Matching VrayPhysicalCam tilt settings in Nuke

[albertus]

Hi,
I´ve been given some architectural renderings that were done using the VrayPhysicalCam with the vertical tilt setting used, and have to match the nuke camera to comp some lights and people.
I have no problem matching the transformation side of the camera with the exr metadata, and I can have acces to the vertical tilt settings used in Vray.

But I have no luck trying to match the output when there is tilt/shift.
On the Nuke side, I can get *aproximate* results with the windows translate parameters, but have to rotate the camera on its X axis as well, to maintain roughly the same framing.
On the Vray side, I have the intuition that the camera tilts (rotates) as well as the *sensor* moves when changing the tilt value. Is this asumption correct?
Using some rudimental trigonometry that I find quite reasonable, I managed to make some expressions, but I am not getting the expected results. For instance, I can match the perspective just when the camera gets parallel to the ground, and the vertical lines are parallel on the frame. Easier, because I have just the window translate parameter to play with. But any intermediate value is not working for me.

Can you please give me some insight on the PhysicalCam implementation on the tilt value or a direction to look forward to make this up?
Thanks in advance and best regards

[albertus]

Sorry, I forgot to confirm that 3dsmax was used to deliver these images.
Best regards

[Svetlozar Draganov]

Tilt/Shift parameters are available in the VRayCamera node here:

[albertus]

Hi Svetlozar,
thanks, but I can´t see the attachement. I don´t have the credentials it seems.
I am not the 3d guy, I am on the comp side of the pipeline and have to match the output from Max.

But assuming from the name of the attachement and your anwer, maybe you understood that I am running Vray for Nuke? Sorry that´s not the case, I am trying to match the Vray from Max outcome with regular Nuke camera node.

Tx and best regards
T.

[Svetlozar Draganov]

Yes, the screenshot is for V-Ray camera, here it is again not as forum attachment:
http://ftp.chaosgroup.com/support/sc...6_15-04-46.jpg

I am not sure if Nuke's regular camera node support those features, might be better to refer to The Foundry's forums.

[albertus]

Thanks again Svetlozar.
Yes, I am aware of those setting both on the Vray side and the Nuke side cameras, my issue is that I cannot get both *similar* characteristics to work in the same way.
The Nuke shift indeed seems to slide the lens plane and the sensor plane, causing an offset of the image framing accordingly. On the other hand, the VrayPhysCam also seems to slide one relative to the other, but the framing of the image remains quite similar -though deformed/corrected-.
I just imagine that there is a rotation operation on the Vray camera at the same time, to maintain roughly the same viewing direction. That´s the only explanation I can imagine, and that´s why I was asking for some light on the math relation between the shift and the rotation (as my trigonometry failed).

For the particular case I originally asked, I just set it by eye and worked, but it will be nice if I can figure out a systematic approach for future cases.
Thanks and best regards

[albertus]

Sorry, I see now that maybe is unclear my point and naming.
I am refering to Nuke standard camera, the vanilla one.
And when I refer to VrayPhysCam, I mean the one inside max.
Sorry and thanks.

[823hal]

Hi,

Did you get any success with converting a VrayPhysCam from max to Nuke?

I have the exact same problem as you.

I can't get the horizontal/vertical shift to match in nuke.

Any ideas?

[albertus]

Hi 823hal,
no, I hadn't luck with this.
I understand that I have to feedle with the projection matrix, but that's too much for my knowledge right now.
Please come back if you have some success at all, as this will be very useful for me too.
Best regards