i’ve been asked to put together a short (20 - 30 sec) walkthrough animation at broadcast quality.
i’ve done walkthroughs before so that’s not a problem, and i’ve been following the recent thread about giving architectural animations a more filmic quality which has some really good ideas that i plan to implement.
where i’m not so hot is the output required so that the animation will look good on tv.
output will be PAL at standard 4:3 ratio (probably…) so 720 x 576 resolution @ 25fps - which i’m pretty sure is correct!
where i’m a bit stuck is how to output (compresion, codecs, stills…), if i need motion blur, interlacing etc. etc.
there will be no moving elements, just pans/zooms…
i know a few of you do animations for tv ads etc so i was just wondering if there are any tips you’d be willing to share before i get started so i don’t end up redoing a load of work when it all goes wrong!
oh and btw, i don’t have access to after effects or combustion etc. hopefully that won;t present too uch of a problem! :?
i’ve read that using motion blur can reduce the need to interlace - is that not true?
also if i do interlace, i have to render out twice the fps? correct?
if so, that would be a bit of a pain as i have to render on a single machine…so anything to reduce the amount of rendering would be a bonus…
also i don’t think that quicktime has an option to output interlaced files…
Most of the work i’ve done for TV passed through the Matrox MJpeg Avi codec.
It can be Lossless (in which case is only compaction taking place) or Lossy (MJpeg) with extreme amounts of quality (over 16MB/sec data rate for PAL: nothing but white noise should exceed the 10 Mb/sec with this method).
It can be of help also if you find the need to edit things later (maybe on an RT card in a facility), or better, to transfer to (digi)beta, as it passes all current EBU specs for D1 signals with flying colours.
In my experience QT has always only been used for quick repacks to be sent to clients before meetings, never as a uncompressed storage system, but that’s prolly only my laziness.
As for Moblur/Fielding, it depends entirely on the kind of target you want to reach.
I don’t remember rendering one shot with interlacing, but couldn’t swear as to what got done in post to legalise the footage (probably nothing, though: recorded as full motion on (digi)beta, sent to client for broadcast adaptation…).
You might wish to check out http://www.ebu.ch for everything concerning the Pal format rules, and for sure ask your client what he plans to do with it.
what i’m hoping to do is render out to a sequence of tiffs, edit + compile using quicktime, then burn the whole lot onto a CD or DVD and pass it over to a video production company for conversion to digibeta and DVD.
i don’t think i’ll be able to interlace with the software i’ve got so i’ll probably be adding some motion blur in the render to make up for this.
need to do a few tests to see what sort of blur parameters work best…
as i’ll be working from static camera positions i was thinking i might be able to render out a single frame 360 degree spherical camera from each position, then render the motion blur frames without the need to render the entire scene each time…
We have sometimes delivered animation sequences in the BlackMagic codec (for Quicktime) - this is Lossless and should work with most editing suites. Plugin to be found here:
i’d recommend to stay away from compression if possible
especially 3rd party codecs will cause problems if the production company doesnt have them installed, too
if you have the space (25mb/sec) go for uncompressed
a 30sec spot wil fit onto a dvd without problems (even as a tif sequence, which most editing systems can read)
ps: PAL is 768x576 with 1:1 Pixel aspect ratio - much easier to work with than non-square pixels
all editing systems i know support this
PAL DV is 720x576
what to use depends on your production company
especially on their editing system (Avid, Avid xpress, FCP, Premiere, whatever)
the field order also depends on the editing system (Avid used to be lower first when i was working on it the last time)
i still dont get it why you cant use fields?
quicktime can handle them for sure
(ps: the adobe premiere trial version is fully functional and its perfectly legal to use it for production work
as the animation’s going to be pretty slow panning / zooming of a static scene i think i can get away without rendering to field, but perhaps i should give it a try as it’s something i’ve not done before and might be handy to know in the future.
is there anythin i need to do to optimise rendering to fields or is it just a case of ticking the box?
and you’re correct - quicktime can interlace - my mistake
if you are only doing horizontal pans fields really aren’t that necessary
zooms and vertical movements are different though
try to avoid horizontal details (both geometry and textures) smaller than 2 pixels in hight (2 pixels in the final image that is)
these will show up in only one of the fields, causing flickering
also, the AA filter should be a little less sharp than for progressive scan
and make sure to ask the production people for the field order they need (we used to render a bouncing ball in both orders just to make sure what works)
given enough time you could even render tests (with and without fields) of the portions that involve most vertical movement and test those on tv
opps yeah. hehe. well i plan from here on to do everything i do at film speed since im on NTSC and there is something to be said for chopping 6 frames a second off my renders especially if it takes a long time to render those frames
you dont really need more comments.. but I’ve output to pal dv digibeta a bit
720x576 is less rendering than 768…
any compositing package will take into account the pixel proportion
the post house will probably prefer targas - you should be able to export a still tga sequence out of QT pro uncompressed, maybe not
there ain’t no substitute for a compositing package - you can check how the video legal requirements may alter the footage for one
dont bother with fields unless its really really fast whiplash motion
yeah, you can export a still tga sequence out of QT pro.
obviously i’d prefer some proper compositing software, but i doubt i’ll be able to convince the powers that be as i don’t do much in the way of animations.
would you recomend using blur as a substitute for fields, or just using a relatively soft AA filter (i’m guessing the video filter as it’s designed for video…)
motion blur instead of fields do you mean?
moblur is nice if you have time and if you need it.. but depends if your client wants smooth cinematic or super crisp architectural
havent used the video filter itself - I just use Mitch Net or Cook for a little softening - there’s plenty of other posts here re. which aa filter
I think adobe do a 1 month demo of some of their products - go and try after effects, you won’t regret it
