Process movie File with Paint Action Sequence

[Frank N]

Hi John and everybody else. I want to process a movie file with a PASeq meanswhile at the same time I like to upconvert from SD to HD or any other higher resolution. Should I just drop the movie file into SA and then change and adjust the canvas into the desired size and set up the right compression settings. Or is it better to change the movie file already when I export the movie from my video editor and then just let it run through SA one to one? What would you recommend? Is there a difference anyway? Thanks in advance.

[David Kaplan]

there are many ways to accomplish this.

what i did most of the time was to build a 'supersizer' step into the PASeq, upping the SD frame to HD size, followed by a warping effect to adjust to the new aspect ratio of 16:9 (if needs be), and then adding a 'canvas to source image' step as well. so then anything that comes after those two steps will render out in HD. however, you'll find that many PASeq steps look and behave very differently at a different frame size, so you'll have to adjust accordingly. by the way, i found that an uncompressed frame size of 720p (1280x720) to be more than adequate for my purposes of projection in a theatrical setting. if you go all the way to 1080p (1920x1080), the frame size becomes too unwieldy to experiment with -- and experimentation is what SA is all about. if you need to output to 1080i or p, you can cross-convert from the 720p very nicely via a kona 3 card or a teranex system or the like. that's what i did for our projection at the sundance film festival and it looked great.

the other way to do it in SA is like you said; simply define the canvas size to HD dimensions (and perhaps add a warp step for the aspect ratio shift) and just play your PASeq and let SA do its thing.

it really depends on the look you're going for and how much detail and sharpness you want to retain from the original source video. try both. also, de-interlacing is very important if you're working with interlaced source footage and there are many different ways to go about that too.

hope this helps.

david

[David Kaplan]

one more way would be to process the movie with your PASeq at SD and then use that processed video as your source video and 'supersize' it up to HD. i did this too and sometimes it's the best way. you'll need to add the warp step again if you're changing aspect ratios from 4:3 to 16:9, and you can also add a few more PASeq steps at the higher definition to give it a little more detail and texture.

[Synthetik]

I agree that it depends on what you are doing. If you are primarily doing paint effects using the paint synthesizer then you can probably just use your original footage and setup the canvas appropriately. If you are doing a lot of Image Operations then you might want to Supersize your footage first and then use that as the source for further processing. As David said, you can integrate the Supersize step into a single PAseq if you wish. If you are going to be doing a lot of experimentation, it might be better to do a separate pass to Supersize and then you can use that footage multiple times with out having to repeat that step. If the canvas size is larger than the source, when you run something like an ip op, the source image is interpolated with bi-cubic interpolation internally when processing the source with the ip op. Sometimes that is fine, but if you are enlarging quite a bit then using supersizer first will give you improved quality. For the paint synthesizer it's less of an issue, since the quality of the paint strokes themselves is what is generating the higher resolution detail.

[Frank N]

Thanks again!