Forum Replies Created
-
Posts
-
hi eric
how will you be using the fire elements? in a photoreal comp? as graphic elements.
you are kinda lucky with fire because in a lot of cases they are almost self-matting. bring them in as a F/M layer in action and CC your matte, crushing the range a bit. then experiment with the blend mode of the clip (change from “blend” to “simple add”, etc). then try adding another layer as unCC F/M and use that as a muliply image on top. depending on the comp you’ll have to monkey with it.
you’l probably have to do it in a few layers (2-3) and build the density of fire up slowly. the big obstacle is trying to keep the subtle darker qualities (the blacks and browns) of the clip while keeping the highlights from blowing out.
all of this is assuming that your fire footage/stock was shot on black.
hope this helps
good luck
k
if you’ve got smoke on the same box you can use the built-in vector/waveform scopes …they’re pretty accurate
however, nothing can beat an external vectorscope/waveform monitor in the suite. they are instantly responsive and you can set them up to give you alarms if you go out of gamut. don’t skimp and get the component model, you’ll need a full SDI version.
if you have more than one suite and dont’ want to spend the $5k per suite for the scopes you can get a facility-wide scope that is routeable over IP…meaning it lives in your master control room and you can patch the output of that device to any PC over your network (this is handy for HD monitoring since HD scopes can be a lot pricier than the SD scope). be big drawback to this method is that the refresh rate of the network-scopes is not too good (the one we have sends a “snapshot” screengrab of the scope every 8seconds)…so they really on work if you have an area of concern in your program and you park on that frame and wait for the scope to update.
k
hi david,
conforming is basically the process of bringing the elements for a project together into some type of meaningful order from their raw form to match pre-existing timings of some sort.
in commercials it is usually taking the EDL (edit decision list) for an Avid or other offline edit system and redigitizing the media at higher resolution from DigiBeta, D5 or HDCAM which will be reassembled (usually semi-automatically) to match the editor’s cut.
in film it could mean taking the EDL or cut-list from the edit (again, usually avid) and linking it to film scans which have been saved as image sequences somewhere.
archving is pretty much anything you do to close out a job and mothball the elements in offline storage in some format that can be retrived later if needed. in discreet-speak this is usually done in an automated process to either DigiBeta, DTF or to a file on the network (which is just a big .tar file). these archives contain the clip media, the audio and the metadata for the media including edits and soft-effects.
for non-discreet systems the process is usually just culling out the junk (intermediate renders, duplicates, tests) and committing the remaining files to some type of bulk storage medium (DVD-R, tape drives, etc).
hope this helps
k
to the original poster:
i can see why the idea of using a cracked version of flame is so appealing….$100k just for flame software can seem like a lot. however, the price tag of FFI/FS reflects not only discreet’s (excuse me, AME’s) opinion of it’s worth but also what the post/fx market will bear for very powerful software running on very powerful machines helmed by very talented artists.
if you are a skilled FFI artist known in the ad/fx community and you are setting up shop for yourself then you should have little problem charging competitive rates for your work and making enough to pay your lease on that gear. clients (particularly ad agencies) know the difference between sitting in a suite for 5 days to get their spot done for $50k and waiting for somebody on combustion/AE to do their spot in 3 weeks for $5k. they pay for the interactive nature, the speed, the iterations, the creative process.
i’m not saying that you are going to have an easy time of it at the outset and you are going to be cash-flush…that is up to you and your ability to market yourself and deliver jobs. doing top-dollar work on cracked software is about the most insane thing i can think of. what happens when you do a great job and your client puts out a press release touting their spot to the ad community and they mention they did it with you on FFI? think discreet’s client base is so huge they won’t notice a new name in the mix working on gear they didn’t sell him?
also, keep in mind that if you are setting up shop for yourself you will probably be doing a lot of the IT/tech support stuff solo until you can grow enough to need other suites and can afford a full-time sysadmin. that first year of paid support with your new discreet (AME!) box will become your very best friend. FFI/FS systems are nothing if not flaky (even the new linux systems)….i personally would not want to debug a used Octane2 and its oh-so-reliable VBOB by myself…you’ve got patches on top of patches on top of patches for every subsystem in the box and every dot-version of the software seems to need a very select OS/driver setup to run optimally.
i’m sure a lot of the posters here will agree and can provide helpful advice on fine-tuning your business model. we’re all in this together.
good luck & be smart.
k
have you looked into buying one of the new linux-based systems? the smoke-flint linux combo was recently selling for around USD$110k turnkey from discreet (this includes a pretty fair sized stone array). the only major drawbacks to flint is that it is currently limited to 8bit and does not have the modular keyer. other than that they give stellar bang for your buck (especially for broadcast work). they are very zippy in terms of both interacting with the UI and for rendering.
i think that would be a much safer bet than buying used gear. all the SGI stuff can be pretty difficult to maintain (read: they break a lot) and the upcoming releases of FFI/FS will no longer be able to run on R10k processors (which is most of what you’ll find on the secondary market).
k
do you mean software raid (like you said) or hardware raid?
i’m pretty sure they have hardware raid. the initial config that the sytems shipped with when we got them were JBOD, but i believe we got the chassis for the hardware raid later
k
mark,
i’ve only used the flinux systems sparingly, but i don’t think you can have ANY media on they system higher than 8bit…from what i understand this is a discreet-imposed limitation to differentiate it from flame/inferno
when you initiate a wire transfer from another system or import an image it is forced into 8bit space
somebody please correct me if i’m mistaken about this
k
the new linux systems are very, very zippy. in general the UI and interface responds much quicker on those boxes than the older SGI’s.
i really like the linux systems for how fast they handle timeline soft-effects in Smoke….they beat my Octane2 hands-down.
as for flint on linux vs. flame on SGI….i think the two main deficiencies having to work in 8-bit (not much of a problem if you are working NTSC/PAL) and lack of the modular keyer. other than that they are basically identical (ok…fine…you don’t have Motion or 3d tracking but as soon as discreet rolls out their new optical flow and tracking technology in the new flame?inferno? releases i think you’ll see features like that ripple down to flint pretty soon.
by way of comparison, i took a complex comp i did on flame (30+ layers, particles, motion-blur, high anti-aliasing etc.) and rendered it on the flinux box and Onyx2…the flint box rendered it 3 times faster!
depending on the type of work you’re doing (broadcast vs. film, etc) i think you’ll be very very happy with the new IBM based linux systems.
be careful when you buy, though…because i know discreet has some near-future plans to release some HD functionality on those IBM platforms, and we were told that our existing linux boxes won’t support the newer cards that will be required
k
actually the best grid i’ve found is the Grid spark from Method Studios:
http://www.methodsoftware.com/index.cgi?f=r&f2=s&f3=6&f4=&c=pxUKJOWa6VVRY
and it’s FREE! (you get a free license from them)
it’s pretty simple and painless to use, although it only works desktop, not batch.
good luck
k
yeah, the Oct2 + VBOB solution is horrible. almost everyone has some type of problem with them
we have 2 systems here with that config and it think each as been through at least 2 vbobs
my best suggestion is to stay on hardware support (either thru discreet or a local SGI support reseller like XS Internationl….we have the latter and they have done right by us many times)
sorry to hear about your troubles….just know you are not alone
k
from what i saw of DF5 at NAB it was really tight. the 3D workspace seemed very well laid out and extremely user friendly. i played with it for an hour or so and it didn’t crash or lock up, so i think the stability is there.
the particle system is pretty amazing for a compositing app…i’d put it easily on par with FFI’s particles but nowhere near where you can go with a true 3d package. really neat features include pre-packaged expressions and behaviour controls like turbulence, flocking, etc.
overall i’ve had a better UI experience with DF4/5 than with Nuke, but Nuke in our shop has seen development of some really sick widgets that i don’t think you can do with DF. i doubt very seriously that D2 will ever include particles in their package but they have promised enhance useability and speed with the 3D workspace (which can be kinda kludgy at times)
3kx3k is a pretty big image….if it is not necessary to use that large of a file i’d use something smaller
if you absolutly need that level of detail (say you are zooming into the image and only seeing 15% of the total area) then you might want to try nesting a cropped hi-res snippet of the image inside of a blown-up lower-res version. if you match up the two images and add some soft-cropping to your insert piece they should blend fine and your renders will go a lot faster, even if you are doing hi anti-aliasing
good luck
k
you can try making an expression for each child that is the negative of the parent’s y-axis rotation
the big problem with lookat is that it only works properly with axes that don’t have any parents…it cannot evaulate the entire chain of inherited transformations of one or more parent axes above it.
NAB was pretty cool. not for any specific thing i saw (lustre FAST 3d tracking was cool, but why was it demo’d in lustre and not inferno?…..clip history was demo’d last year…..smart moblur was exciting, etc etc)….
what really got me jazzed was the overall sense that I LOVE WHAT I DO and that i feel truly blessed to be working on a discreet box. FFI/FS artists are a breed apart and when you stood at the confluence of the discreet/apple/avid booths and looked at each respective vendor the discreet (oh, sorry…autodesk) booth just made me S M I L E. something about the toolsets and the integration and throughput simply puts them on top. true, Apple is making very positive and creative inroads but discreet is matching stride with opening up their API (viz: Wiretap) and reaching out to the workgroup market with Toxik. i hope that the autodesk acquistion and rebranding will ultimately mean increased R&D spending like they said at the show. discreet has huge street cred and they have their work cut out for them to stay ahead of the curve.
so…to sum up…thanks to all the people who went above and beyond to present, demonstrate, impress, talk, communicate, listen and respond. john, mike & jeff from fxG for their session, phillipe for his contribution, the autodesk brass for being so accomodating and especially to martin for his charming ways & honest insights into FFI.
it is a pleasure to be part of this community and a joy to be doing this work.
k
ps-> if anyone manages to get a copy of the showreel i’d like to get one too to show everyone at the shop who didn’t get to go to NAB.
i think if you open the archive and save one frame of white (or bars or whatever) to it and close it then it will force a re-write of all the TOC’s
k