Welcome to TVPaint Community forums !

On vimeo: https://vimeo.com/77061828

This was done from beginning of September until now. About 2000 images prepared in Photoshop and TVPaint, assembled in AnimeStudio. The cylindrical text overlay was animated in PS CS5 - this was a nightmare because it took ages to render and wasn't even in good quality. And then I couldn't get PS to render it! No way. Finally I remembered that TVP can open PS files - wrote about this already.

This production was cursed - I had more than a dozen complete crashes (hardware reasons), and even data loss. And I'm not so happy with AS' performance this time: numerous times it would crash while rendering, preferrably after 90% done, and then the resulting QT file wouldn't open.

TVP to the rescue once again: I had to render PNG sequences and put them into TVP anyway because I had to correct some glitches with the 3D: the walls in the background kept blinking, undecisive about which one was in front. Nice thing: TVP didn''t crash even once, although I had file sizes up to 3,5 GB.

Each of the 18 "characters" is composed of at least 2 QT movies, one for the head, one for the body, each between 1000*1000 and 3000*3000 px large. The whole concept of putting 2D elements into a 3D environment worked out nicely.

TVPs new ability of dragging frames around in the timeline was a big help. My workflow was:
- construct master file in AS with empty group layers for the 18 characters, with 3D walls, camera movement and animation of characters so they always face camera
- import a bunch of images from a folder into PS via script, regardless of name or size
- re-size roughly, trim
- open PSD in TVP, cut out elements with modified airbrush, or lasso, or magic wand tool
- resize and position exactly to template
- arrange frames
- render QT video
- import video to AS master project file into pre-constructed group layers
- repeat 18x
- render.

Not mentioned here: repeated work due to data loss (PSD files only) or errors, and repeated renders because of multiple program crashes. I estimate it took 1 month to produce 50 sec, and 2 weeks to correct the remaining 8 sec.

I had much fun with the music, again it was a little bit of melody lying around useless for a decade, now spiced up a bit (the style is called "cumbia" and I only learned about it two months ago). Best part always is client feedback, yesterday they mailed me "this is the first trailer which we can dance to - and we do right now!"

The trailer will be shown at dozens of cinemas in Berlin, in the Berlin subway TV, on local TV, online, and during the festival at the beginning of each program.

Nice work and the tedious effort you describe comes through and is easy to imagine and appreciate. It's one of those things that one can look at over and over -- for instance I tried to remember the names of some of the old film cameras I think I recognized -- difficult. Is there an old 16mm Arriflex among them (I forgot the East German model name). Is there anything significant about the faces; people who would be recognized by the audiences? I think this is an example of proper use of the cutout collage technique because it isn't used as a workaround replacement of proper animation. Same goes for the special effects tricks; nothing cheesy there.

Thx Paul. Best part of this is that I conceived the whole trailer concept in a dream right be fore the initial meeting with the client ... I dreamt that camera move in the cinema as well as the picture replacement animation, and I'm quiite satisfied it turned out true to my vision.

As for the faces: I tried to not get any celebrity in there (although some might have creeped in due to me not knowing them) and definitely no stock photo - most faces are from advertising, beauty surgeons, make-up blogs, and porn pictures. Yes, porn. It would've been much easier to do a trailer for a porn film festival with this approach! When I searched for "from behind", guess what popped up? It took more time to find all images than to proepare them. (I did the logo and characters for the Berlin porn film festival indeed, 8 years ago, and they still use them: http://www.pornfilmfestivalberlin.de/)

The camera is most likely a Bolex - I still owe one myself.

The hardest part was to get enuogh pictures of heads in different stages of turing around. I cheated a lot there. One day I was biking around when I spotted a mannequin's head in a shop window. On the way back I stopped there and asked the clerk if I could borrow that head for some minutes and take pictures of it for my animation, since I had my camera with me. She allowed and gave it to me, I put it on my bike's saddle and took a dozen pics with different angles, which are spread among the other pics now.

I spotted the Bolex instantly; went through several of those myself, and I spotted the Bell & Howell with the three lenses. I think what I saw was the real Arriflex, not the East German ripoff.

Thanks for sharing your workflow !

It's always quite complicated when one has to switch between several software ...