I’m in the middle of an uncharacteristic foray into animation.. with 3d people models.. and could do with a helping hand.
I thought I could export a loop of a person walking as a vrayproxy and then use that in my scene to get that VERY motion blurred look that is often used in architectural photography.
the problem is, no matter what I try I only get a very small amount of motion blur in the scene with the imported vrayproxy. In the orginal max file with the axyz person, it all works as it should.
I’m exporting 60 frames to a vrayproxy, loading it up, same camera settings, but no worky. Is there a limitation with doing it this way?
can you override the number of moblur samples with vrayproxies? anything else to take note of?
Try using the point cache modifier instead making sure you use the pc2 and not xml file format for it - you can pull off similar things. For game of thrones we got our lovely animator to make a loop on the spot, figures out how quickly the people were walking per second and animated them along splines at an appropriate speed - great for crappy crowd stuff for backgrounds.
thanks John, maybe a stupid question.. but can I use a point cache modifier on lots of objects at once? (the HD axyz character is made up of several objects and groups).
Ah so all of them have separate meshes for things like clothes / accessories? In that case no, point cache doesn’t care about meshes with multiple elements but it’ll only work on a single object at the one time. Are you likely to use a lot of characters? If vray proxy isn’t working for you what you could do is copy all the objects and collapse them to a single mesh keeping their material assignments. If you skin wrap this new mesh to the various parts of the original parts it’ll map vertex to vertex and pull the motion of your original animation, and you can then stick a point cache on this. Bit of a workaround but might help?
trying that now and looking promising. If skin wrap wasn’t so incredibly slow and non multi threaded i’d be able to tell you if it worked or not by now.
Yeah it’s doing a fair bit of work - there’s alot of bits in max that kill it - spring controllers being my current nemesis. Once you’ve got your point cache though you can do a lot with it, like edit the frame ranges or animate what frame it’s on specifically using a function curve which means if you’ve a 24 frame loop of a walk, you can animate between frame 1 and 24 so it gives you a perfect movement, but then set that curve to loop in the trackview so you get perpetual movement. You’ve also got custom start frames so if you have 50 characters, you can give them random offsets each for a big crowd of people. On game of thrones we’d maybe 20 characters for the backgrounds each skinned simply with biped. then we’d animated bip files of various different actions that could be loaded into the characters to give more variation again, then these were randomly applied and offset using maxscript to create a wally load of variation!
if anyone has any tips on the skin wrap modifier they would be gladly received. So far I have added 6 of 12 objects and it seems to be getting exponentially slower. Do I need to ‘reset’ between adding objects or anything like that?
Would love to see the results of this experiment, Peter, when you’re done. I’ve been looking into doing something just like this to get those heavily blurred people. They’re next to impossible to buy from stock libraries and are a huge pain to mask, so I think this would be a nice solution. Ideally it would be a workflow that allows you to slap on a quick mocap file to get some different motions going on, so if you can find a workaround to all this collapsing business that would obviously be best.
Really the only reason I am wanting to use point cache or vrmesh is to keep my scene file free of messy animation stuff, and because I dont understand curve editors etc. Vray proxies would be ideal for me, but sadly just now they don’t work for my purposes.
John’s idea of using the skin wrap modifier didn’t work for some reason, but might try again.
Not yet; we will probably do this for Alembic files only. The reason is that the proxy only stores a bounding box for one frame of movement and if we have to load the entire proxy to figure out the correct box, things might get too slow.
I have a vrmesh sequence of an animated pool, would I have to import this as a mesh somehow?
it has velocity info, so i guess i can do small amounts of mb with that. With your -fps trick using ply2vrmesh, i was able to adjust the amount of mb with my particles. How can I edit the fps of a vrmesh?
