Jump to content

How to use 16.5 video volume optical flow with flipsolver


aoisaki

Recommended Posts

4 minutes ago, Noobini said:

you could study this one while you wait i guess...(i think theres' a 2nd part too)

https://www.sidefx.com/tutorials/driving-fluid-simulations-with-video-data/

 

Thank you for reply, I watched this before, but it's really different kind workflow with the 16.5 lanch event's example.....

Still can't understand how set up optical flow node and use that data for volumetrail and flipsolver

Link to comment
Share on other sites

load the video into a Cop2Sop, set that to volume slice then use a timeoffsetSOP set to $F+2 for the second input for the opticalflowSOP. You can then source that in DOPs setting the SOP to DOP as C.* -> vel.

Not sure if its accurate but a test seemed to work.

Link to comment
Share on other sites

39 minutes ago, marty said:

load the video into a Cop2Sop, set that to volume slice then use a timeoffsetSOP set to $F+2 for the second input for the opticalflowSOP. You can then source that in DOPs setting the SOP to DOP as C.* -> vel.

Not sure if its accurate but a test seemed to work.

 You can then source that in DOPs setting the SOP to DOP as C.* -> vel. 

how can I do this? I get C.bx C.by C.bz 3 volume,how to convert these 3 volume to vel.

Link to comment
Share on other sites

2 minutes ago, aoisaki said:

 You can then source that in DOPs setting the SOP to DOP as C.* -> vel. 

how can I do this? I get C.bx C.by C.bz 3 volume,how to convert these 3 volume to vel.

in SourceVolumeDOP, SOP to Drop Bindings tab, in the 'Velocity Volume' put C.*

Link to comment
Share on other sites

Many thanks for sharing your approach @mestela, it helps a lot understanding the basic idea.

Since siming takes ages on my machine, is there any way to visualize the velocity streamers properly? I already tried adding a 'Volume Trail' to the current setup, but I don't get the streamers to display correctly.

Why that's not built into the node itself?

Edited by p2or
Link to comment
Share on other sites

3 hours ago, mestela said:

oflow will generate vectors per channel of your volume, I've found it cleaner and faster to just use alpha (so convert your image to luminance for example, and copy to A), then you get clean vectors to push to dops or whatever.

optical_flow_flip.hipnc

flip_oflow.gif.22395138afbfa9186ec2e1ec90c7bbea.gif

Yeah dude - that's a heaps better approach. I'm always going to wait for your files now :) 

Link to comment
Share on other sites

That is a nice setup Matt, but the file exhibits the strangest characteristic. When I am looking through the locked camera and I move it around, the entire setup needs to re-cook. I'm not sure why the cache becomes invalidated just because of a camera move? Is that a side effect of leveraging COPs?

Link to comment
Share on other sites

Cool setup Matt!

Hey @Atom, I believe the re-ccoking is because of the pointwrangle (2nd node from top) that is transforming the points of the mocapbiped in to the camera space of cam1. I didn't know that this technique could be used to to be able to center and tumble around the target for the Geometry COP!

oflow_recook.gif.dccabc46960a1f3468f4a37902e58252.gif

Thank you for the file Matt!

Edited by galagast
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...