Pixel aspect ratio

Hallo There,

A nasty issue which is bothering me since this morning…
Maybe I’m neglecting a trivial detail but,
there’s nothing else I can figure out to try.

I imported in v4 a collada model with an animated camera,
and the same scene rendered as video out of the 3d software,
(video zize 800x600 pixel aspect squared)
they do not match :(

And it’s not just the size/proportion which doesn’t match,
the orientation of the object in space seems distorted as well…

Exporting a video sequence out of v4 and importing it into the 3d software also leads to the same result: doesn’t match.

here’s a very simple patch showing the problem.
TNX for any hints!


pixel_aspectratio.rar (277.9 kB)

well you just used aspectratio transform badly. also i couldn’t see that .dae file neither from Collada neither from Assimp so i replaced it with a simple box. also there are some difference in the animation of the video and the camera. quite strange (280.0 kB)

Micro’s spot on with the aspect ratio. Also the collada is there in the new patch- it just needs it’s transform from the mesh node attached again. I think it might be that the collada plugin is interpolating it’s key frames in a liner fashion, even when they should be soft and spliney.

Are you using Cinema 4D by any chance solanina?

I ask because ran in to something like this and it might well be c4d’s fault, and not the plugin. You can workaround by baking your keyframes before exporting.

it’s bad workaround, better not to use it normally

you are a c4d guy right Antokhio? what would you suggest instead of baking?

everyoneishappy is right.
It’s a matter of interpolation. c4d always interpolates between points in time (in a bezier style, if I remember correctly), as default. In your patch you have the linear “behavior” of the time of the movie operating on collada camera, but the camera into c4d is not linearly moving from point to point.

So here you have to set in a linear style the animation of the camera (you’ll know you got it when the handles of the spline representing the animation disappear), or reproduce interpolation of c4d camera.

I’ve seen also that FOV, FL, AW of a c4d camera are not so intuitive to reproduce in vvvv - last time I had a camera with specific, meaning not default, cam params: to make things overlap I applied skew to the geometry in vvvv.

Baking can be a good solution for many issues - exception made for particular post production tunings - but how can it solve the problem represented by FL (not in solanina’s case, but in general)?


What FL and AW stand for?

I’m not familiar with c4d, but I can assure that baking the animation is the solution for a lot of problems, when you’re trying to export 3d data to other programs.


They stand for camera’s focal length and aperture width, both helpful to reconstruct real camera lens: the higher the FL the higher the “flattening” of different planes and viceversa, as with real cameras.
So, if solanina changed e.g. AW could meet some issues, imho.

p.s. reading again above post I see I expressed my thought in a bad way, please check again, sorry.

Ah ok…
Just camera’s parameters…

Here I’m using Maya 2012 and the new Collada Exporter works really well.
But I always use to not export the Camera and use the VVVV Camera instead.
It’s better to leave it interactive for me.

But I also like the way to manage the pre-animated objects in VVVV…

It’s strange how two italians have to force themself to write in english on the forum, instead of speaking directly in italian…

But you’re are better than me in english, so I think the problem is just mine…

@ robe - tis good for the sharing of knowledge beyond your conversation.

Struggle on :)

Ti giuro che non sapevo tu fossi italiano :D
Ed il mio inglese non è migliore del tuo, in ogni caso.

@h99 Yes baking won’t help at all with perspective issues, but I’m pretty sure it’s the solution to OP’s problem.

I also think it’s nice to have interactive cam. I’m still very interested in using imported ones though. Having some set cams does not mean you can’t still switch to a live cam, or for that matter chain transformations under the baked data.

I would really love to get the C4D > collada camera workflow sorted, but maybe will start a new thread- I’m not even sure if C4D is what solanina is using so don’t want to hijack (more).
Includes an example cake