[Kjell]

i want to thank you both for making my head asplode

[Kjell]

Mylar (or more properly mirror-plex) is only useful for when you really want *reflections* for your comp (ie. standing on a tile floor, on ice, etc)

shadows can be extracted from keying or they can be added in later, depending on what your lighting setup is like. these days the default lighting setup for greenscreen seems to be a buncha spacelights up above for the base and adding floor/back units for shaping on your subject. result? crazy multiple shadows with one or two dominant ones. a good comp/key artist can usually minimize the lighter shadows cast from the multiple spacelights and keep at least some of the shadow from the keylight. (sometimes with spacelights i will black off the very bottom of the unit to prevent directional shadows being cast from this…the net result is less light overall but generally a more diffuse and even illumination of the cyc.

hope this helps

k

[Kjell]

@pgill 25757 wrote:

kevin,

do you reckon the lto or sait tapes would be a more solid archive system if you archive to file and then just write the file to tape in the command line rather than through the autodesk archiving system. (we currently use conform).

that is exactly what we are doing. our Conform system *was* our archive system until we decided having to tether our tape machine directly to a discreet box (that may or may not be supported in the future and that needs frequent software & hardware updates) was a Very Bad Idea.

our IT guys wrote some custom scripts to automate the process and integrate it into a database backend so searching & retrieving archives is a lot easier than it used to be. it leaves all archived jobs on the RAID as long as possible, but when the capacity gets 80% full it automatically deletes the oldest jobs first that have been verified as written to LTO. actually a handy policy because there have been more than a few occasions when a job has suddenly reactivated and i’ve found it still on the archive RAID instead of having to restore it from tape.

k

[Kjell]

@pgill 25679 wrote:

we have also been having corrupt archive problems and we are using sait2….we are moving over to lto 4 in the hope of reducing the problem…but it sounds like it is the same issue when appending to archives.

This didn’t seem to happen with the dtf’s so much…

is their a solution to avoiding these corrupt archives?

paul

actually we had several DTF2 archives get corrupted. and true to form those are the tapes that have projects that clients came calling on later. DTF2 was a horrible format in my opinion. very slow and expensive.

i think corrupted archives are a danger no matter what format you write to. i recall having resort to the ATOC several times on digibeta archives in the past.

k

[Kjell]

i second the recommendation to avoid archiving to hard disk for long-term storage. it seems like a great idea when you think of cost and speed, but you’ll change your tune when one of them fails and you lose a job you need to recover. we looked into it as a viable way of keeping things nearline, but all the anecdotal evidence said that was asking for trouble. archiving to a more stable format (disc, tape) is infinitely safer.

the LTO4 should be super speedy for you; we use it and it seldom takes more than half a day to move a job off/on the framestore through the attached NAS. the bottleneck is usually the gigE connecting the flame/smoke to the network not the NAS itself.

[Kjell]

try adding some noise and ghosting

also separate the R/G/B channels and offset them slightly to create some fringing and abberation

a/c roll bars are a nice touch

[Kjell]

um don’t know how “>Steve Hullfish, Big Idea Productions” got appended as my signature…that’s not me. will look into it. sorry all.

[Kjell]

most shots that look like nodal pans are probably not unless they were shot with a meticulously calibrated motion control rig or properly centered camera.

99% of the time when a production camera (film or HD) is set up on-set the nodal center of the optical system is NOT the nodal center of movement for the pan/tilt of the dolly or tripod head. depending on the lens and camera configuration the assistant is required to weight balance the load on the head, rather than setting it up for optical work.

for some loose 3D tracking software can sometimes “assume” a shot is a nodal pan and get away with it (but maybe not in your case).

you might also want to remove optical distortion before solving the camera move. that can often make or break a 3D track.

good luck

k

[Kjell]

the modular keyer creates an analogue to a 3D LUT for the colors you pick (for both tolerance and softness, plus any intersecting “patches” that you add). if you turn on your scope/graph HUD viewer for the keyer you can see it displayed graphically and even edit it directly.

the power of modular keyer really comes into play when you wring out it’s “modular” capabilities (a good key for a difficult scene in MK can involve a very complex MK schematic in much the same way a good key in nuke or shake can involve many nodes and operators). both the MatteBlend and ColorBlend trees in the schematic can have multiple levels which essentially stack several small selective operations on top of each other. AND..each of those levels can also have an individual matte/alpha that limits the effect of the operation to a particular part of the image (it can be a gmask-based limitation, another key or an operation from any other source on the pipe…particularly useful for edge control, etc).

[Kjell]

the modular keyer is exclusive to flame&inferno; flint and smoke only have the master keyer (plus all the other standard keyers).

one big thing about modular keyer is that it is so very tactile and responsive.

[Kjell]

that’s a shame! (he probably switched ISP’s or something…hopefully his stuff will come back online soon at a new provider)

some of his tips and tricks are archived here on fxguide:

http://www.fxguide.com/fxtips-135.html

http://www.fxguide.com/fxtips-127.html

http://www.fxguide.com/fxtips-254.html

be advised that the pincushion thing doesn’t work as expected as of v9.x because they fixed a bug that changed some behaviours

k

[Kjell]

if you are tryin to do extractions (keys, sky replacements, etc) and the CC footage is too crunchy see if you can get a flat pass out of them. be advised that the CC pass and flat pass might ntt line up exactly if they are scanned in different sessions (so you might want to get them to re-scan the CC pass in the same session)

good luck

k

[Kjell]

are the scans of color corrected footage or flat?

if the scans come from a colorist who is being heavy handed with color correction he could be crunching things to a point that accentuates the grain of the film (and the film itself could be a very high speed stock which could mean more grain….or it could be under/over exposed and the colorist is trying to squeeze detail back into the shot)

whenever i do work from 2k we almost always get a CC pass and a flat pass

k

[Kjell]

check out jake parker’s tip on “lookat”:

http://www.fxguide.com/fxtips-323.html

also, be careful…lookat doesn’t work when you have a bunch of stuff parented above the image’s main axis node

k

[Kjell]

i have not used that camera but i have used other hi-speed video cameras.

recently used the PhantomHD:

http://www.visionresearch.com/index.cfm?sector=htm/files&page=camera_HD_new

dont’ know how much you are planning on spending but that camera is pretty amazing. it shoots 1000fps at 1920×1080 and can go 2k at lower frame rates (i think 400 is the max at 2k)

those guys at Vision Research have about 10 other models to choose from too

k

[Author]

Posts