Re: Stereo Render differences between versions

[Graham D Clark]

The decade plus old backplate update issue (why I needed to use SCOPs to drive 
them, not ICE, which Softimage replicated in their stereo rig) I believe was 
fixed recently, is perhaps why Soft 2015 matches Maya, dunno, I roll my own in 
each and match, highly recommended.

Graham D Clark, Head of Stereography, Deluxe 3D dba Stereo D
phone: why-I-stereo
http://www.linkedin.com/in/grahamclark

> On Jun 16, 2014, at 7:22 PM, Andre Zazzera  wrote:
> 
> Hey guys - we're not really sure what the problem was coming down to.   
> Unfortunately I've been off-site and this is a high security job (nothing 
> goes outside the building), so I've been trying to talk my guys through it 
> over the phone.  
> 
> We're using a standard stereo rig, off-axis, with the zero-parallax plane 
> being set manually.   Nothing's moving or animated.  The most vexing problem 
> is that the cameras do different things in Softimage 2013 and 2014 than they 
> do when we open in 2015 - which blows my mind.  I don't think I've ever had 
> Softimage break something on me from version to version (well, maybe 1 or 2 
> other things).  What's weirder is that the Maya imports were matching the 
> Softimage 2015 version.
> 
> We found a workaround, though, to at least get this thing out the door.   We 
> just created a couple single cameras, matched all the film aperture and focal 
> length settings and all that jazz, and just constrained those cameras to the 
> original left and rights, treating them as straight parallel cameras, and 
> then we handled the convergence in Nuke to match the original off-axis 
> converged renders..
> 
> It matches alllmost perfectly (off maybe half a pixel due to rendering in a 
> different position, but otherwise great).
> 
> Thanks for the tips, guys!   I'm definitely going to investigate further when 
> I'm back on-site.
> 
> 
>> On Sun, Jun 15, 2014 at 11:15 PM, Ed Manning  wrote:
>> convergence? 
>> 
>> How are you controlling that?  If through a constraint system, maybe 
>> something to check as well.
>> 
>> Or maybe even more basic -- parallel vs. converged cameras?
>> 
>> 
>> 
>> 
>> 
>> 
>>> On Fri, Jun 13, 2014 at 4:23 AM, Tim Leydecker  wrote:
>>> The usual suspects for trouble in camera transfer between Maya and 
>>> Softimage,
>>> depending on the Crosswalk/*.fbx/or else version used:
>>> 
>>> *Field of View Angle calculation, Horizontal or Vertical? Can also give 
>>> grief with 3DSMax.
>>> 
>>> *Default XSI Picture Ratio vs. Maya Film Aspect Ratio,  Softimage defaults 
>>> to a 16:9 ratio, Maya to a 3:2.
>>> 
>>> When I create a Stereo rig in both Maya and XSI 2014, they differ in Angle 
>>> of View,
>>> Film Aspect Ratio and a Film Offset in Maya of +/- 0.017 (mm?) vs a Optical 
>>> Center Shift of
>>> +/- -0,0043 inch in XSI. Also note the overall 10:1 scene unit related 
>>> values...
>>> 
>>> Enough to make my head hurt by itself already. Which is why I generally 
>>> don´t like Stereo much anyway. 
>>> 
>>> I´d check for inch/Millimeter issues in values as well as rounding errors 
>>> and Film Aperture woes.
>>> 
>>> 
>>> Cheers,
>>> 
>>> 
>>> tim
>>> 
>>> 
>>> 
>>> 
>>> 
>>> 
>>> 
>>> 
>>> 
>>> 
>>> 
>>> 
>>> Am 13.06.2014 01:32, schrieb Athanasios Pozantzis:
 just speculating here...
 is there a film shift (tilt shift) or film offset setting in the mix?
 that could be off if everything else is spot on
 
 my 2 cents
 
 On 12 Jun 2014, at 18:16, Andre Zazzera  wrote:
 
> Hey all,
> 
> In an effort to start transitioning over to Maya, *sigh*, we've been 
> doing look dev on our current project in Softimage 2013 with the intent 
> to render in Maya.   We got approval on the stereo from client based on 
> our Softimage animation, but now we're having a big discrepancy between 
> the stereo renders in Softimage and the stereo in Maya - Maya has a biiig 
> shift in the depth of the renders and it looks like a massive difference 
> in the camera interaxial.
> 
> But we're measuring, and all the distances from camera to camera and 
> camera to subject are exactly the same in Maya as they are in Softimage, 
> and all the camera settings are the same.
> 
> So we opened the scene in Softimage 2015 and exported to Maya from there, 
> and then they match each other. Great!  But here's the thing - we checked 
> and the renders from Softimage 2015 don't match the renders from 
> Softimage 2013. So we tried in 2014, and those match the originals from 
> 2013.   
> 
> Did something change between 2014 and 2015 in the way Softimage stereo 
> cameras work?  
> 
> For the moment we're just trying to eyeball the cameras to try and get 
> something that matches the approved shot, but something just isn't making 
> sense.   I don't understand how cameras in the same place in space could 
> yield different results.
> 
> Do you guys have any insight?
> 
> Thanks!

Re: Stereo Render differences between versions

[Andre Zazzera]

Hey guys - we're not really sure what the problem was coming down to.
Unfortunately I've been off-site and this is a high security job (nothing
goes outside the building), so I've been trying to talk my guys through it
over the phone.

We're using a standard stereo rig, off-axis, with the zero-parallax plane
being set manually.   Nothing's moving or animated.  The most vexing
problem is that the cameras do different things in Softimage 2013 and 2014
than they do when we open in 2015 - which blows my mind.  I don't think
I've ever had Softimage break something on me from version to version
(well, maybe 1 or 2 other things).  What's weirder is that the Maya imports
were matching the Softimage 2015 version.

We found a workaround, though, to at least get this thing out the door.
We just created a couple single cameras, matched all the film aperture and
focal length settings and all that jazz, and just constrained those cameras
to the original left and rights, treating them as straight parallel
cameras, and then we handled the convergence in Nuke to match the original
off-axis converged renders..

It matches alllmost perfectly (off maybe half a pixel due to rendering in a
different position, but otherwise great).

Thanks for the tips, guys!   I'm definitely going to investigate further
when I'm back on-site.


On Sun, Jun 15, 2014 at 11:15 PM, Ed Manning  wrote:

> convergence?
>
> How are you controlling that?  If through a constraint system, maybe
> something to check as well.
>
> Or maybe even more basic -- parallel vs. converged cameras?
>
>
>
>
>
>
> On Fri, Jun 13, 2014 at 4:23 AM, Tim Leydecker  wrote:
>
>>  The usual suspects for trouble in camera transfer between Maya and
>> Softimage,
>> depending on the Crosswalk/*.fbx/or else version used:
>>
>> *Field of View Angle calculation, Horizontal or Vertical? Can also give
>> grief with 3DSMax.
>>
>> *Default XSI Picture Ratio vs. Maya Film Aspect Ratio,  Softimage
>> defaults to a 16:9 ratio, Maya to a 3:2.
>>
>> When I create a Stereo rig in both Maya and XSI 2014, they differ in
>> Angle of View,
>> Film Aspect Ratio and a Film Offset in Maya of +/- 0.017 (mm?) vs a
>> Optical Center Shift of
>> +/- -0,0043 inch in XSI. Also note the overall 10:1 scene unit related
>> values...
>>
>> Enough to make my head hurt by itself already. Which is why I generally
>> don´t like Stereo much anyway.
>>
>> I´d check for inch/Millimeter issues in values as well as rounding errors
>> and Film Aperture woes.
>>
>>
>> Cheers,
>>
>>
>> tim
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> Am 13.06.2014 01:32, schrieb Athanasios Pozantzis:
>>
>> just speculating here...
>> is there a film shift (tilt shift) or film offset setting in the mix?
>> that could be off if everything else is spot on
>>
>>  my 2 cents
>>
>> On 12 Jun 2014, at 18:16, Andre Zazzera  wrote:
>>
>>   Hey all,
>>
>> In an effort to start transitioning over to Maya, *sigh*, we've been
>> doing look dev on our current project in Softimage 2013 with the intent to
>> render in Maya.   We got approval on the stereo from client based on our
>> Softimage animation, but now we're having a big discrepancy between the
>> stereo renders in Softimage and the stereo in Maya - Maya has a biiig shift
>> in the depth of the renders and it *looks *like a massive difference in
>> the camera interaxial.
>>
>> But we're measuring, and all the distances from camera to camera and
>> camera to subject are exactly the same in Maya as they are in Softimage,
>> and all the camera settings are the same.
>>
>> So we opened the scene in Softimage 2015 and exported to Maya from there,
>> and then they match each other. Great!  But here's the thing - we checked
>> and the renders from Softimage 2015 don't match the renders from Softimage
>> 2013. So we tried in 2014, and those match the originals from 2013.
>>
>> Did something change between 2014 and 2015 in the way Softimage stereo
>> cameras work?
>>
>> For the moment we're just trying to eyeball the cameras to try and get
>> something that matches the approved shot, but something just isn't making
>> sense.   I don't understand how cameras in the same place in space could
>> yield different results.
>>
>> Do you guys have any insight?
>>
>> Thanks!
>> Andy
>>
>>
>>
>

Re: Stereo Render differences between versions

[Ed Manning]

convergence?

How are you controlling that?  If through a constraint system, maybe
something to check as well.

Or maybe even more basic -- parallel vs. converged cameras?






On Fri, Jun 13, 2014 at 4:23 AM, Tim Leydecker  wrote:

>  The usual suspects for trouble in camera transfer between Maya and
> Softimage,
> depending on the Crosswalk/*.fbx/or else version used:
>
> *Field of View Angle calculation, Horizontal or Vertical? Can also give
> grief with 3DSMax.
>
> *Default XSI Picture Ratio vs. Maya Film Aspect Ratio,  Softimage defaults
> to a 16:9 ratio, Maya to a 3:2.
>
> When I create a Stereo rig in both Maya and XSI 2014, they differ in Angle
> of View,
> Film Aspect Ratio and a Film Offset in Maya of +/- 0.017 (mm?) vs a
> Optical Center Shift of
> +/- -0,0043 inch in XSI. Also note the overall 10:1 scene unit related
> values...
>
> Enough to make my head hurt by itself already. Which is why I generally
> don´t like Stereo much anyway.
>
> I´d check for inch/Millimeter issues in values as well as rounding errors
> and Film Aperture woes.
>
>
> Cheers,
>
>
> tim
>
>
>
>
>
>
>
>
>
>
>
>
> Am 13.06.2014 01:32, schrieb Athanasios Pozantzis:
>
> just speculating here...
> is there a film shift (tilt shift) or film offset setting in the mix?
> that could be off if everything else is spot on
>
>  my 2 cents
>
> On 12 Jun 2014, at 18:16, Andre Zazzera  wrote:
>
>   Hey all,
>
> In an effort to start transitioning over to Maya, *sigh*, we've been doing
> look dev on our current project in Softimage 2013 with the intent to render
> in Maya.   We got approval on the stereo from client based on our Softimage
> animation, but now we're having a big discrepancy between the stereo
> renders in Softimage and the stereo in Maya - Maya has a biiig shift in the
> depth of the renders and it *looks *like a massive difference in the
> camera interaxial.
>
> But we're measuring, and all the distances from camera to camera and
> camera to subject are exactly the same in Maya as they are in Softimage,
> and all the camera settings are the same.
>
> So we opened the scene in Softimage 2015 and exported to Maya from there,
> and then they match each other. Great!  But here's the thing - we checked
> and the renders from Softimage 2015 don't match the renders from Softimage
> 2013. So we tried in 2014, and those match the originals from 2013.
>
> Did something change between 2014 and 2015 in the way Softimage stereo
> cameras work?
>
> For the moment we're just trying to eyeball the cameras to try and get
> something that matches the approved shot, but something just isn't making
> sense.   I don't understand how cameras in the same place in space could
> yield different results.
>
> Do you guys have any insight?
>
> Thanks!
> Andy
>
>
>

Re: Stereo Render differences between versions

[Tim Leydecker]

The usual suspects for trouble in camera transfer between Maya and 
Softimage,

depending on the Crosswalk/*.fbx/or else version used:

*Field of View Angle calculation, Horizontal or Vertical? Can also give 
grief with 3DSMax.


*Default XSI Picture Ratio vs. Maya Film Aspect Ratio,  Softimage 
defaults to a 16:9 ratio, Maya to a 3:2.


When I create a Stereo rig in both Maya and XSI 2014, they differ in 
Angle of View,
Film Aspect Ratio and a Film Offset in Maya of +/- 0.017 (mm?) vs a 
Optical Center Shift of
+/- -0,0043 inch in XSI. Also note the overall 10:1 scene unit related 
values...


Enough to make my head hurt by itself already. Which is why I generally 
don´t like Stereo much anyway.


I´d check for inch/Millimeter issues in values as well as rounding 
errors and Film Aperture woes.



Cheers,


tim












Am 13.06.2014 01:32, schrieb Athanasios Pozantzis:

just speculating here...
is there a film shift (tilt shift) or film offset setting in the mix?
that could be off if everything else is spot on

my 2 cents

On 12 Jun 2014, at 18:16, Andre Zazzera > wrote:



Hey all,

In an effort to start transitioning over to Maya, *sigh*, we've been 
doing look dev on our current project in Softimage 2013 with the 
intent to render in Maya.   We got approval on the stereo from client 
based on our Softimage animation, but now we're having a big 
discrepancy between the stereo renders in Softimage and the stereo in 
Maya - Maya has a biiig shift in the depth of the renders and it 
/looks /like a massive difference in the camera interaxial.


But we're measuring, and all the distances from camera to camera and 
camera to subject are exactly the same in Maya as they are in 
Softimage, and all the camera settings are the same.


So we opened the scene in Softimage 2015 and exported to Maya from 
there, and then they match each other. Great!  But here's the thing - 
we checked and the renders from Softimage 2015 don't match the 
renders from Softimage 2013. So we tried in 2014, and those match the 
originals from 2013.


Did something change between 2014 and 2015 in the way Softimage 
stereo cameras work?


For the moment we're just trying to eyeball the cameras to try and 
get something that matches the approved shot, but something just 
isn't making sense.   I don't understand how cameras in the same 
place in space could yield different results.


Do you guys have any insight?

Thanks!
Andy

Re: Stereo Render differences between versions

[Athanasios Pozantzis]

just speculating here...
is there a film shift (tilt shift) or film offset setting in the mix?
that could be off if everything else is spot on

my 2 cents

> On 12 Jun 2014, at 18:16, Andre Zazzera  wrote:
> 
> Hey all,
> 
> In an effort to start transitioning over to Maya, *sigh*, we've been doing 
> look dev on our current project in Softimage 2013 with the intent to render 
> in Maya.   We got approval on the stereo from client based on our Softimage 
> animation, but now we're having a big discrepancy between the stereo renders 
> in Softimage and the stereo in Maya - Maya has a biiig shift in the depth of 
> the renders and it looks like a massive difference in the camera interaxial.
> 
> But we're measuring, and all the distances from camera to camera and camera 
> to subject are exactly the same in Maya as they are in Softimage, and all the 
> camera settings are the same.
> 
> So we opened the scene in Softimage 2015 and exported to Maya from there, and 
> then they match each other. Great!  But here's the thing - we checked and the 
> renders from Softimage 2015 don't match the renders from Softimage 2013. So 
> we tried in 2014, and those match the originals from 2013.   
> 
> Did something change between 2014 and 2015 in the way Softimage stereo 
> cameras work?  
> 
> For the moment we're just trying to eyeball the cameras to try and get 
> something that matches the approved shot, but something just isn't making 
> sense.   I don't understand how cameras in the same place in space could 
> yield different results.
> 
> Do you guys have any insight?
> 
> Thanks!
> Andy

Stereo Render differences between versions

[Andre Zazzera]

Hey all,

In an effort to start transitioning over to Maya, *sigh*, we've been doing
look dev on our current project in Softimage 2013 with the intent to render
in Maya.   We got approval on the stereo from client based on our Softimage
animation, but now we're having a big discrepancy between the stereo
renders in Softimage and the stereo in Maya - Maya has a biiig shift in the
depth of the renders and it *looks *like a massive difference in the camera
interaxial.

But we're measuring, and all the distances from camera to camera and camera
to subject are exactly the same in Maya as they are in Softimage, and all
the camera settings are the same.

So we opened the scene in Softimage 2015 and exported to Maya from there,
and then they match each other. Great!  But here's the thing - we checked
and the renders from Softimage 2015 don't match the renders from Softimage
2013. So we tried in 2014, and those match the originals from 2013.

Did something change between 2014 and 2015 in the way Softimage stereo
cameras work?

For the moment we're just trying to eyeball the cameras to try and get
something that matches the approved shot, but something just isn't making
sense.   I don't understand how cameras in the same place in space could
yield different results.

Do you guys have any insight?

Thanks!
Andy