'm currently using Nov '5 CTP bits.
>>
>> Thanks in advanced.
>>
>> CK

So first, (Daniel?,) I tried applying a negative scale ImageBrush transform,
but then I ended up with an image that was so opaque that I couldn't even
tell if it was mirrored it or not. Not sure if the resulting opacity is a
bug, or expected behavior. I'm using Nov '05 CTP.

Adam, I see what you mean about the rotating about the X-axis and a
translate not mirroring. Sounds like your suggestion of using a -1 scale for
the camera would require multiple viewports? I had written off multiple
viewports because I didn't think it would work to achieve the effects found
here:
http://www.microsoft.com/max/

There, some of the objects like the 'Max' logo are positioned ontop of where
other objects are reflecting. Or have I missed something?

Assuming that I'm wrong about multiple viewports (pretty safe bet here),
when you talk about -1 scale for the camera, do you mean changing the
UpDirection? And about sharing models, I noticed that I have had to
explicitly call 'Clone' to add the reflected objects. How do I set that up
so that they are using references instead of copies?

Thanks so much for your help guys

-CK

"Adam Smith [MS]" wrote:

> That works to mirror the contents, but if the mesh itself isn't symmetric
> about the Y-Z plane (or the plane perpendicular to the "mirror" - I'm
> mentally picturing a "reflective table" mirror for this example), I think
> this will still look incorrect. Consider a mesh with 1 equilateral triangle
> in the X-Y plane, point to the +X direction (so there's a flat side in
> the -X direction). Rotating this 180 about Z and translating -Y will end up
> with a triangle with a point in the -X direction.
>
> My suggestion was indeed to use a -1 scale in the camera (in whatever
> direction you want as the normal for the mirror). You may also need a
> center, if you don't want to mirror about the origin. You should then be
> able to re-use (share) all of the Model3Ds in your scene, which'll reduce
> the working set (duplicating meshes can be expensive, if they're large).
>
> -Adam Smith [MS]
>
> "TheRHogue" <TheRHogue@discussions.microsoft.com> wrote in message
> news:41EC828F-BEF9-4A05-BCF1-F6DEF9673ACB@microsoft.com...
> > Use a negate scale on the image itself, then apply the imagebrush to the
> > plane.
> >
> > "Craig Kelly" wrote:
> >
> >> Hello,
> >>
> >> I'm trying to achieve some mirror effects like those seen here:
> >> http://www.microsoft.com/max/
> >>
> >> From Adam Smith's post a while back, he suggested a couple ways to go
> >> about
> >> this:
> >> 1) Add each element twice to your viewport. For copied items, flip em,
> >> and
> >> apply a gradient, blending the original to an opaque color.
> >> 2) Add a second viewport which holds the same items the first viewport,
> >> but
> >> flip the camera.
> >>
> >> I'm a 3D newb, so please do correct me if I'm wrong, but it seems as
> >> though
> >> the (2) approach would not work to achieve reflections like those in Max
> >> as
> >> items are place at several different z values.
> >>
> >> So I'm trying approach (1) right now, but I'm unsure of which transforms
> >> I
> >> would need to apply for the reflection. I'm currently adding these
> >> transformations to my cloned Model3DGroup in the OnLoaded()
> >>
> >> a) rotation 180 around z axis
> >> b) translate it down
> >>
> >> But how do I obtain the inverted image? Do I need to apply a
> >> ScaleTransform3D with negative values? Or is there a much simpler way to
> >> go
> >> about this? I'm currently using Nov '5 CTP bits.
> >>
> >> Thanks in advanced.
> >>
> >> CK
>
>
>

You're right about using multiple Viewports if you transform the Camera - I
believe that because of the way a reflective surface works, all "real"
objects will be on top of all mirrored objects. As a result, you can use
two Viewports, where the reflected contents are in one Viewport and the
"real" contents are in another Viewport on top (in 2d) of the reflection's
Viewport. This technique allows you to place a 2d gradient between the two
to simulate an attenuated reflection, if you like.

You should also be able to do this within the 3D scene if you create two
root groups - one will contain the entire "real" scene, and the other will
contain one child with an appropriate -1 scale which will then also have (as
children) the entire real scene again. If you want to place a reflective
surface into the world in this choice, you'll need to make a Mesh to do so -
an advantage here is that the surface will be lit.

As for sharing - you shouldn't have to Clone in general (though there may be
reasons in your code - databinding and dynamic resource references come to
mind). If you don't have databinding/resource references, does something
not work if you just share?

-Adam Smith [MS]


"Craig Kelly" <CraigKelly@discussions.microsoft.com> wrote in message
news:F47CD1EC-2C10-4AA3-B19C-EE5AF43EC4B0@microsoft.com...
> So first, (Daniel?,) I tried applying a negative scale ImageBrush
> transform,
> but then I ended up with an image that was so opaque that I couldn't even
> tell if it was mirrored it or not. Not sure if the resulting opacity is a
> bug, or expected behavior. I'm using Nov '05 CTP.
>
> Adam, I see what you mean about the rotating about the X-axis and a
> translate not mirroring. Sounds like your suggestion of using a -1 scale
> for
> the camera would require multiple viewports? I had written off multiple
> viewports because I didn't think it would work to achieve the effects
> found
> here:
> http://www.microsoft.com/max/
>
> There, some of the objects like the 'Max' logo are positioned ontop of
> where
> other objects are reflecting. Or have I missed something?
>
> Assuming that I'm wrong about multiple viewports (pretty safe bet here),
> when you talk about -1 scale for the camera, do you mean changing the
> UpDirection? And about sharing models, I noticed that I have had to
> explicitly call 'Clone' to add the reflected objects. How do I set that
> up
> so that they are using references instead of copies?
>
> Thanks so much for your help guys
>
> -CK
>
> "Adam Smith [MS]" wrote:
>
>> That works to mirror the contents, but if the mesh itself isn't symmetric
>> about the Y-Z plane (or the plane perpendicular to the "mirror" - I'm
>> mentally picturing a "reflective table" mirror for this example), I think
>> this will still look incorrect. Consider a mesh with 1 equilateral
>> triangle
>> in the X-Y plane, point to the +X direction (so there's a flat side in
>> the -X direction). Rotating this 180 about Z and translating -Y will end
>> up
>> with a triangle with a point in the -X direction.
>>
>> My suggestion was indeed to use a -1 scale in the camera (in whatever
>> direction you want as the normal for the mirror). You may also need a
>> center, if you don't want to mirror about the origin. You should then be
>> able to re-use (share) all of the Model3Ds in your scene, which'll reduce
>> the working set (duplicating meshes can be expensive, if they're large).
>>
>> -Adam Smith [MS]
>>
>> "TheRHogue" <TheRHogue@discussions.microsoft.com> wrote in message
>> news:41EC828F-BEF9-4A05-BCF1-F6DEF9673ACB@microsoft.com...
>> > Use a negate scale on the image itself, then apply the imagebrush to
>> > the
>> > plane.
>> >
>> > "Craig Kelly" wrote:
>> >
>> >> Hello,
>> >>
>> >> I'm trying to achieve some mirror effects like those seen here:
>> >> http://www.microsoft.com/max/
>> >>
>> >> From Adam Smith's post a while back, he suggested a couple ways to go
>> >> about
>> >> this:
>> >> 1) Add each element twice to your viewport. For copied items, flip em,
>> >> and
>> >> apply a gradient, blending the original to an opaque color.
>> >> 2) Add a second viewport which holds the same items the first
>> >> viewport,
>> >> but
>> >> flip the camera.
>> >>
>> >> I'm a 3D newb, so please do correct me if I'm wrong, but it seems as
>> >> though
>> >> the (2) approach would not work to achieve reflections like those in
>> >> Max
>> >> as
>> >> items are place at several different z values.
>> >>
>> >> So I'm trying approach (1) right now, but I'm unsure of which
>> >> transforms
>> >> I
>> >> would need to apply for the reflection. I'm currently adding these
>> >> transformations to my cloned Model3DGroup in the OnLoaded()
>> >>
>> >> a) rotation 180 around z axis
>> >> b) translate it down
>> >>
>> >> But how do I obtain the inverted image? Do I need to apply a
>> >> ScaleTransform3D with negative values? Or is there a much simpler way
>> >> to
>> >> go
>> >> about this? I'm currently using Nov '5 CTP bits.
>> >>
>> >> Thanks in advanced.
>> >>
>> >> CK
>>
>>
>>

Aha, having two viewports on top of each other - that makes sense now. I had
them stacked like in the 'Boing' sample I've looked at.

The other good news, is that now the reference stuff seems to be working
just fine now. Too little coffee yesterday (or perhaps too much).

But I'm unfortunately still confused about 'inverting' the camera in the 2nd
viewport. I can think of a couple ways how one might achieve the inversion:
1) Through scale transformations done on the root of the 3dmodel.
I get strange results with this approach - might be a coffee thing

2) or by adjusting the UpDirection on the camera.
The UpDirection in my source viewport's camera is "0,1,0". For the 2nd
viewport, I would think that I should have at least some portion of the
UpDirection would have to be in the z-direction, but the "0,1,0" looks
exactly the same as "0,1,1". Likewise with "1,0,1" and "1,0,0" . By
specifiying "0,-1,0" the object looks flipped, but to take your triangle
example, the 'point' is pointing the wrong way.

Have I missed something again?

Really appreciate it

CK

"Adam Smith [MS]" wrote:

> You're right about using multiple Viewports if you transform the Camera - I
> believe that because of the way a reflective surface works, all "real"
> objects will be on top of all mirrored objects. As a result, you can use
> two Viewports, where the reflected contents are in one Viewport and the
> "real" contents are in another Viewport on top (in 2d) of the reflection's
> Viewport. This technique allows you to place a 2d gradient between the two
> to simulate an attenuated reflection, if you like.
>
> You should also be able to do this within the 3D scene if you create two
> root groups - one will contain the entire "real" scene, and the other will
> contain one child with an appropriate -1 scale which will then also have (as
> children) the entire real scene again. If you want to place a reflective
> surface into the world in this choice, you'll need to make a Mesh to do so -
> an advantage here is that the surface will be lit.
>
> As for sharing - you shouldn't have to Clone in general (though there may be
> reasons in your code - databinding and dynamic resource references come to
> mind). If you don't have databinding/resource references, does something
> not work if you just share?
>
> -Adam Smith [MS]
>
>
> "Craig Kelly" <CraigKelly@discussions.microsoft.com> wrote in message
> news:F47CD1EC-2C10-4AA3-B19C-EE5AF43EC4B0@microsoft.com...
> > So first, (Daniel?,) I tried applying a negative scale ImageBrush
> > transform,
> > but then I ended up with an image that was so opaque that I couldn't even
> > tell if it was mirrored it or not. Not sure if the resulting opacity is a
> > bug, or expected behavior. I'm using Nov '05 CTP.
> >
> > Adam, I see what you mean about the rotating about the X-axis and a
> > translate not mirroring. Sounds like your suggestion of using a -1 scale
> > for
> > the camera would require multiple viewports? I had written off multiple
> > viewports because I didn't think it would work to achieve the effects
> > found
> > here:
> > http://www.microsoft.com/max/
> >
> > There, some of the objects like the 'Max' logo are positioned ontop of
> > where
> > other objects are reflecting. Or have I missed something?
> >
> > Assuming that I'm wrong about multiple viewports (pretty safe bet here),
> > when you talk about -1 scale for the camera, do you mean changing the
> > UpDirection? And about sharing models, I noticed that I have had to
> > explicitly call 'Clone' to add the reflected objects. How do I set that
> > up
> > so that they are using references instead of copies?
> >
> > Thanks so much for your help guys
> >
> > -CK
> >
> > "Adam Smith [MS]" wrote:
> >
> >> That works to mirror the contents, but if the mesh itself isn't symmetric
> >> about the Y-Z plane (or the plane perpendicular to the "mirror" - I'm
> >> mentally picturing a "reflective table" mirror for this example), I think
> >> this will still look incorrect. Consider a mesh with 1 equilateral
> >> triangle
> >> in the X-Y plane, point to the +X direction (so there's a flat side in
> >> the -X direction). Rotating this 180 about Z and translating -Y will end
> >> up
> >> with a triangle with a point in the -X direction.
> >>
> >> My suggestion was indeed to use a -1 scale in the camera (in whatever
> >> direction you want as the normal for the mirror). You may also need a
> >> center, if you don't want to mirror about the origin. You should then be
> >> able to re-use (share) all of the Model3Ds in your scene, which'll reduce
> >> the working set (duplicating meshes can be expensive, if they're large).
> >>
> >> -Adam Smith [MS]
> >>
> >> "TheRHogue" <TheRHogue@discussions.microsoft.com> wrote in message
> >> news:41EC828F-BEF9-4A05-BCF1-F6DEF9673ACB@microsoft.com...
> >> > Use a negate scale on the image itself, then apply the imagebrush to
> >> > the
> >> > plane.
> >> >
> >> > "Craig Kelly" wrote:
> >> >
> >> >> Hello,
> >> >>
> >> >> I'm trying to achieve some mirror effects like those seen here:
> >> >> http://www.microsoft.com/max/
> >> >>
> >> >> From Adam Smith's post a while back, he suggested a couple ways to go
> >> >> about
> >> >> this:
> >> >> 1) Add each element twice to your viewport. For copied items, flip em,
> >> >> and
> >> >> apply a gradient, blending the original to an opaque color.
> >> >> 2) Add a second viewport which holds the same items the first
> >> >> viewport,
> >> >> but
> >> >> flip the camera.
> >> >>
> >> >> I'm a 3D newb, so please do correct me if I'm wrong, but it seems as
> >> >> though
> >> >> the (2) approach would not work to achieve reflections like those in
> >> >> Max
> >> >> as
> >> >> items are place at several different z values.
> >> >>
> >> >> So I'm trying approach (1) right now, but I'm unsure of which
> >> >> transforms
> >> >> I
> >> >> would need to apply for the reflection. I'm currently adding these
> >> >> transformations to my cloned Model3DGroup in the OnLoaded()
> >> >>
> >> >> a) rotation 180 around z axis
> >> >> b) translate it down
> >> >>
> >> >> But how do I obtain the inverted image? Do I need to apply a
> >> >> ScaleTransform3D with negative values? Or is there a much simpler way
> >> >> to
> >> >> go
> >> >> about this? I'm currently using Nov '5 CTP bits.
> >> >>
> >> >> Thanks in advanced.
> >> >>
> >> >> CK
> >>
> >>
> >>
>
>
>

Craig,

Would you be willing to show how you were able to do this? Or maybe give
some pointers on how to start? I've been reviewing Xaml and find this very
interesting.




"Craig Kelly" <CraigKelly@discussions.microsoft.com> wrote in message
news978432C-9E70-4E6E-995F-42CB1BA4B596@microsoft.com...
> Aha, having two viewports on top of each other - that makes sense now. I
> had
> them stacked like in the 'Boing' sample I've looked at.
>
> The other good news, is that now the reference stuff seems to be working
> just fine now. Too little coffee yesterday (or perhaps too much).
>
> But I'm unfortunately still confused about 'inverting' the camera in the
> 2nd
> viewport. I can think of a couple ways how one might achieve the
> inversion:
> 1) Through scale transformations done on the root of the 3dmodel.
> I get strange results with this approach - might be a coffee thing
>
> 2) or by adjusting the UpDirection on the camera.
> The UpDirection in my source viewport's camera is "0,1,0". For the 2nd
> viewport, I would think that I should have at least some portion of the
> UpDirection would have to be in the z-direction, but the "0,1,0" looks
> exactly the same as "0,1,1". Likewise with "1,0,1" and "1,0,0" . By
> specifiying "0,-1,0" the object looks flipped, but to take your triangle
> example, the 'point' is pointing the wrong way.
>
> Have I missed something again?
>
> Really appreciate it
>
> CK
>
> "Adam Smith [MS]" wrote:
>
>> You're right about using multiple Viewports if you transform the Camera -
>> I
>> believe that because of the way a reflective surface works, all "real"
>> objects will be on top of all mirrored objects. As a result, you can use
>> two Viewports, where the reflected contents are in one Viewport and the
>> "real" contents are in another Viewport on top (in 2d) of the
>> reflection's
>> Viewport. This technique allows you to place a 2d gradient between the
>> two
>> to simulate an attenuated reflection, if you like.
>>
>> You should also be able to do this within the 3D scene if you create two
>> root groups - one will contain the entire "real" scene, and the other
>> will
>> contain one child with an appropriate -1 scale which will then also have
>> (as
>> children) the entire real scene again. If you want to place a reflective
>> surface into the world in this choice, you'll need to make a Mesh to do
>> so -
>> an advantage here is that the surface will be lit.
>>
>> As for sharing - you shouldn't have to Clone in general (though there may
>> be
>> reasons in your code - databinding and dynamic resource references come
>> to
>> mind). If you don't have databinding/resource references, does something
>> not work if you just share?
>>
>> -Adam Smith [MS]
>>
>>
>> "Craig Kelly" <CraigKelly@discussions.microsoft.com> wrote in message
>> news:F47CD1EC-2C10-4AA3-B19C-EE5AF43EC4B0@microsoft.com...
>> > So first, (Daniel?,) I tried applying a negative scale ImageBrush
>> > transform,
>> > but then I ended up with an image that was so opaque that I couldn't
>> > even
>> > tell if it was mirrored it or not. Not sure if the resulting opacity is
>> > a
>> > bug, or expected behavior. I'm using Nov '05 CTP.
>> >
>> > Adam, I see what you mean about the rotating about the X-axis and a
>> > translate not mirroring. Sounds like your suggestion of using a -1
>> > scale
>> > for
>> > the camera would require multiple viewports? I had written off multiple
>> > viewports because I didn't think it would work to achieve the effects
>> > found
>> > here:
>> > http://www.microsoft.com/max/
>> >
>> > There, some of the objects like the 'Max' logo are positioned ontop of
>> > where
>> > other objects are reflecting. Or have I missed something?
>> >
>> > Assuming that I'm wrong about multiple viewports (pretty safe bet
>> > here),
>> > when you talk about -1 scale for the camera, do you mean changing the
>> > UpDirection? And about sharing models, I noticed that I have had to
>> > explicitly call 'Clone' to add the reflected objects. How do I set
>> > that
>> > up
>> > so that they are using references instead of copies?
>> >
>> > Thanks so much for your help guys
>> >
>> > -CK
>> >
>> > "Adam Smith [MS]" wrote:
>> >
>> >> That works to mirror the contents, but if the mesh itself isn't
>> >> symmetric
>> >> about the Y-Z plane (or the plane perpendicular to the "mirror" - I'm
>> >> mentally picturing a "reflective table" mirror for this example), I
>> >> think
>> >> this will still look incorrect. Consider a mesh with 1 equilateral
>> >> triangle
>> >> in the X-Y plane, point to the +X direction (so there's a flat side in
>> >> the -X direction). Rotating this 180 about Z and translating -Y will
>> >> end
>> >> up
>> >> with a triangle with a point in the -X direction.
>> >>
>> >> My suggestion was indeed to use a -1 scale in the camera (in whatever
>> >> direction you want as the normal for the mirror). You may also need a
>> >> center, if you don't want to mirror about the origin. You should then
>> >> be
>> >> able to re-use (share) all of the Model3Ds in your scene, which'll
>> >> reduce
>> >> the working set (duplicating meshes can be expensive, if they're
>> >> large).
>> >>
>> >> -Adam Smith [MS]
>> >>
>> >> "TheRHogue" <TheRHogue@discussions.microsoft.com> wrote in message
>> >> news:41EC828F-BEF9-4A05-BCF1-F6DEF9673ACB@microsoft.com...
>> >> > Use a negate scale on the image itself, then apply the imagebrush to
>> >> > the
>> >> > plane.
>> >> >
>> >> > "Craig Kelly" wrote:
>> >> >
>> >> >> Hello,
>> >> >>
>> >> >> I'm trying to achieve some mirror effects like those seen here:
>> >> >> http://www.microsoft.com/max/
>> >> >>
>> >> >> From Adam Smith's post a while back, he suggested a couple ways to
>> >> >> go
>> >> >> about
>> >> >> this:
>> >> >> 1) Add each element twice to your viewport. For copied items, flip
>> >> >> em,
>> >> >> and
>> >> >> apply a gradient, blending the original to an opaque color.
>> >> >> 2) Add a second viewport which holds the same items the first
>> >> >> viewport,
>> >> >> but
>> >> >> flip the camera.
>> >> >>
>> >> >> I'm a 3D newb, so please do correct me if I'm wrong, but it seems
>> >> >> as
>> >> >> though
>> >> >> the (2) approach would not work to achieve reflections like those
>> >> >> in
>> >> >> Max
>> >> >> as
>> >> >> items are place at several different z values.
>> >> >>
>> >> >> So I'm trying approach (1) right now, but I'm unsure of which
>> >> >> transforms
>> >> >> I
>> >> >> would need to apply for the reflection. I'm currently adding these
>> >> >> transformations to my cloned Model3DGroup in the OnLoaded()
>> >> >>
>> >> >> a) rotation 180 around z axis
>> >> >> b) translate it down
>> >> >>
>> >> >> But how do I obtain the inverted image? Do I need to apply a
>> >> >> ScaleTransform3D with negative values? Or is there a much simpler
>> >> >> way
>> >> >> to
>> >> >> go
>> >> >> about this? I'm currently using Nov '5 CTP bits.
>> >> >>
>> >> >> Thanks in advanced.
>> >> >>
>> >> >> CK
>> >>
>> >>
>> >>
>>
>>
>>

I believe it's new with Nov CTP, but all Camera classes now have a
"Transform" property - so you can use this to perform the scale-inversion.

-Adam Smith [MS]


"Craig Kelly" <CraigKelly@discussions.microsoft.com> wrote in message
news978432C-9E70-4E6E-995F-42CB1BA4B596@microsoft.com...
> Aha, having two viewports on top of each other - that makes sense now. I
> had
> them stacked like in the 'Boing' sample I've looked at.
>
> The other good news, is that now the reference stuff seems to be working
> just fine now. Too little coffee yesterday (or perhaps too much).
>
> But I'm unfortunately still confused about 'inverting' the camera in the
> 2nd
> viewport. I can think of a couple ways how one might achieve the
> inversion:
> 1) Through scale transformations done on the root of the 3dmodel.
> I get strange results with this approach - might be a coffee thing
>
> 2) or by adjusting the UpDirection on the camera.
> The UpDirection in my source viewport's camera is "0,1,0". For the 2nd
> viewport, I would think that I should have at least some portion of the
> UpDirection would have to be in the z-direction, but the "0,1,0" looks
> exactly the same as "0,1,1". Likewise with "1,0,1" and "1,0,0" . By
> specifiying "0,-1,0" the object looks flipped, but to take your triangle
> example, the 'point' is pointing the wrong way.
>
> Have I missed something again?
>
> Really appreciate it
>
> CK
>
> "Adam Smith [MS]" wrote:
>
>> You're right about using multiple Viewports if you transform the Camera -
>> I
>> believe that because of the way a reflective surface works, all "real"
>> objects will be on top of all mirrored objects. As a result, you can use
>> two Viewports, where the reflected contents are in one Viewport and the
>> "real" contents are in another Viewport on top (in 2d) of the
>> reflection's
>> Viewport. This technique allows you to place a 2d gradient between the
>> two
>> to simulate an attenuated reflection, if you like.
>>
>> You should also be able to do this within the 3D scene if you create two
>> root groups - one will contain the entire "real" scene, and the other
>> will
>> contain one child with an appropriate -1 scale which will then also have
>> (as
>> children) the entire real scene again. If you want to place a reflective
>> surface into the world in this choice, you'll need to make a Mesh to do
>> so -
>> an advantage here is that the surface will be lit.
>>
>> As for sharing - you shouldn't have to Clone in general (though there may
>> be
>> reasons in your code - databinding and dynamic resource references come
>> to
>> mind). If you don't have databinding/resource references, does something
>> not work if you just share?
>>
>> -Adam Smith [MS]
>>
>>
>> "Craig Kelly" <CraigKelly@discussions.microsoft.com> wrote in message
>> news:F47CD1EC-2C10-4AA3-B19C-EE5AF43EC4B0@microsoft.com...
>> > So first, (Daniel?,) I tried applying a negative scale ImageBrush
>> > transform,
>> > but then I ended up with an image that was so opaque that I couldn't
>> > even
>> > tell if it was mirrored it or not. Not sure if the resulting opacity is
>> > a
>> > bug, or expected behavior. I'm using Nov '05 CTP.
>> >
>> > Adam, I see what you mean about the rotating about the X-axis and a
>> > translate not mirroring. Sounds like your suggestion of using a -1
>> > scale
>> > for
>> > the camera would require multiple viewports? I had written off multiple
>> > viewports because I didn't think it would work to achieve the effects
>> > found
>> > here:
>> > http://www.microsoft.com/max/
>> >
>> > There, some of the objects like the 'Max' logo are positioned ontop of
>> > where
>> > other objects are reflecting. Or have I missed something?
>> >
>> > Assuming that I'm wrong about multiple viewports (pretty safe bet
>> > here),
>> > when you talk about -1 scale for the camera, do you mean changing the
>> > UpDirection? And about sharing models, I noticed that I have had to
>> > explicitly call 'Clone' to add the reflected objects. How do I set
>> > that
>> > up
>> > so that they are using references instead of copies?
>> >
>> > Thanks so much for your help guys
>> >
>> > -CK
>> >
>> > "Adam Smith [MS]" wrote:
>> >
>> >> That works to mirror the contents, but if the mesh itself isn't
>> >> symmetric
>> >> about the Y-Z plane (or the plane perpendicular to the "mirror" - I'm
>> >> mentally picturing a "reflective table" mirror for this example), I
>> >> think
>> >> this will still look incorrect. Consider a mesh with 1 equilateral
>> >> triangle
>> >> in the X-Y plane, point to the +X direction (so there's a flat side in
>> >> the -X direction). Rotating this 180 about Z and translating -Y will
>> >> end
>> >> up
>> >> with a triangle with a point in the -X direction.
>> >>
>> >> My suggestion was indeed to use a -1 scale in the camera (in whatever
>> >> direction you want as the normal for the mirror). You may also need a
>> >> center, if you don't want to mirror about the origin. You should then
>> >> be
>> >> able to re-use (share) all of the Model3Ds in your scene, which'll
>> >> reduce
>> >> the working set (duplicating meshes can be expensive, if they're
>> >> large).
>> >>
>> >> -Adam Smith [MS]
>> >>
>> >> "TheRHogue" <TheRHogue@discussions.microsoft.com> wrote in message
>> >> news:41EC828F-BEF9-4A05-BCF1-F6DEF9673ACB@microsoft.com...
>> >> > Use a negate scale on the image itself, then apply the imagebrush to
>> >> > the
>> >> > plane.
>> >> >
>> >> > "Craig Kelly" wrote:
>> >> >
>> >> >> Hello,
>> >> >>
>> >> >> I'm trying to achieve some mirror effects like those seen here:
>> >> >> http://www.microsoft.com/max/
>> >> >>
>> >> >> From Adam Smith's post a while back, he suggested a couple ways to
>> >> >> go
>> >> >> about
>> >> >> this:
>> >> >> 1) Add each element twice to your viewport. For copied items, flip
>> >> >> em,
>> >> >> and
>> >> >> apply a gradient, blending the original to an opaque color.
>> >> >> 2) Add a second viewport which holds the same items the first
>> >> >> viewport,
>> >> >> but
>> >> >> flip the camera.
>> >> >>
>> >> >> I'm a 3D newb, so please do correct me if I'm wrong, but it seems
>> >> >> as
>> >> >> though
>> >> >> the (2) approach would not work to achieve reflections like those
>> >> >> in
>> >> >> Max
>> >> >> as
>> >> >> items are place at several different z values.
>> >> >>
>> >> >> So I'm trying approach (1) right now, but I'm unsure of which
>> >> >> transforms
>> >> >> I
>> >> >> would need to apply for the reflection. I'm currently adding these
>> >> >> transformations to my cloned Model3DGroup in the OnLoaded()
>> >> >>
>> >> >> a) rotation 180 around z axis
>> >> >> b) translate it down
>> >> >>
>> >> >> But how do I obtain the inverted image? Do I need to apply a
>> >> >> ScaleTransform3D with negative values? Or is there a much simpler
>> >> >> way
>> >> >> to
>> >> >> go
>> >> >> about this? I'm currently using Nov '5 CTP bits.
>> >> >>
>> >> >> Thanks in advanced.
>> >> >>
>> >> >> CK
>> >>
>> >>
>> >>
>>
>>
>>