| Hello, Have you tried using the Generate button in Ox Hair Clustering at frame 0, then rendering the scene, then saving it? This way clusters will be cached with the scene. Marsel Khadiyev (Software Developer, EPHERE Inc.) |
| Yep, the only way to make it is to have at frame 0 the sphere away from each other. I can't have them close at frame 0. http://tatprod.com/ |
| Yes, but you only need to do it once to generate the cache. You do need to generate it in a default pose. Marsel Khadiyev (Software Developer, EPHERE Inc.) |
| That means the frame 0 is ALWAYS have to be a T-pose and not a neutral pose ? Didn't think we had thoses problem with Ox_v3 Another disturbing stuff is changing viewport hair count make cluster blow up. For hair rendering i can understand that we ahve to re-generete cluster, but for viewport, that pretty anoying http://tatprod.com/ |
| Once you've cached the clusters you don't need to generate them at frame 0. It doesn't have to be a T pose as long as when you generate the clusters it is neutral, this is a part of modeling process. Changing the cluster count should automatically update them inside the viewport (and later in render). If that's not the case it could be a bug. Marsel Khadiyev (Software Developer, EPHERE Inc.) |
| Ok here's another simple scene that represent what we want with characters : - a simple tube with basic fur. - When skin is on, frame 0 is T pose. I generate cluster on this, save & reload/merge file, every thing is OK. - At this point, if i change hair count in viewport, cluster are re-generated, it's OK. - If a save a point cache and change playback graph to invert the animation (aka frame 0 is a non-T-pose) every goes wrong at this time. - If i reopen my scene, clusters are wrong - if i try yo change hair count (viewport), clusters goes sticky Every scene where frame 0 is a non Tpose, i have problems... -p- http://tatprod.com/ |
| Thank you for sending the scene. Just to clarify, viewport clusters get re-generated every time you open the scene on the frame at which the scene opens. So your description sounds like it would be the case (we'll need to address that separately). Importantly, render clusters are cached when you generate clusters and render scene first time and stay the same even after scene is reopened. Have you tried if this problem also happens when you render the scene, or is it just in the viewport? Marsel Khadiyev (Software Developer, EPHERE Inc.) |
| - Do you cache cluster at render launch or at render end ? - Do you cache cluster in max file or in Ox object ? in case of max file, that would explain why it fail at merge. - I've try my scene after doing a generate + render, but each time i reopen or merge, it re-bug. - The cluster regeneration is a real problem. How can we deal with a standard workflow : I prepare a character in an empty scene ready to be merged in X shots scene. Thanks -p- http://tatprod.com/ |
| The caching is done at render start. It is stored with the object, not the max file. I will try to experiment with merging here, perhaps something will become apparent, thank you. Marsel Khadiyev (Software Developer, EPHERE Inc.) |
| Hello there !
We are entering in full production soon and i really need to find a solution for sticky clusters. Here's a last try : http://pingus.tatprod.com/BUG/Ornatrix/Ox_Cluster3/Ox_Cluster_bug.avi I've try severals things : - File to merge is T pose in 0, all clusters regenerated & render done. - Also tryed to do it directly in scene : un load pointCache, generate then reload - It looks good in viewport & local rendering, but when i submit a sequence rendering over network (Deadline 8 / max 2016 / vray) it's still a fail I've attached a simplified scene so you can confirmed you have bug too : http://pingus.tatprod.com/BUG/Ornatrix/Ox_Cluster3/Igor_Rendu_09_Clean.zip We don't have problem on few others characters, could it comes from the stack of 4 Ox_Hair_Clustering modifiers ? Thanks ! http://tatprod.com/ |
| I downloaded the cluster bug scene and video. I'll look deeper into this tomorrow, thank you. The glued cluster must be coming from the fact that cluster indices are not cached. I assume you cache them properly as per our discussion before? Marsel Khadiyev (Software Developer, EPHERE Inc.) |
| Well the saftest way i cached them is : off point cache (so T-pose) then Clear All & Generate on all clusters modifiers from bottom to top http://tatprod.com/ |
| We're working on this problem right now. Marsel Khadiyev (Software Developer, EPHERE Inc.) |
| Hello, I think I have a build which could fix this. Could you please email me (marsel.khadiyev (at) ephere.com) so I can send it back to you for testing? Marsel Khadiyev (Software Developer, EPHERE Inc.) |
