[REQ] Splitting screen for render

HFM

Thanx for the info, I sure will look there

[can't believe I missed it the last time ]

Didier Bur

Hi,

The idea is to create a ruby that can split the screen into 4 or 9 separate screens divided over pages. This means that every screen has to have its camera position placed so, that when combined again [after rendering for example] they form 1 image.

Are you speaking of a VR object image for the final image (made of several images put together ?)

If that's the case, go here http://www.crai.archi.fr/RubyLibraryDepot/Ruby/em_cam_page.htm and d/l "vr_exporter.rb"
Hope this helps,

HFM

While pretty cool, it is not what I meant...

Updated the first post with an image to show what I mean

HFM

Okay bringing this up one more time. Is there anyone who could make this or explain how to create this in ruby [if it's very simple ]

Thanks in advance...

Fletch

just wanted to 'second' this request, as it would be helpful for KT users, as KT does not have a render region feature, if I could use SU to set up 4 or 8 regions for doing extremely large renders or something, or just for having some 'test region' cameras - that would be very cool.

Of course, I think it would have to be done with 2 point cams... like with a photo match cam for example... where when you pan around the image, the point of view does not change.

tbd

you just need to fiddle with the camera position and then add pages. the tricky part is creating the formula of dividing the screen and map that to camera parameters to keep everything aligned.

solo

I have nothing to add to the conversation as tecnically I do not know how it could be achieved, but do want to comment on the excellent render.

HFM

@solo said:

I have nothing to add to the conversation as tecnically I do not know how it could be achieved, but do want to comment on the excellent render.

Thanks! Wasn't very hard to do, I used bunkspeed's hypershot which is very quick when you want to render single object [only GI though... ]

jopsa2

hi! first time poster here, but have been lurking for a while, and have finally decided to join in on the discussion.
I believe the solution for this wouldn't involve panning, since that moves the camera position relative to the focal points, horizon line, etc, and the result would be a series of images that don't quite match up when stitching them together.

The way it could work would be using the "look around" feature, which doesn't change the perspective of the image, just the direction the camera is pointing.
The downside to this would be that you end up with some sort of "fisheye lens" render (meaning the "field of view" would be greater than the one used in each scene)

To resolve this, each scene would need to have a smaller field of view than normal, in order to compensate.

Have absolutely no clue how to code in Ruby, so I'm not sure this would even be possible to automate, but surely it can be done easily by hand?

In any case, hope this is helpful, and look forward to posting again in the future.

tomasz

The problem is that,it seams, there is no method available in SU Ruby to read 2D panning - same used for foto-matched images. If you investigate the photo-match tool - the camera stays exactly in the same position, only SU performs 'OpenGL region render'. I am afraid Google has not allowed us to read the 'cropped' image coordinates to prevent SU Free users exporting High Res images from SU (stitched from smaller parts).

It is the reason why I (SU2KT) cannot export photo matched perspectives correctly when user has panned the view.

Maybe more experienced Ruby Masters will know a workaround?

Tomasz