Jump to content
tfreitag

Packed Prims and UVs

Recommended Posts

hi everyone, I making a fishccrowd with copy- sopsetup with packed prims. Every fish have an uv-attribute. But after setting "packed before copying", I loosing all my attributes.

How can get back my UV. Since every fish is stored in a single point, I cant transfer it back.

I will write out the hole crowd in alambic file (alembic ropout).

Any solution out ther?

 

Thanks.

Share this post


Link to post
Share on other sites

Transfer the uv's where? Even though they're packed, rendering in mantra should be fine.

Share this post


Link to post
Share on other sites

Yes, you won't see the uv attribute information in MMB "geometry info" pop-up after packing, but Mantra unpacks primitives upon rendering, thus restoring the uvs.

Share this post


Link to post
Share on other sites

There is an "Unpack" sop if you wish to access the data at some point after your copy, or you should just not pack it in the first place,

the entire point of packed geometry is the assumption that the data isn't going to change, I think.

Share this post


Link to post
Share on other sites

take care if you packed them more than once, your uvs will be deeper into packed hierarchy, in this case you will need more iteration on the unpack node iteration parameter.

 

 

cheers

Share this post


Link to post
Share on other sites

hey big thanks everyone for help. I think i missunderstood the concept of packed prim. Its not stored in its "packed" point but inside.

So everything works fine.

But now I have  other big trouble with packed prim -> alembic -> to 3dsmax/vray.... (20.000 birds with 3.500 faces per birds, one bird

so its my first job with houdini. Until now it was fun...but now its the hell to get everthing out to 3dmax.

 

 

https://www.sidefx.com/index.php?option=com_forum&Itemid=172&page=viewtopic&p=172361#172361

 

happy about every tips with packed prims/alembic

Share this post


Link to post
Share on other sites
Guest tar

Hey Tom - not sure for 3DsMax but previously for .abc files exporting them out of Maya in Ogawa format made things fly, in comparison to exporting from Houdini.

 

Do you have a file we can test this with?

Share this post


Link to post
Share on other sites

hey marty :)

yes I tried some minutes ago the ogawa type. but If I loading in houdini it looks totally damaged -> some point artefacts popping out. but also default type some fishes are jumping/jittering. I attached scenefile (with baked out pointcloud and fish seq) and I also attached one kind of buggy swarm.abc for testing.  The only thing whats nice about alembic (if "use instance where possible" checked in alamebic sop) in in conjunction with packed prim (copy sop) it became VERY small :) ...

but loading in (in 3dsmax and houdini) it becomes a hell  :(

thanks for any tips for getting. i getting a little nervous becuase I have to get it to render on the client farm tomorrow) oh-my-god

 

this is my alembic test file:

https://db.tt/V6qpZlY0

 

here is my hipfile:

https://db.tt/tYHb06dB

 

thanks in advanced. tom

Share this post


Link to post
Share on other sites
Guest tar

Unfortunately no luck yet with quick testing export Owaga out of Maya - will have to remember the 'weird trick' that was used. 

 

When it worked playback goes from 1fps to 30fps etc. IIRC it's to do with the way timestamps are written into alembic file. The slower version has each frame written whilst the fast one has a kind of compression.  

 

Will test more if I have the time. Can you please try H13 too - as it handled .abc files a tad differently.

Share this post


Link to post
Share on other sites

woow...thanks for quick response and help.

 

in h13 is the same. if move the timebar with hold mouse and left and right it seems ok. but if frame by frame, its jumping like hell...

 

edit: i also try to check off in alembic rop the "use visibility" ... no dissapperaing, but  jittering is still there

Edited by tfreitag

Share this post


Link to post
Share on other sites
Guest tar

Using abcinfo shows the TimeSamples 'issue' I believe; see below: http://www.sidefx.com/docs/houdini14.0/ref/utils/abcinfo

 

 

abcinfo -t /fishpack_1_baracuda.abc 

 (Ogawa), version: 10502 (2 timings)

   TimeSample[0]      1 samples:  t0=0 (frame 0.0 @ 24fps)

   TimeSample[1]    751 samples:  t0=3.33333 (frame 80.0 @ 24fps)

 

Edited by tar

Share this post


Link to post
Share on other sites

thanks marty.

i have an offset of 100 frames (in alembic rop out). so th seq goes from 100-850 at 30fps. (also this setting in hipfile i sent)

Yes a friend tried in modo, and there is same jittering. Maybe a point order or some timesample issue.

Is this bug ? ..or something that result in wrong setup?

Share this post


Link to post
Share on other sites
Guest tar

I don't think it's a bug, but the goal, I believe,is to get those time samples down to much less - I'm super busy but some tests are needed! 

 

Try writing out very simple animating geo and check the time samples - sorry I don't have an exact answer but a bit of RnD might be the only answer!

 

Edit: Looks like I'm totally wrong :)  MountainSOP on a Box for 25 frames does:

 

abcinfo -t Box.abc 

 (Ogawa), version: 10502 (2 timings)

   TimeSample[0]      1 samples:  t0=0 (frame 0.0 @ 24fps)

   TimeSample[1]     25 samples:  t0=0.0416667 (frame 1.0 @ 24fps)

 

Edited by tar

Share this post


Link to post
Share on other sites

25 samples:  t0=0.0416667 ...What is this telling me. Is this 25 time sample with time difference of 0.041sec ?

Share this post


Link to post
Share on other sites
Guest tar

Yup 4/100 of a second - the fish have 3.3??? something to investigate

Share this post


Link to post
Share on other sites
Guest tar

I noticed it's 1 million polys - I'm not sure I've tested .abc to that level. Perhaps this is what happens at 1 million polys.

 

Edit: got the same slowdown with a test of boxes   looks like the boxes test are fast 

https://www.dropbox.com/s/ewn7meggfvswyo6/BoxTest_Archive.zip?dl=0

 

Edit 2: With some more light testing, substituting a box for the fish - it looks like using 'Pack Geometry Before Copying' in the CopySop makes Alembic slow to read in.

Edited by tar

Share this post


Link to post
Share on other sites

marty, thanks :D for investing time to help me!! ...after a bad night, I tried totally simple setup...only copysop with animated fishgeoseq (no randomoffset) and the pointcloud....this seems to work...so I guess I have made something very evil in my setup...so maybe my (frame)timecounter stamping into timeshift, for offseting and speed-ups my fishes....but in my old setup have only integer timecounter (integer timeshift for better instanceing).

 

I will do RnD some more...and I will post it  if I found some results.

 

..I tried also  FBX/ obj seq and its nogo.. no instanceing....so alembic have 50mb for 300frames and fbx have 2GB(for hole sequence). My birds (30.000) is 5gb as *.abc and as bgeo its 3gb per frame (nearly 1TB compare to 5gb alembic).

In my sitiation alembic is the only way I quess...

Edited by tfreitag

Share this post


Link to post
Share on other sites

thx for the file. Its much faster now, but I guessed with packed primitiv it uses every frame(29 frames on fishcycle) of fish animtion as an instance.

But with simple timeshift stamping expression the jittering is gone :)

Viewport/render speed issue: I thing the most importer have problem to read the instance hierachrie (alembic instances), i guess.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×