Jump to content
J.Santos

From Low Res to High - Adding detail to chunks after the sim...

Recommended Posts

Hi all!

 

I know this question has been asked and answered but I don't really seem to be able to make it work...

 

So I put together an amazing (not really) scene, with a voronoi pre-fractured geometry (a simple polygon box). I then call it into Dop's to sim and after that I should use the points of the sim to drive higher res chunks...

But how? Should I pick the chunks, get a centroid out of each of them and copy onto it a new high-res chunk? Should I go back to the original geometry (the box) and give it a lot of detail (subD's and noise/mountain/...) and break it again... but this will change the geo so the chunks will not match the previous ones...

 

Questions, questions...

 

Can someone elaborate a bit more on the process or pick my (basic) scene and add the detail so I can study the how to?

 

Because up 'til now what I've done is get all  the detail I can before sending it to Dop's (and also use a displacement shader...) and I do understand that that is a waste of resources...

 

Thanks guys!

 

Cheers!

rbd_addDetailAfterSim.hipnc

Edited by J.Santos

Share this post


Link to post
Share on other sites

Hi again guys!

 

I'm still facing the same issue... In another question someone did (about VDB fracturing) I ended up being able to understand a small step further but I'm still lost...

 

If I start with the same object and have 2 versions of it, one low Res and another High, and apply the exact same Voronoi/Cutter to it and simulate the low res... and fetch the points from the Sim... how do I go about applying them to the High Res? And how well will they match?

 

Because in my new and improved attached file I also noticed at least one error (?) -> The low res has 35 pieces and the High Res has 38... and in that model one is just a bit more subDivided... If I went more extreme, how would the model/difference behave?

 * Do you just delete/fix by hand the "bad" pieces?

 * Do you use a for each or something the like to pair a point from the low Res sim to the High res equivalent?

 

Any hints/help/examples to look at would be invaluable!

 

Cheers guys!

lowRes_sim_V02_01.hipnc

Edited by J.Santos

Share this post


Link to post
Share on other sites

I think it has something to do with your input geometry (teapot).
Maybe its a problem that its not one connected geometry.

I gave it a quick try and after converting it to a closed geometry it worked perfectly.

Heres my test.

low_high_rbd.hip

Edited by LouisXIV
  • Like 1

Share this post


Link to post
Share on other sites

Hi Martin!

 

Thanks for your input and file!

 

But the thing I want to know (and sorry if my rant mislead you) is to apply to an higher definition object the DOP information from a simpler/faster sim done to a low resolution version of the object!

 

I know that you must fetch the points from the low res sim and copy the high res pieces onto them but... not sure how! ;)

 

Cheers!

 

ps - But I got the open geo versus closed one! That was messing my number of pieces! Cool to know!

 

This is what I would like to know (and a nice Sim also ;) ), watch the breakdown and the bit where all the sim and fixes are done to the low-res and later copied into a High-Res...

 

Edited by J.Santos

Share this post


Link to post
Share on other sites

 

I know that you must fetch the points from the low res sim and copy the high res pieces onto them but... not sure how! ;)

 

 

the dop import node can do it. If you set it to "transform input geometry"

Thats whats happening in my file.

 

The left side is the lowres geo and the right side is the highres one(just added interior detail ).

Edited by LouisXIV
  • Like 1

Share this post


Link to post
Share on other sites

I'm sorry Martin... so I'm totally lost now!

 

I see that in your DOPs you use the OUT_low as RBD fractured object so yes, you are using the low res for sim...

Now things get a bit fuzzy for me...

 

You use a vertex to cusp the Normals, then the DOPS kick in, with a rest and the dopimport... but if this is below the chain shouldn't it be forcing everything to re-sim? I kind of get the "NO" because you explicitly tell to the sim that the fractured object is the low...

So in fact it is using the low res to sim and "copying"/changing the low for the high res... right? The explanation for the sim being slower is not because of the sim itself but rather the system having to calculate the extra geo/detail, right?

 

Sorry for this stream of questions, this is confusing me because I had set in my head that if did the sim, wrote it out as points and then copied the chunks it would be faster/different somehow...

Edited by J.Santos

Share this post


Link to post
Share on other sites

Hm i think you are right with all your assumptions. I think the dop import uses the lowres sim still in catch. Probably it would be smarter to write out the point representation of the pieces and copy the high res ones onto them.

I must say i m also a beginner in the rbd topic so I'm not really sure whats the "best" way.

Edited by LouisXIV

Share this post


Link to post
Share on other sites

Well... your way of addressing this works and I'll test it out (as soon as I finish some renders that are taking most of my machine away :( )

 

And about the best or not... I tried using a forEach SOP, inside of which I would do the copy from points obtained at the low res sim to the High res pieces but unfortunately I have no idea how to cycle between the pieces and I had the issue (solved thanks to you) of mismatching geo... I remember something about a sideFx tutorial (the cornfield or something) where they talked about the expression to get the number of the object...  something like getting the name but the using arg??? or something to only get the number...

Will check it out as soon as possible...

I do however (hence the "best or not") have doubts about how effective it is compared to your approach... forEachs aren't exactly fast. But everything is better than simulating all again, right?

 

I'll try to post more as soon as I know more!

 

Thanks for all your help! Keep your thoughts coming! ;)

Share this post


Link to post
Share on other sites

Just t say that I did a quick test (very empirical) and:

 

 * The hole in my geo was the culprit for the mismatching number of pieces!

 

 * As far as I could tell Martin's approach works really well - I tested it using the performance monitor but I couldn't really "get it"; it starts complaining that the High Res is draining more resources in DOP's BUT... if you reset the simulation and then you enter a specific value in the timeline, like say, 96 frames, the sim will cook for about 2 seconds (this is a light simulation ;) ) and display the result. You rewind, switch to High res, enter 96 again and... zzzzap! Instant result! So, in an empiric way, I guess this works really well :)

 

Now onto the real deal, instead of the tests... ;)

 

Here's the file with my scene and using Martin's approach! I changed the platonic to get rid of holes!

lowRes_sim_V02_03.hipnc

  • Like 1

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

×