rendering nulls?

[wesware]

Hi there,


Well here I go talking about something I haven't the faintest clue about :)

In Mograph, you could place all the nulls as a child of a Fracture object, the use a tracer to create a spline, then use a cloner to place balls or whatever on that spline. I'm having trouble getting the balls to sit just on the points but I guess you could do a current state to the tracer to get the spline then use a plug in or expresso to get the balls where they need to be.

Hope that helps.

[wesware]

I guess there might also be a way to throw all the nulls into a group in expresso and then assign a cube or sphere to them as a whole, I just don't know...my head hurts.

[LucentDreams]

wioclest way I can think of simon won't give you spheres :/


select all nulls
in the character menu under conversion use the Convert to joints command
Set the joitns to Axis mode, and set the length to say 10m.
Int he bone display parameter just below the length setting set it to box and then to polygon.
Render.

[Simon Wicker]

kai's method seems quite neat - i just need to get my head around how the bones are being aligned so i can get the tips of the polys this method generates pointing down at the nulls.

cheers, everyone for your pointers!

[LucentDreams]

sorry I hadn't thought about them being generated from the axis instead of on the axis I guess you could offset them all by half their Length value in the Z axis (the joints by default should be aligned on Z.)

guess this solution might not be that feasible

[LucentDreams]

haha so hows this for really bizarre retarded million step workflows.

after converting to the joints instead of setting up the joitns like I said there is another ocmmand that coverts joints to splines, then once they are all spline objects, use theocnnect command to make a single spline object. then use mograph cloner to clone a sphere or disc to the points on the spline object.


If you need more info or details simon you know where you can find me for more immediate response.

[JoelOtron]

Just an idea.

Stormtracer? create a marker image in photoshop with an alpha. If possible, put your nuls nin one group. Apply stormtracer to that group. With the right settings you will have all your nulls rendered with whatever marker image you want.

Damn--nevermind--ST doesnt work on nulls--thought it did.

[ChrisCousins]

Hi chaps - just another method thrown into the mix, using TP. It just spawns the right number of particles at frame zero, then each particle tracks the position of each null. Then you can add geometry the normal way.

Hope this helps - C

[bobtronic]

why don't you make simply a "render as editor" rendering?

cheers,
Bob

[Simon Wicker]

why don't you make simply a "render as editor" rendering?

cheers,
Bob

the results need to be sub-pixel accurate to see there is no sliding - this needs to be composited over the original footage in ae to fully check this out so you not only need good anti-aliasing but also an alpha channel.

rendering as editor doesn't use any anti-aliasing.

there are a few things cinema would need to be usable in this respect (several of which i have previously mentioned to tilo) but of course small stuff that only benefits me and sebastien doesn't really bubble to the surface when other higher profile stuff like CA needs fixing.

cheers, simon w.

[bobtronic]

the results need to be sub-pixel accurate to see there is no sliding - this needs to be composited over the original footage in ae to fully check this out so you not only need good anti-aliasing but also an alpha channel.

rendering as editor doesn't use any anti-aliasing.

there are a few things cinema would need to be usable in this respect (several of which i have previously mentioned to tilo) but of course small stuff that only benefits me and sebastien doesn't really bubble to the surface when other higher profile stuff like CA needs fixing.

cheers, simon w.

ah, I see

cheers,
Bob

[Per-Anders]

Here's a quick way:

Create a Mograph Fracture object
Put all your nulls into the fracture object
Create a MoGraph cloner object
Put a sphere in the cloner object
With the cloner oebjct selected create an Inheritance effector
In the inheritance effector set the "Object" link to the Fracture object, then check "Morph Motion Object"
In the cloner up the number of clones to match the number of nulls

Second mograph method (using dark arts knowledge)

Create a MoGraph fracture object
Put all your nulls into the fracture object
Create a mograph Cache tag on the fracture object
Bake the object for the animation
Create a mograph cloner object and put a sphere in it
Drag the Cache tag from the fracture object onto the cloner object

And finally you could use a COFFEE script to do this relatively easily too.

[ernia]

Hello Simon,

Wouldn't it be easier to export all of your trackers together under one obj file out of Syntheyes? Apparently each tracker comes in as an x-plane.

Just a thought.

ernia

[basti]

hey simon,

in a normal matchmove workflow you would render out a wireframe
of the rough scene geometry.
rendering just the nulls only gives you the calculated 3d positions
of the markers which should already be perfect even before exporting
from your tracking software. instead use these points to build a rough
scene geometry and render it as wireframe (cel render with motionblur).
this is the best way to check a tracking :)


another workaround if you really need the points:
render as editor with 4x the resolution and the bg set to black and scale
the rendering down in your comp tool. (self made AA ;) )

or use the renderer in syntheyes to render the 3d solve directly on your
plate - although it's kind of slow with high AA settings


one hint about matchmoving: you wrote 'sometimes.. hundrets of points'
sounds like you used automatic tracking which is often way less accurate
than manual tracking.
it's always better to have only a few manual points (6-10) which are tracked
with a precision of 100% than hundrets of points with a lot of little errors.
these autotracking errors will get smaller because of the huge number of
trackers but the errors are still there and can be seen in the final solve
- especially when working in hd or feature film res (2k)!

basti

[Simon Wicker]

thanks for all the input - lots to test out now. clearly a lot of this is down to opening up syntheyes for the first time yesterday and then trying to figure out how everything works for a shot that needs to be finished next week!

unfortunately the syntheyes manual leaves a lot to be desired and i'm sure that while this could be cleared up with a bit of trial and error this extra time is something you sometimes can't afford.

i tried following basti's advice of manually tracking some points to perform the solve but despite my best efforts the residual error in my manually tracked scene is always worse than a cleaned up auto solve. grrr.

i can see why companies have people that specialise in this because it goes right over my head. i still can't work out why the trackers in syntheyes seem to track everything automatically when you step through or play the shot but they also have a key frame button. if they've already tracked the feature why the need for keyframes? and if the keyframes help with the solve should i key frame every frame or just a few?

right, we are now going a bit off-topic here so i'll keep testing stuff out and thanks for everyones help. i'll report back when i've played some more!

cheers, simon w.

[marshalartist]

I am not the most experienced with Syntheyes, but what you have to watch out for is all the redundant tracking points as basti said bring the number right down and for some of them you will have to determine in and out points so it knows when to pay attension to them and when not to.
Sytheyes must have the worst user interface of any program, I recently upgraded I have not used it in a couple of years and was shocked to see it had not changed.
What are you trying to achieve? it does not sound like you are trying to matchmove a piece of footage.

[Simon Wicker]

i am matching some footage but it is some rather dark digibeta from the third season of doctor who, not HD or film.

it is being matchmoved by someone else now who will hopefully do a better job of it than i will but i would still like to know for my own satisfaction how syntheyes (or any other matchmover) works as it seems like quite a useful program for photomodelling and such like.

since this has been a piece of problematic footage for myself then it seems ideal for me to practice on now, chipping away until i get a good result.

syntheyes is pretty hard to get to grips with, the lack of a decent manual doesn't help, many buttons and options are described in that typical manual way: 'set key - press to set a key'.

hmm. how about telling me why i need to set a key and what it does for the matchmoving process.

i'm sure that it makes sense to a TD who does this for a living but i need something that is actually going to help me get the right result.

cheers, simon w.

[marshalartist]

So I take it you were trying to render the nulls imported from Syntheyes. I do not see how this will help you. You simply just need to place your geometry and see if it has worked. As far as getting better results from Sytheyes I don't know, you could check out pixelcorps at DV garage I think they have some Tutorials, they certainly have the right people to give you pointers. Good luck

[basilisk]

I must say this makes me a bit reluctant to invest in Syntheyes. While it has good credentials, and is relatively cheap, it sounds a bit of a headache to the occasional user. As someone who got OK results out of PFHoe by just clicking the start button, do I really need the power? I notice PFHoe2, and PFHoePro are now out, now working with HD footage and multiple exports, and user defined features to track. At about half the price of Syntheyes.
Probably too early to tell if they are any good, and no use for 2k resolution.

[Simon Wicker]

well i'm sure that syntheyes works and works well (you wouldn't get basti using it as a tool in his matchmoving arsenal otherwise) but it is certainly more like maya than cinema. for me this means that the complexity is exposed to the user rather than hidden. this makes for a slightly more complex learning curve for the application. i prefer an application that grows in complexity as your knowledge grows.

i've been playing around some more with syntheyes and have produced a better manual track than before by being extremely exact about which tracks i use for the final solve. as basti said you are better off having a few exact tracks and making sure you delete all tracks with any errors in them.

for straightforwards tracks i don't doubt you could just use the auto-tracker and use the results in cinema. the problem i am having is that i'm trying to be too ambitious without really knowing the app well enough. i have to solve the camera, extend the move, and create match geometry for the set so i can camera map the set back for the extended move.

cheers, simon w.

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.