Matt Silverman
Forum Replies Created
-
If you want it interlaced you need to render it interlaced as you have been doing. Even though SD video is an interlaced format, you do not need to render interlaced. We render almost everything progressive with motion blur… this produces more of a “film-like” feel to match film sources for spots. It’s really an aesthetic choice for TV… Corporate jobs usually play back from projectors or Plasma/LCD, so progressive is the only option. The only time we field render is if we something is not playing smoothly and field rendering is the only choice to fix the problem.
-
Matt Silverman
December 12, 2006 at 9:54 pm in reply to: AE render to rgb+alpha with BlackMagic codec OSXFrom my understanding YUV codecs can not support alpha channels.
-
Matt Silverman
October 2, 2006 at 10:53 pm in reply to: How did they do the new ipod spots with those glowing trails?Best guess is Maya Paint Effects.
-
Matt Silverman
October 2, 2006 at 10:51 pm in reply to: Has ANYONE had succes with After Effects 5.5 & DeckLink?AE 5.5 was long before Decklink supported Windows, so I would guess that it is not compatible.
-
When you name the file, just replace the underscore with a period. If you want to permanently change this, I think that there is a way in the text prefs, but not positive…
-
I always try to get dark green tape. It should be dark enough to have a visible luminance difference between the greenscreen, yet still be keyable (or taken out with something like Key Correct’s Alpha Cleaner). The green tape really helps when there is motion blur… If you have white tape you should avoid intersections with the actors, since white tape will have to be painted from the edges of the foreground motion blur. After getting burned a few times, I now bring my own tape. Usually the production companies do not know what to do… I have requested dark green and got fllourescent green which looked identical to the greenscreen. I made them re-tape the whole thing (60’x60′ 2 screens took all night). Recently the grips only had “electrical” tape which was highly reflective and turned white when lights hit them, so do not use reflective tape.
If the shot is pretty planer, then you should use crosses for tracking marks. If you have rotation or severe craning, then triangle markers might work better.
Make sure that you have enough trackers in frame at all times. This will depend on what you are tracking. If you only are tracking position, then you only need one marker in frame. If you need to track rotation or scale you need two, and a corner-pin will need four. If one goes out of frame, make sure there is one to continue the offscreen tracking. In reality, the more the merrier if you have good green tape. I sometimes find myself rushed to get set up, and will get trackers in the specific area for the first shot, then continuously add trackers throughout the day to the rest of the screen… it doesn;t need to be pretty… sometimes you just have to rip some tape and slap it up there.
If you are planning on doing a 2D track, then you should look at the video assist and make the trackers the appropriate size depending on how far they are from the camera. I suggest something like 50px x 50px lines 5px thick on a D1 frame, 100×100 on HD. This should be ideal for a 2D tracker. I like 1/4 inch tape for most shots… usually the gaffer’s tape is 1/2 or 1″ which is too wide.
3D trackers like a lot of points to grab onto. In general I like smaller tracking marks than 2D tracks, and a lot more of them. I have had large tracking marks and found that the Boujou’s trackers end up clinging to the corners of the marks (ie. 50 trackers stuck to one cross on the edges). 3D trackers do a statistical analysis of a ton of points, then throw away bad tracks automatically. This makes them a lot more lenient than 2D tracks which can suffer from jitter. Therefore, I have found that shifting channels and crushing the image prior to 3D tracking will help sometimes even though it is deteriorating the image (ie. do not do this with 2D tracks).
3D trackers need to be placed on planer floors and walls. Do not put any markers on the transition between the floor and the wall.
Try not to do zooms if possible.
If you are shooting on film, get a simultaneous flat pass out of the telecine. This should be a “raw” pass that the telecine’s color corrector never touches. This should be the shot that you track, then apply the data to the beauty pass. Any processing done by the telecine will deteriorate your image quality… even though the raw pass will look low contrast it still contains more information. We have also had success doing red passes in telecine to pull out hard to see bright green trackers that would have been impossible to track without crushing it. It’s much better to do this in telecine than in AE…
Anyway, those are some tips that I learned the hard way.
-Matt Silverman
http://www.phoenixedit.com -
BTW, Panasonic has an FCP Frame Rate convertor plug-in which allows you to manually set the flagged frames (ie. capture 59.94, then later tell it “you are 23.98).
-
We use After Effects for this. Export the QT ref out of FCP and import into AE. Go to interpret footage and change the default pixel aspect ratio from the DVCProHD setting to square pixels. Drop this into a new comp 23.976fps D1 720×486. Hit Command/Option/F to fit this to the comp for anamorphic output, or hit command/option/shift/H to letterbox. Then render to 10bit QT with the render setting set to lower field first and choose any of the pulldown orders. You will then end up with a D1 QT running 29.97 with pulldown.
-Matt Silverman
-
To clarify, there are two frame rates for 720p… 60 and 59.94 progressive. There is no interlacing. Our tri-level sync generator only had these two options which is how we first discovered the nightmare called 24p 720p which as mike pointed out does not exist. Everything is being caaptured either 60 or 59.94 on tape, then when captured the metadata is read by FCP and the extra frames are ignored. When played back these frames are added.
Take a 23.98 DVCProHD Sequence and try to feed it into another FCP machine with a 23.98 capture template and you will end up with your picture playing back twice as slow, since the playback is really 59.94.
720p has been the bane of our existence. 1080 is much closer to what we are used to working in NTSC (once you understand pSf is not fps ;-)), and the first format we started with when we did our HD upgrade. We thought we knew what we were doing until along came 24fps 720p
-Matt