[Kevin Camp]

i think i agree, you need two rates… my 13 pixel move idea does work to create whole pixel move in the 2k and the dv, however the move is too fast, so it appears jerky, and it still looks awful when you add pulldown.

i think a major issue is the pulldown, they way interlacing has to be generated using even lines from the frame in front and odd lines from the frame behind seems to be a real problem. going 24p for d1 would help… do you need to put the d1 to tape? is it necessary to interlace to bring the piece up to 29.97?

does anyone know how they get movie credits to 29.97 for dvd/tape and avoid this problem? do they create a new rate for the credits?

Kevin Camp
Designer – KCPQ, KMYQ & KRCW

[Kevin Camp]

i wondered how one might go about creating whole pixel movement for a film and also having that movement work for when scaled down to a dvd… and the only thing i could think of was figuring out an increment for the film rez that could be divisible by the scale factor…

1556 / 480 = 3.241667

lets just call it 3.25 (although it wouldn’t quite be accurate and it may cause problems later). you would need a film rez movement of 13 pixels per frame. 13 / 3.25 = 4, so your scaled down version would be 4 pixels per frame.

unfortunately i don’t have ae at home to try this (i’ll test it tomorrow), but i’m not sure how it will work with a pulldown added to the d1 24p.

i think the best solution would be to create one for film and another for dvd.

[Kevin Camp]

what’s more interesting is the macpro seems to get better with windows….

here’s a link someone posted earlier this month:
macvideopro.digitalmedianet.com/articles/viewarticle.jsp?id=167680

[Kevin Camp]

you may just be experiencing ae’s not-as-good-as-it-should-be anti-aliasing. have you tried rendering back to the dvcprhd native 1.33 par and 960×720 frame size?

check that against the original, that will tell you if you are losing quality somewhere in the workflow. if it looks the same, then your workflow is good, and it may be just that ae’s scaling algorithms are not as good as you’d like.

your best option then would be to keep it in the native par and frame size, or try to get a 1280×720 square px export/conversion of the dvcprohd out of your nle or other tool, or from who ever gave you the dvcprohd quicktimes, hoping that another piece of software will do a better job scaling the 1.33 par to square.

[Kevin Camp]

i’m a bit confused by your question….

but, if you have an earlier version of ae and want to upgrade to version 7 (note that the latest version is cs3 also known as ae8), but you need to make sure you are purchasing the correct upgrade. there are different upgrade paths based on the version number (6 or 6.5, etc) and bundle (profesional/production or standard) of ae that you currently have.

you would need to check with adobe or read the packaging for requirements to see if it is the correct upgrade for the version that you have.

if you can upgrade to cs3 (ae8) there is only one version — proffesional. they no longer produce standard, and, they are not charging more to upgrade from a previous version of the standard ae. so upgrading to cs3 may be a better option.

all upgrades are complete versions (it’s not a patch or update), you just need to have a previous version of ae installed, or enter the serial number of the previous version to install the upgrade.

[Kevin Camp]

make sure the quality for you footage layer is set to best (its the check box under the diagonal line in the switches panel in the timeline.

also make sure your not using opengl… i’ve never had good results with opengl scaling to fix pixel aspect ration (despite ae’s pixel aspect ratio correction being off, ae is still scalling the 1.33 pixel to fit your square pixel comp… i’ve always noticed opengl having problems with this). turn off opengl in the preview preference and make sure the render settings are not using opengl (the default is normall set to off, but that can get changed).

if none of that helps, try setting the comp to the dvcprohd native 1.33 pixel and rendering that, then compare it to the original in your nle or quicktime.

[Kevin Camp]

i have a 4-core macpro at work and just got ae8 (cs3) last week. if i have time i’ll try the totalbenchamrk test next week. i’m a little more ram challenged with only 4gb, i’m planning on hounding my boss for more when she gets back from vacation…

purely for the sake of speed, you might try disabling the disk cache.. ram’s much faster than even sata ii drives, so if the render is hitting the ram limit that may slow down multiprocessing. you may also need to decrease the default memory to around 2gb or maybe even 1.5gb to allow all the cores to have enough ram. likewise, i think you would also need to reduce the ram cache to half the memory setting, around 1gb or 750mb.

just a thought, it may not make any difference… i’ll try to do some experimenting next week.

[Kevin Camp]

so…. what you have are two pieces of footage (or images). you want to have one be used as the front of plane that will get broken into diamond shapes. the other footage or image will be used as the back of the plane. each diamond will flip over and show the back side of the plane. correct?

are all diamonds the same size? they’ll need to be to be used in card wipe/dance.

if they are, then you will need to break up your image into equal diamond shapes and (to use card wipe or dance) you will need to make sure the diamond shapes are whole (you can’t have partial ‘cards’). also the diamonds will need to be separated into columns and rows of whole diamonds. there for you will need two ‘fronts’ and two ‘backs’ to make the whole image.

what i would do first is create a diamond shape (solid over transparent), then use motion tile to fill the comp area with the diamond shapes. make adjusments to the tile size to get full tiles within the comp edges (they must be full diamonds to the edges of the comp).

now you will need to figure out the size of the diamonds (x and y) and enlarge the comp by that size. so if your diamond was 72×96 px (after adjusting the size of the tile) and your comp was 720×480, then change the comp to 892×576). you need to do this to get full ‘negative’ diamond shapes to fill the holes.

this diamond shape comp will be used as a track matte for your images. make your footage or image into a new comp and drag the diamond comp into it. set the image layer to use the diamond layer as a alpha track matte.

rename the comp as front 1 and duplicate it. then open the duplicate comp and change the comp size to the size of the diamond layer (892×576, by my example). don’t scale or change the footage layer, it will need to line up with the other comp. set the footage layer to use the alpha inverted of the diamond layer as a track matte.

duplicate those two comps, and rename them back 1 and back 2, respectively. open those and replace the front footage with the back footage.

create a new comp that will be used to assemble the diamond cards (the size should be the size of your final destination, i’ll say 720×480 for my example. drag the fron and back comps into this comp. hide back 1 and 2, and at this point it might be easiest to hide front 2 also. apply card wipe/dance to front 1 and set the number of cards for rows and columns to the number of diamonds, and the back layer to the back 1 comp. do the same for front 2, note that this is the larger comp and has 1 more diamond per row and column. set back 2 as the back layer.

whether you use card wipe or card dance to flip the cards, they will flip as diamond tiles and reveal the back image as the turn. you may want to link some of the card wipe/dance attributes together via expression if it makes it easier to control the two effects as one. if you have slight gaps between the diamonds, create an adjustment layer and add simple choker with a slight negative choke.

Kevin Camp
Designer – KCPQ, KMYQ & KRCW

[Kevin Camp]

i’ve used both sorenson squeeze and apple’s compressor, both on mac. i like both of them, they are both easy to use and you get good quality for file size…. however, i was not encoding to wmv.

i know there are several choices for wmv encoding software for pc. but sorry to say i can’t really make any recommednations. but finding software that will make two passes, one to analyze and the other to compress, will help create good quality, low file size media.

Kevin Camp
Designer – KCPQ, KMYQ & KRCW

[Kevin Camp]

if you want to test your machines against the ae benchmark used by barefeats.com, you can download the total benchmark project files that brian maffitt developed a few years ago.

but like you had mentioned, there may be an effect that you’re using in you project that is not multiprocessor friendly.

Kevin Camp
Designer – KCPQ, KMYQ & KRCW