gdalgorithms-list

[Algorithms] negative UV coords

[skunkwerk <sku...@gm...>]

hi,
   i'm looking for an explanation of what negative UV coordinates do when
using texture mapping.  my initial thought was that negative coordinates are
just a continuation in the same U-V plane of the tiled texture, so a value
of -1.6 would be equivalent to a positive value of 1.4

however, based on tests i did opening a simple 3d model of a plane with
positive, and then negative uv coordinates, it seems like the negative sign
is actually flipping the texture around the u/v axis.  is this correct?

thanks

Re: [Algorithms] negative UV coords

[Jason H. <jas...@di...>]

There's a number of issues involved in mapping a texture to a triangle, 
and given the nature of your question, I have to assume you may not be 
aware of some of them, so I'll try to explain the very fundamental 
components.

The UV coordinates of a vertex are just that... a location in a texture 
mapped from 0..1 in each axis.  Graphics hardware typically have 
different settings to handle what happens when UV values are outside of 
that range.  The most basic are Clamp, Wrap, and Mirror.  If you set it 
to Clamp, any value outside 0..1 becomes either 0 or 1, so you can't 
exceed the given map on a model.  On a typical cube with default UV 
mapping, that means you'll have one face with a texture on it and all 
the other faces will have the perimeter color stretched through them.  
If you set it to Mirror, every odd modulus is frac(U) and every even 
modulus is 1-frac(U), so 1.1 is remapped to 0.9... thus you can 
essentially tile a texture across a plane and have it unfold the texture 
forward and backward forever.  With Wrap set, you get the texture 
repeating as frac(U), so 1.1 is the same as 0.1.  Each axis is 
configured separately.

There's also a texture matrix that can be applied to the UVs, which I 
believe (correct me if I'm wrong) is applied before the 
wrap/clamp/mirror operation, so you can scale or rotate the UVs on a 
model without actually modifying the input data.

Now, what you were asking about is the direction of a mapping.  A 
geometric triangle has a face normal that can be computed by cross 
product, or using the right-hand rule from edges 0-1 to 0-2.  Let's 
pretend the texture is in the same plane as the triangle.  The three UVs 
of the triangle vertices form a triangle in the texture plane.  Since 
it's a triangle, it has a facing as well.  Computing the sign of the 
cross product, if it's positive (the right hand rule of edge 0-1 to 0-2 
leaves you with your thumb pointing the same direction as the triangle's 
face normal), the triangle's texture will be mapped as you see it in the 
texture.  However, if you have a negative sign, or your thumb points 
opposite of the face normal, it will appear to be mirrored/backwards.  
It's not wrong, just the way it works.

One thing is certain, if you are just poking at UV values and trying to 
figure it out, it will seem complicated.  Here's a few things to think 
about.  If you map a triangle, translating the UVs will never cause the 
direction of the mapping to flip (unless you set Mirror mode).  But 
simply negating them will, because that reverses the face normal.  
Negating either the positions or UVs will make the texture appear 
mirrored/backward.  So, if you negate both the vertex positions and UVs, 
you have flipped both triangle normals, and it will appear to be mapped 
correctly, but facing the opposite direction.  :-)

I hope that helps.

JH

skunkwerk wrote:
> hi,
>    i'm looking for an explanation of what negative UV coordinates do 
> when using texture mapping.  my initial thought was that negative 
> coordinates are just a continuation in the same U-V plane of the tiled 
> texture, so a value of -1.6 would be equivalent to a positive value of 1.4
>
> however, based on tests i did opening a simple 3d model of a plane 
> with positive, and then negative uv coordinates, it seems like the 
> negative sign is actually flipping the texture around the u/v axis.  
> is this correct?
>

Re: [Algorithms] negative UV coords

[Tony B. <ton...@kr...>]

In other words, yes.


> There's a number of issues involved in mapping a texture to a triangle,
> and given the nature of your question, I have to assume you may not be
> aware of some of them, so I'll try to explain the very fundamental
> components.
>
> The UV coordinates of a vertex are just that... a location in a texture
> mapped from 0..1 in each axis.  Graphics hardware typically have
> different settings to handle what happens when UV values are outside of
> that range.  The most basic are Clamp, Wrap, and Mirror.  If you set it
> to Clamp, any value outside 0..1 becomes either 0 or 1, so you can't
> exceed the given map on a model.  On a typical cube with default UV
> mapping, that means you'll have one face with a texture on it and all
> the other faces will have the perimeter color stretched through them.
> If you set it to Mirror, every odd modulus is frac(U) and every even
> modulus is 1-frac(U), so 1.1 is remapped to 0.9... thus you can
> essentially tile a texture across a plane and have it unfold the texture
> forward and backward forever.  With Wrap set, you get the texture
> repeating as frac(U), so 1.1 is the same as 0.1.  Each axis is
> configured separately.
>
> There's also a texture matrix that can be applied to the UVs, which I
> believe (correct me if I'm wrong) is applied before the
> wrap/clamp/mirror operation, so you can scale or rotate the UVs on a
> model without actually modifying the input data.
>
> Now, what you were asking about is the direction of a mapping.  A
> geometric triangle has a face normal that can be computed by cross
> product, or using the right-hand rule from edges 0-1 to 0-2.  Let's
> pretend the texture is in the same plane as the triangle.  The three UVs
> of the triangle vertices form a triangle in the texture plane.  Since
> it's a triangle, it has a facing as well.  Computing the sign of the
> cross product, if it's positive (the right hand rule of edge 0-1 to 0-2
> leaves you with your thumb pointing the same direction as the triangle's
> face normal), the triangle's texture will be mapped as you see it in the
> texture.  However, if you have a negative sign, or your thumb points
> opposite of the face normal, it will appear to be mirrored/backwards.
> It's not wrong, just the way it works.
>
> One thing is certain, if you are just poking at UV values and trying to
> figure it out, it will seem complicated.  Here's a few things to think
> about.  If you map a triangle, translating the UVs will never cause the
> direction of the mapping to flip (unless you set Mirror mode).  But
> simply negating them will, because that reverses the face normal.
> Negating either the positions or UVs will make the texture appear
> mirrored/backward.  So, if you negate both the vertex positions and UVs,
> you have flipped both triangle normals, and it will appear to be mapped
> correctly, but facing the opposite direction.  :-)
>
> I hope that helps.
>
> JH
>
> skunkwerk wrote:
>> hi,
>>    i'm looking for an explanation of what negative UV coordinates do
>> when using texture mapping.  my initial thought was that negative
>> coordinates are just a continuation in the same U-V plane of the tiled
>> texture, so a value of -1.6 would be equivalent to a positive value of 1.4
>>
>> however, based on tests i did opening a simple 3d model of a plane
>> with positive, and then negative uv coordinates, it seems like the
>> negative sign is actually flipping the texture around the u/v axis.
>> is this correct?
>>
>
> -------------------------------------------------------------------------
> This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
> Build the coolest Linux based applications with Moblin SDK & win great prizes
> Grand prize is a trip for two to an Open Source event anywhere in the world
> http://moblin-contest.org/redirect.php?banner_id=100&url=/
> _______________________________________________
> GDAlgorithms-list mailing list
> GDA...@li...
> https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list
> Archives:
> http://sourceforge.net/mailarchive/forum.php?forum_name=gdalgorithms-list
>
>





This message and its attachments may contain legally privileged or confidential information. This message is intended for the use of the individual or entity to which it is addressed. If you are not the addressee indicated in this message, or the employee or agent responsible for delivering the message to the intended recipient, you may not copy or deliver this message or its attachments to anyone. Rather, you should permanently delete this message and its attachments and kindly notify the sender by reply e-mail. Any content of this message and its attachments, which does not relate to the official business of the sending company must be taken not to have been sent or endorsed by the sending company or any of its related entities. No warranty is made that the e-mail or attachment(s) are free from computer virus or other defect.

Re: [Algorithms] negative UV coords

[Mat N. <mat...@bu...>]

Or no, depending on the render state.

MSN
-----Original Message-----
From: gda...@li... [mailto:gda...@li...] On Behalf Of Tony Ball
Sent: Saturday, July 26, 2008 2:15 PM
To: Game Development Algorithms
Subject: Re: [Algorithms] negative UV coords

In other words, yes.


> There's a number of issues involved in mapping a texture to a triangle,
> and given the nature of your question, I have to assume you may not be
> aware of some of them, so I'll try to explain the very fundamental
> components.
>
> The UV coordinates of a vertex are just that... a location in a texture
> mapped from 0..1 in each axis.  Graphics hardware typically have
> different settings to handle what happens when UV values are outside of
> that range.  The most basic are Clamp, Wrap, and Mirror.  If you set it
> to Clamp, any value outside 0..1 becomes either 0 or 1, so you can't
> exceed the given map on a model.  On a typical cube with default UV
> mapping, that means you'll have one face with a texture on it and all
> the other faces will have the perimeter color stretched through them.
> If you set it to Mirror, every odd modulus is frac(U) and every even
> modulus is 1-frac(U), so 1.1 is remapped to 0.9... thus you can
> essentially tile a texture across a plane and have it unfold the texture
> forward and backward forever.  With Wrap set, you get the texture
> repeating as frac(U), so 1.1 is the same as 0.1.  Each axis is
> configured separately.
>
> There's also a texture matrix that can be applied to the UVs, which I
> believe (correct me if I'm wrong) is applied before the
> wrap/clamp/mirror operation, so you can scale or rotate the UVs on a
> model without actually modifying the input data.
>
> Now, what you were asking about is the direction of a mapping.  A
> geometric triangle has a face normal that can be computed by cross
> product, or using the right-hand rule from edges 0-1 to 0-2.  Let's
> pretend the texture is in the same plane as the triangle.  The three UVs
> of the triangle vertices form a triangle in the texture plane.  Since
> it's a triangle, it has a facing as well.  Computing the sign of the
> cross product, if it's positive (the right hand rule of edge 0-1 to 0-2
> leaves you with your thumb pointing the same direction as the triangle's
> face normal), the triangle's texture will be mapped as you see it in the
> texture.  However, if you have a negative sign, or your thumb points
> opposite of the face normal, it will appear to be mirrored/backwards.
> It's not wrong, just the way it works.
>
> One thing is certain, if you are just poking at UV values and trying to
> figure it out, it will seem complicated.  Here's a few things to think
> about.  If you map a triangle, translating the UVs will never cause the
> direction of the mapping to flip (unless you set Mirror mode).  But
> simply negating them will, because that reverses the face normal.
> Negating either the positions or UVs will make the texture appear
> mirrored/backward.  So, if you negate both the vertex positions and UVs,
> you have flipped both triangle normals, and it will appear to be mapped
> correctly, but facing the opposite direction.  :-)
>
> I hope that helps.
>
> JH
>
> skunkwerk wrote:
>> hi,
>>    i'm looking for an explanation of what negative UV coordinates do
>> when using texture mapping.  my initial thought was that negative
>> coordinates are just a continuation in the same U-V plane of the tiled
>> texture, so a value of -1.6 would be equivalent to a positive value of 1.4
>>
>> however, based on tests i did opening a simple 3d model of a plane
>> with positive, and then negative uv coordinates, it seems like the
>> negative sign is actually flipping the texture around the u/v axis.
>> is this correct?
>>
>
> -------------------------------------------------------------------------
> This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
> Build the coolest Linux based applications with Moblin SDK & win great prizes
> Grand prize is a trip for two to an Open Source event anywhere in the world
> http://moblin-contest.org/redirect.php?banner_id=100&url=/
> _______________________________________________
> GDAlgorithms-list mailing list
> GDA...@li...
> https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list
> Archives:
> http://sourceforge.net/mailarchive/forum.php?forum_name=gdalgorithms-list
>
>





This message and its attachments may contain legally privileged or confidential information. This message is intended for the use of the individual or entity to which it is addressed. If you are not the addressee indicated in this message, or the employee or agent responsible for delivering the message to the intended recipient, you may not copy or deliver this message or its attachments to anyone. Rather, you should permanently delete this message and its attachments and kindly notify the sender by reply e-mail. Any content of this message and its attachments, which does not relate to the official business of the sending company must be taken not to have been sent or endorsed by the sending company or any of its related entities. No warranty is made that the e-mail or attachment(s) are free from computer virus or other defect.


-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
GDAlgorithms-list mailing list
GDA...@li...
https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list
Archives:
http://sourceforge.net/mailarchive/forum.php?forum_name=gdalgorithms-list

Re: [Algorithms] negative UV coords

[Tom P. <ga...@fa...>]

> however, based on tests i did opening a simple 3d model of a plane
> with positive, and then negative uv coordinates, it seems like the
> negative sign is actually flipping the texture around the u/v
> axis.  is this correct?

Based on the variety of answers you're getting, I'd suggest that it
depends how it's implemented.

Flipping when going negative seems bizarre to me, and I'm trying to
remember what happened on the PS2 (which is the last platform where
I even thought about it), however it had enough challenges with
tiling in the first place that there had to be significant magic in
the data conditioner to make the data look like something that the
machine could handle at all.

-tom!

Re: [Algorithms] negative UV coords

[John P. <jo...@3d...>]

Assuming you don't have texture clamping turned on, negating uv's will 
cause them to be mirrored along the negated axis.  If you think about 
it, by negating u, you will be walking backwards along the u axis, which 
means those texels will be encountered in reverse order, which is the 
order they will be mapped onto the triangle.


Tom Plunket wrote:
>> however, based on tests i did opening a simple 3d model of a plane
>> with positive, and then negative uv coordinates, it seems like the
>> negative sign is actually flipping the texture around the u/v
>> axis.  is this correct?
>>     
>
> Based on the variety of answers you're getting, I'd suggest that it
> depends how it's implemented.
>
> Flipping when going negative seems bizarre to me, and I'm trying to
> remember what happened on the PS2 (which is the last platform where
> I even thought about it), however it had enough challenges with
> tiling in the first place that there had to be significant magic in
> the data conditioner to make the data look like something that the
> machine could handle at all.
>
> -tom!
>
> -------------------------------------------------------------------------
> This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
> Build the coolest Linux based applications with Moblin SDK & win great prizes
> Grand prize is a trip for two to an Open Source event anywhere in the world
> http://moblin-contest.org/redirect.php?banner_id=100&url=/
> _______________________________________________
> GDAlgorithms-list mailing list
> GDA...@li...
> https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list
> Archives:
> http://sourceforge.net/mailarchive/forum.php?forum_name=gdalgorithms-list
>
>   

Re: [Algorithms] negative UV coords

[Mat N. <mat...@bu...>]

To be even more anal, D3D defines 5 separate texture addressing modes. So, again, it depends.

MSN
-----Original Message-----
From: gda...@li... [mailto:gda...@li...] On Behalf Of John Pollard
Sent: Saturday, July 26, 2008 3:34 PM
To: Game Development Algorithms
Subject: Re: [Algorithms] negative UV coords

Assuming you don't have texture clamping turned on, negating uv's will
cause them to be mirrored along the negated axis.  If you think about
it, by negating u, you will be walking backwards along the u axis, which
means those texels will be encountered in reverse order, which is the
order they will be mapped onto the triangle.


Tom Plunket wrote:
>> however, based on tests i did opening a simple 3d model of a plane
>> with positive, and then negative uv coordinates, it seems like the
>> negative sign is actually flipping the texture around the u/v
>> axis.  is this correct?
>>
>
> Based on the variety of answers you're getting, I'd suggest that it
> depends how it's implemented.
>
> Flipping when going negative seems bizarre to me, and I'm trying to
> remember what happened on the PS2 (which is the last platform where
> I even thought about it), however it had enough challenges with
> tiling in the first place that there had to be significant magic in
> the data conditioner to make the data look like something that the
> machine could handle at all.
>
> -tom!
>
> -------------------------------------------------------------------------
> This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
> Build the coolest Linux based applications with Moblin SDK & win great prizes
> Grand prize is a trip for two to an Open Source event anywhere in the world
> http://moblin-contest.org/redirect.php?banner_id=100&url=/
> _______________________________________________
> GDAlgorithms-list mailing list
> GDA...@li...
> https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list
> Archives:
> http://sourceforge.net/mailarchive/forum.php?forum_name=gdalgorithms-list
>
>


-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
GDAlgorithms-list mailing list
GDA...@li...
https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list
Archives:
http://sourceforge.net/mailarchive/forum.php?forum_name=gdalgorithms-list

Re: [Algorithms] negative UV coords

[Tony B. <ton...@kr...>]

Another way of thinking about it that might help some people is to realize
that only the fractional part of the UV coordinates are used to look up the
texture map. So going stepping from -1 to 0 the fractional component is moving
the opposite direction to when stepping from 0 to 1.


> Assuming you don't have texture clamping turned on, negating uv's will
> cause them to be mirrored along the negated axis.  If you think about
> it, by negating u, you will be walking backwards along the u axis, which
> means those texels will be encountered in reverse order, which is the
> order they will be mapped onto the triangle.
>
>
> Tom Plunket wrote:
>>> however, based on tests i did opening a simple 3d model of a plane
>>> with positive, and then negative uv coordinates, it seems like the
>>> negative sign is actually flipping the texture around the u/v
>>> axis.  is this correct?
>>>
>>
>> Based on the variety of answers you're getting, I'd suggest that it
>> depends how it's implemented.
>>
>> Flipping when going negative seems bizarre to me, and I'm trying to
>> remember what happened on the PS2 (which is the last platform where
>> I even thought about it), however it had enough challenges with
>> tiling in the first place that there had to be significant magic in
>> the data conditioner to make the data look like something that the
>> machine could handle at all.
>>
>> -tom!
>>
>> -------------------------------------------------------------------------
>> This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
>> Build the coolest Linux based applications with Moblin SDK & win great
>> prizes
>> Grand prize is a trip for two to an Open Source event anywhere in the world
>> http://moblin-contest.org/redirect.php?banner_id=100&url=/
>> _______________________________________________
>> GDAlgorithms-list mailing list
>> GDA...@li...
>> https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list
>> Archives:
>> http://sourceforge.net/mailarchive/forum.php?forum_name=gdalgorithms-list
>>
>>
>
>
> -------------------------------------------------------------------------
> This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
> Build the coolest Linux based applications with Moblin SDK & win great prizes
> Grand prize is a trip for two to an Open Source event anywhere in the world
> http://moblin-contest.org/redirect.php?banner_id=100&url=/
> _______________________________________________
> GDAlgorithms-list mailing list
> GDA...@li...
> https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list
> Archives:
> http://sourceforge.net/mailarchive/forum.php?forum_name=gdalgorithms-list
>
>





This message and its attachments may contain legally privileged or confidential information. This message is intended for the use of the individual or entity to which it is addressed. If you are not the addressee indicated in this message, or the employee or agent responsible for delivering the message to the intended recipient, you may not copy or deliver this message or its attachments to anyone. Rather, you should permanently delete this message and its attachments and kindly notify the sender by reply e-mail. Any content of this message and its attachments, which does not relate to the official business of the sending company must be taken not to have been sent or endorsed by the sending company or any of its related entities. No warranty is made that the e-mail or attachment(s) are free from computer virus or other defect.

Re: [Algorithms] negative UV coords

[Jason H. <jas...@di...>]

In this one specific case, where the U or V has been negated (but not 
both), and the texture addressing mode is set to wrap, yes the texture 
is mirrored.  However, I suggest that it is a poor mental model, 
thinking about stepping as a property of certain coordinate regions, as 
if the coordinate system defines directionality of a mapping--it does not.

Flow is a function of the geometry triangle, UV triangle, and the 
texture addressing modes--not the property of any one component in 
isolation...

After fighting to generate really solid tangents (which depends on all 
these things), I began to appreciate the way it all works together.  :-)

Thanks,
JH

Tony Ball wrote:
> Another way of thinking about it that might help some people is to realize
> that only the fractional part of the UV coordinates are used to look up the
> texture map. So going stepping from -1 to 0 the fractional component is moving
> the opposite direction to when stepping from 0 to 1.
>
>
>   
>> Assuming you don't have texture clamping turned on, negating uv's will
>> cause them to be mirrored along the negated axis.  If you think about
>> it, by negating u, you will be walking backwards along the u axis, which
>> means those texels will be encountered in reverse order, which is the
>> order they will be mapped onto the triangle.
>>
>>
>> Tom Plunket wrote:
>>     
>>>> however, based on tests i did opening a simple 3d model of a plane
>>>> with positive, and then negative uv coordinates, it seems like the
>>>> negative sign is actually flipping the texture around the u/v
>>>> axis.  is this correct?
>>>>
>>>>         
>>> Based on the variety of answers you're getting, I'd suggest that it
>>> depends how it's implemented.
>>>
>>> Flipping when going negative seems bizarre to me, and I'm trying to
>>> remember what happened on the PS2 (which is the last platform where
>>> I even thought about it), however it had enough challenges with
>>> tiling in the first place that there had to be significant magic in
>>> the data conditioner to make the data look like something that the
>>> machine could handle at all.
>>>
>>> -tom!
>>>
>>> -------------------------------------------------------------------------
>>> This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
>>> Build the coolest Linux based applications with Moblin SDK & win great
>>> prizes
>>> Grand prize is a trip for two to an Open Source event anywhere in the world
>>> http://moblin-contest.org/redirect.php?banner_id=100&url=/
>>> _______________________________________________
>>> GDAlgorithms-list mailing list
>>> GDA...@li...
>>> https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list
>>> Archives:
>>> http://sourceforge.net/mailarchive/forum.php?forum_name=gdalgorithms-list
>>>
>>>
>>>       
>> -------------------------------------------------------------------------
>> This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
>> Build the coolest Linux based applications with Moblin SDK & win great prizes
>> Grand prize is a trip for two to an Open Source event anywhere in the world
>> http://moblin-contest.org/redirect.php?banner_id=100&url=/
>> _______________________________________________
>> GDAlgorithms-list mailing list
>> GDA...@li...
>> https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list
>> Archives:
>> http://sourceforge.net/mailarchive/forum.php?forum_name=gdalgorithms-list
>>
>>
>>     
>
>
>
>
>
> This message and its attachments may contain legally privileged or confidential information. This message is intended for the use of the individual or entity to which it is addressed. If you are not the addressee indicated in this message, or the employee or agent responsible for delivering the message to the intended recipient, you may not copy or deliver this message or its attachments to anyone. Rather, you should permanently delete this message and its attachments and kindly notify the sender by reply e-mail. Any content of this message and its attachments, which does not relate to the official business of the sending company must be taken not to have been sent or endorsed by the sending company or any of its related entities. No warranty is made that the e-mail or attachment(s) are free from computer virus or other defect.
>
>
> -------------------------------------------------------------------------
> This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
> Build the coolest Linux based applications with Moblin SDK & win great prizes
> Grand prize is a trip for two to an Open Source event anywhere in the world
> http://moblin-contest.org/redirect.php?banner_id=100&url=/
> _______________________________________________
> GDAlgorithms-list mailing list
> GDA...@li...
> https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list
> Archives:
> http://sourceforge.net/mailarchive/forum.php?forum_name=gdalgorithms-list
>
>   

Re: [Algorithms] negative UV coords

[Tony B. <ton...@kr...>]

True, but the original post was a very simple question, so I thought it
deserved a simple answer. :)


> In this one specific case, where the U or V has been negated (but not
> both), and the texture addressing mode is set to wrap, yes the texture
> is mirrored.  However, I suggest that it is a poor mental model,
> thinking about stepping as a property of certain coordinate regions, as
> if the coordinate system defines directionality of a mapping--it does not.
>
> Flow is a function of the geometry triangle, UV triangle, and the
> texture addressing modes--not the property of any one component in
> isolation...
>
> After fighting to generate really solid tangents (which depends on all
> these things), I began to appreciate the way it all works together.  :-)
>
> Thanks,
> JH
>
> Tony Ball wrote:
>> Another way of thinking about it that might help some people is to realize
>> that only the fractional part of the UV coordinates are used to look up the
>> texture map. So going stepping from -1 to 0 the fractional component is
>> moving
>> the opposite direction to when stepping from 0 to 1.
>>
>>
>>
>>> Assuming you don't have texture clamping turned on, negating uv's will
>>> cause them to be mirrored along the negated axis.  If you think about
>>> it, by negating u, you will be walking backwards along the u axis, which
>>> means those texels will be encountered in reverse order, which is the
>>> order they will be mapped onto the triangle.
>>>
>>>
>>> Tom Plunket wrote:
>>>
>>>>> however, based on tests i did opening a simple 3d model of a plane
>>>>> with positive, and then negative uv coordinates, it seems like the
>>>>> negative sign is actually flipping the texture around the u/v
>>>>> axis.  is this correct?
>>>>>
>>>>>
>>>> Based on the variety of answers you're getting, I'd suggest that it
>>>> depends how it's implemented.
>>>>
>>>> Flipping when going negative seems bizarre to me, and I'm trying to
>>>> remember what happened on the PS2 (which is the last platform where
>>>> I even thought about it), however it had enough challenges with
>>>> tiling in the first place that there had to be significant magic in
>>>> the data conditioner to make the data look like something that the
>>>> machine could handle at all.
>>>>
>>>> -tom!
>>>>
>>>> -------------------------------------------------------------------------
>>>> This SF.Net email is sponsored by the Moblin Your Move Developer's
>>>> challenge
>>>> Build the coolest Linux based applications with Moblin SDK & win great
>>>> prizes
>>>> Grand prize is a trip for two to an Open Source event anywhere in the
>>>> world
>>>> http://moblin-contest.org/redirect.php?banner_id=100&url=/
>>>> _______________________________________________
>>>> GDAlgorithms-list mailing list
>>>> GDA...@li...
>>>> https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list
>>>> Archives:
>>>> http://sourceforge.net/mailarchive/forum.php?forum_name=gdalgorithms-list
>>>>
>>>>
>>>>
>>> -------------------------------------------------------------------------
>>> This SF.Net email is sponsored by the Moblin Your Move Developer's
>>> challenge
>>> Build the coolest Linux based applications with Moblin SDK & win great
>>> prizes
>>> Grand prize is a trip for two to an Open Source event anywhere in the world
>>> http://moblin-contest.org/redirect.php?banner_id=100&url=/
>>> _______________________________________________
>>> GDAlgorithms-list mailing list
>>> GDA...@li...
>>> https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list
>>> Archives:
>>> http://sourceforge.net/mailarchive/forum.php?forum_name=gdalgorithms-list
>>>
>>>
>>>
>>
>>
>>
>>
>>
>> This message and its attachments may contain legally privileged or
>> confidential information. This message is intended for the use of the
>> individual or entity to which it is addressed. If you are not the addressee
>> indicated in this message, or the employee or agent responsible for
>> delivering the message to the intended recipient, you may not copy or
>> deliver this message or its attachments to anyone. Rather, you should
>> permanently delete this message and its attachments and kindly notify the
>> sender by reply e-mail. Any content of this message and its attachments,
>> which does not relate to the official business of the sending company must
>> be taken not to have been sent or endorsed by the sending company or any of
>> its related entities. No warranty is made that the e-mail or attachment(s)
>> are free from computer virus or other defect.
>>
>>
>> -------------------------------------------------------------------------
>> This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
>> Build the coolest Linux based applications with Moblin SDK & win great
>> prizes
>> Grand prize is a trip for two to an Open Source event anywhere in the world
>> http://moblin-contest.org/redirect.php?banner_id=100&url=/
>> _______________________________________________
>> GDAlgorithms-list mailing list
>> GDA...@li...
>> https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list
>> Archives:
>> http://sourceforge.net/mailarchive/forum.php?forum_name=gdalgorithms-list
>>
>>
> -------------------------------------------------------------------------
> This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
> Build the coolest Linux based applications with Moblin SDK & win great prizes
> Grand prize is a trip for two to an Open Source event anywhere in the world
> http://moblin-contest.org/redirect.php?banner_id=100&url=/_______________________________________________
> GDAlgorithms-list mailing list
> GDA...@li...
> https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list
> Archives:
> http://sourceforge.net/mailarchive/forum.php?forum_name=gdalgorithms-list





This message and its attachments may contain legally privileged or confidential information. This message is intended for the use of the individual or entity to which it is addressed. If you are not the addressee indicated in this message, or the employee or agent responsible for delivering the message to the intended recipient, you may not copy or deliver this message or its attachments to anyone. Rather, you should permanently delete this message and its attachments and kindly notify the sender by reply e-mail. Any content of this message and its attachments, which does not relate to the official business of the sending company must be taken not to have been sent or endorsed by the sending company or any of its related entities. No warranty is made that the e-mail or attachment(s) are free from computer virus or other defect.

Re: [Algorithms] negative UV coords

[Jon W. <jw...@gm...>]

Jason Hughes wrote:
> In this one specific case, where the U or V has been negated (but not 
> both), and the texture addressing mode is set to wrap, yes the texture 
> is mirrored.

That is only true when you have MIRROR texture addressing mode turned 
on. This is easy to verify in your modeling tool, which assumes WRAP: 
the texture is tiled continuously in all directions, including the 
negative directions.

However, the texture MAP may be mirrored on the SCREEN if your texture 
coordinate gradients don't match your texture memory lay-out and 
triangle mapping. Most people think that in Direct3D, V goes down-to-up, 
and in OpenGL, V goes up-to-down. However, that's just an artifact of 
D3D typically loading BMP textures, and OpenGL typically loading JPG or 
PNG textures (in the samples that are around). The actual definition is 
simple: V grows in the direction of the data you pass to the texture. If 
you pass data with the bottom-most scanline first, V grows upwards. If 
you pass data with the topmost scanline first, V grows downwards.

Sincerely,

jw

Re: [Algorithms] negative UV coords

[Tom P. <ga...@fa...>]

> Assuming you don't have texture clamping turned on, negating uv's
> will cause them to be mirrored along the negated axis.  If you
> think about it, by negating u, you will be walking backwards along
> the u axis, which means those texels will be encountered in reverse
> order, which is the order they will be mapped onto the triangle.

Again, it would seem (to me) to depend on the implementation, since
it depends how you interpret the fractional portion of the values. 
E.g. I seem to remember hearing that different hardware implemented
modulus differently, so -3.5 % 2.0 could yield 1.5 or .5 (or
possibly a negative version of one of those?), depending on your
implementation.  So then the question is one of whether you
interpret -.3 U as .3 or .7.  Can one say with certainty that one
way is wrong and the other right?  I mean, to me if I have the mode
"wrap" set, I sure wouldn't expect it'd implicitly mean "flip."  For
that, I'd choose a "mirror" mode instead.

-tom!

Re: [Algorithms] negative UV coords

[Emil P. <hu...@co...>]

With wrap mode it indeed wraps, rather than flip. The fractional part is
defined x - floor(x), which is always in [0, 1), so frac(-0.3) = 0.7.

I think the confusion in this thread is because some are talking about
negating texture coordinates (which flips) and some are talking about what
happens when the texture coordinate is interpolated from the positive to
negative side of zero (where it's just wrapping, like it does across any
other integer coordinate).

-Emil


-----Original Message-----
From: gda...@li...
[mailto:gda...@li...] On Behalf Of Tom
Plunket
Sent: 29 July 2008 00:45
To: Game Development Algorithms
Subject: Re: [Algorithms] negative UV coords


> Assuming you don't have texture clamping turned on, negating uv's
> will cause them to be mirrored along the negated axis.  If you
> think about it, by negating u, you will be walking backwards along
> the u axis, which means those texels will be encountered in reverse
> order, which is the order they will be mapped onto the triangle.

Again, it would seem (to me) to depend on the implementation, since
it depends how you interpret the fractional portion of the values. 
E.g. I seem to remember hearing that different hardware implemented
modulus differently, so -3.5 % 2.0 could yield 1.5 or .5 (or
possibly a negative version of one of those?), depending on your
implementation.  So then the question is one of whether you
interpret -.3 U as .3 or .7.  Can one say with certainty that one
way is wrong and the other right?  I mean, to me if I have the mode
"wrap" set, I sure wouldn't expect it'd implicitly mean "flip."  For
that, I'd choose a "mirror" mode instead.

-tom!

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great
prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
GDAlgorithms-list mailing list
GDA...@li...
https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list
Archives:
http://sourceforge.net/mailarchive/forum.php?forum_name=gdalgorithms-list