FLIP UVs, again..

   18204   17   5
User Avatar
Member
8 posts
Joined: 11月 2013
Offline
Hi,

I'm trying to create a flowing lava effect and I'm having some trouble texturing the meshed simulation.

After searching for a while both here and at odforce (as well as a lot of googling), I found a popular approach illustrated generously here by spev (getting UVs from emitter, re-apply them after sim):
http://forums.odforce.net/topic/15866-melting-a-textured-object-with-flips/ [forums.odforce.net]

Unfortunately, this solution doesn't work well when you are using a constantly emitting fluid, as you are quickly getting a very stretched and unusable UVs.

I've read about a blended UVs method, which uses 2 sets to achieve this, but couldn't find any more technical info about that.

I did find these amazing tests and explanations from Igor Kharitonov:
http://vimeo.com/88171240 [vimeo.com]
http://www.gerbertgosch.com/# [gerbertgosch.com]!rndtextureflowpt2/cqaj
But as a relatively new Houdini user I find them a bit hard to understand.

Anything you can share to shed some light on the matter would be greatly appreciated.

(I've uploaded a modified scene with constant emission)

Thanks.

Attachments:
flipUV_v001_ConstantEmission.hipnc (1.6 MB)

User Avatar
Member
250 posts
Joined: 2月 2013
Offline
You might ask Alejandro !
https://vimeo.com/obreadytom [vimeo.com]
User Avatar
Member
373 posts
Joined: 3月 2009
Offline
http://fx-td.com/?p=329 [fx-td.com] << I have a file here set up for noise and dual rest fields, but using uv's instead of of the rest positions is pretty easy. Probably not exactly what you need, but it's very similar to Igor's video. Hope it helps!
Ian Farnsworth
User Avatar
Member
8 posts
Joined: 11月 2013
Offline
Solitude
http://fx-td.com/?p=329 [fx-td.com] << I have a file here set up for noise and dual rest fields, but using uv's instead of of the rest positions is pretty easy. Probably not exactly what you need, but it's very similar to Igor's video. Hope it helps!
Hey Ian,
I did came across your scene in a thread about this topic, but the video seemed to focus on vorticity advection (which is super cool btw), so I didn't dig further into the hip file.
Will do so now, and try to change it to uv as you said.
Thank you for sharing!

Lewul
You might ask Alejandro !
I guess you are referring to this:
http://vimeo.com/47189796 [vimeo.com]
looks crazy good, I wonder how the uv's look on a less viscous fluid.
User Avatar
Member
8 posts
Joined: 11月 2013
Offline
Solitude
http://fx-td.com/?p=329 [fx-td.com] << I have a file here set up for noise and dual rest fields, but using uv's instead of of the rest positions is pretty easy. Probably not exactly what you need, but it's very similar to Igor's video. Hope it helps!

I tried to mix the rest fields with the vorticity fade ramp, and piping it to the point uv attribute, but the uv's I'm getting seems less accurate and stable than the displacement you did.
Is this the correct result? Should the rest ratio come into play somehow?

I'm not sure I'm doing this right, as everything gas-microsolver is way over my head at the moment, perhaps you or someone else who's familiar with this technique could point me in the right direction?

Again,
thanks a lot for the help, I do appreciate it.
User Avatar
Member
3 posts
Joined: 1月 2014
Offline
Howdy,

Having the same issue here.

Ian, thanks for the link, your files were extremely helpful (as all the links in this thread). I just can't seem to get the process to use the uvs instead of the rest positions in your setup though. Has anybody cracked the issue yet?

Thanks!
User Avatar
Member
373 posts
Joined: 3月 2009
Offline
Sorry guys, I'll see if I can do a uv setup this week sometime. Basically instead of using gas rest, I'll try using a sop solver to set uvs every X frames… and go from there.
Ian Farnsworth
User Avatar
Member
8 posts
Joined: 11月 2013
Offline
Ian, that would be awesome!

yalla,
The way I solved this issue at the moment is kinda hacky I feel..
I tried to mimic Alejandro's results by animating the uv project on the emitter, before transferring the uv's to the scatter points.
That way, at least I didn't get a consistently stretched uv's, but of course you'd still get a uv mess when the particles mix.
For a very viscous liquid I guess this is a somewhat ok approach, but it's not a solid solution.
I've attached a test I did with this approach (it's very basic, but tell me if you need more info).

Looking forward to see what Ian comes up with.

Attachments:
LavaFlow_1.mp4 (1.2 MB)

User Avatar
Member
373 posts
Joined: 3月 2009
Offline
I should be able to try it out tonight. I thought of a better solution. I have to test it though… I think it should be doable in sops after I have the rest positions from the field the same way i ha e it set up already. Just put all particles to rest position 1,and 2 set a max size for the uvs and then apply uv1 and 2 accordingly then just copy the UV over. It works in my head… Should work…
Ian Farnsworth
User Avatar
Member
373 posts
Joined: 3月 2009
Offline
Hey Guys,

Attached a file that has the concept working. You could do it on points first, but I figured what the hey… and just did it on the mesh since I already had it cached. I set up a vop in sops to put the color onto points, but I also put it into the shader with it (just quickly) set up for rendering.

Attachments:
h13_fliptricks_uv.hip (4.9 MB)

Ian Farnsworth
User Avatar
Member
8 posts
Joined: 11月 2013
Offline
Hey Ian,

Thank you very much for taking the time to create the scene.
The uv's looks great in your setup, but using the rest fields in a more velocity varied liquid seems to create an issue.
It looks like it switches uv sets abruptly based on the general velocity in the field..?
(You can see this in the attached video)

Is this a limitation of the approach, or am I missing something (very likely..)?

Attachments:
uv_test.hip (2.2 MB)
uv_test.mp4 (4.3 MB)

User Avatar
Member
373 posts
Joined: 3月 2009
Offline
Can't check out the file till tonight, but have you tried using the ‘other’ rest ratio? I had the same issue, but then I plugged in rest ratio2 I think, and it magically worked.
Ian Farnsworth
User Avatar
Member
8 posts
Joined: 11月 2013
Offline
Ok, I did do something silly.. I deleted the Cd and used only one of the uv's.. sorry about that :\

So now I only scaled the uv's to see the distortion better, and as a result I think I understand Igor's solution just a tiny bit better.
I get most of what you did in sops, but I still need to dive into the rest field dop setup and dissect exactly what you did there. so I'll do that!

Good news is that the displacement looks pretty good right off the bat and even kinda lava-like

Thanks again for your help in demystifying this approach!

Attachments:
uv_test2_Disp_1.mp4 (2.2 MB)
uv_test2_Cd_1.mp4 (3.9 MB)

User Avatar
Member
3 posts
Joined: 1月 2014
Offline
Thank you so much Ian, you're the best!

Avivbarami, those tests look really cool! Mine not quite so nice. In your test it looks like the fluid area that “settle” on the ground doesn't get the uv blending after a certain point of time (when it's almost still), while the area closer to the emitter keeps getting blended. Is this is even remotely correct? Would you mind sharing your file?

What I get is an intermittent uv blend that doesn't really look dependent on the velocity in any way (maybe I'm misinterpreting the setup).

I have attached a quick flipbook and a modified version of your file (H12.5). Call me blind and thick but I can't tell whether my dual uvs are working the way they should. Maybe you guys do?

Thanks!

Attachments:
DualUVs.mov (14.1 MB)
uv_test_166_yalla_forum.hip (2.3 MB)

User Avatar
Member
29 posts
Joined: 1月 2011
Offline
It looks like the dual uvs are working fine, but you don't appear to have any rest field data, or the positions in the rest field aren't being updated at all. Did you sim out the rest fields as well?
Harsh Mistri
Effects TD
Method Studio Vancouver
User Avatar
Member
8 posts
Joined: 11月 2013
Offline
yalla
Thank you so much Ian, you're the best!

Avivbarami, those tests look really cool! Mine not quite so nice. In your test it looks like the fluid area that “settle” on the ground doesn't get the uv blending after a certain point of time (when it's almost still), while the area closer to the emitter keeps getting blended. Is this is even remotely correct? Would you mind sharing your file?

What I get is an intermittent uv blend that doesn't really look dependent on the velocity in any way (maybe I'm misinterpreting the setup).

I have attached a quick flipbook and a modified version of your file (H12.5). Call me blind and thick but I can't tell whether my dual uvs are working the way they should. Maybe you guys do?

Thanks!

Hey yalla,

I actually didn't change anything important from Ian's setup, just did a small scale change in the vop that uses the rest values to color the points.
To copy the setup from Ian example I copied the restField sim network, and everything below the geo cache in the fluid import sop.
And then I just re-cached everything (points from sim, geo from points, rest fields).

I'm guessing that the result I'm getting is correct, because the visible uv blending is very similar to the results Igor Kharitonov showed on his site, before he implemented his solution (see above).
I still want to texture it and see how visible the switching actually is, but as Ian show in his original scene, some types of sims are really suited to this kind of setup.

Attachments:
uv_test_noChangeReally.hip (2.2 MB)

User Avatar
Member
373 posts
Joined: 3月 2009
Offline
hcmistryvfx
It looks like the dual uvs are working fine, but you don't appear to have any rest field data, or the positions in the rest field aren't being updated at all. Did you sim out the rest fields as well?

^That was my colleague logged in on this machine. oops!

It is definitely not suited for some situations, and the can occasionally require some fancier setups (igor's setup for instance). I think mine works well for flowing liquid like in my river scene, but probably not too much else.
Ian Farnsworth
User Avatar
Member
8 posts
Joined: 11月 2013
Offline
Solitude
hcmistryvfx
It looks like the dual uvs are working fine, but you don't appear to have any rest field data, or the positions in the rest field aren't being updated at all. Did you sim out the rest fields as well?

^That was my colleague logged in on this machine. oops!

It is definitely not suited for some situations, and the can occasionally require some fancier setups (igor's setup for instance). I think mine works well for flowing liquid like in my river scene, but probably not too much else.

I think using it as a displacement make this setup way more versatile, even for a non-flowing liquid.
It's very hard to see rest field blending in all of the extra detail and liquid motion.
Either way, as you said, it opens the door for further refinement, and that alone makes it useful.

Thanks again,
Aviv.
  • Quick Links