Part 1 done & long take self harm
Added 2021-02-22 22:54:16 +0000 UTCThe good news is the first part is done, minus a 15 second bit I still have to work out the dynamics of. Otherwise, we are now up to 2min34s of episode, and this thing is getting split because there is no way this is going to fit within file limits if I don't do it in some smaller parts. I know where to break, and I have an idea for how to do it, so it's all good.
On the subject of length, the 2nd part has now been blocked out. Literally. I haven't got completed models yet, so I've spent the evening working out movement and timings for speech using blocks. I might even release it as a preview soon. I got it all together, all finished and just needing the characters added and animated which is 100% not a small job, but all their movement is along paths, so I just have to animate where they are along the path and map where they move in relation to each other. The models then get dropped in and made to follow the path.
Both models should have walkcycles added once I am ready to add them, so the bulk is timing tweaks, stopping foot slippage and the bits in between the walking. Then with the blocks animated, the camera was added and all camera movement, angles and so on have been added. Happy with tonights work, I saved it and then realised that in a monumental act of self-harm, I've made a 1870 frame long single take. For reference, that's just over 62 seconds of solid animation with no cuts. Rendering it is going to be roughly 3 days non stop, once I've finished it.
I genuinely don't know why I do this to myself. Add extra cameras. Cut a few times. Be sensible. No, fuck it, a minute of animation. And not even anything exciting. Just dialogue. Hooray.
Comments
The 8 part would be split into something roughly 4/4. Keeps time reduced for each episode, gives viewers time to handle a couple of concepts before throwing a bunch more stuff at them straight after. At the moment, the script introduces something then almost straight away has something happen involving it, so it might not have the same impact as it would if it was allowed to breathe for a while. The blocks method is pretty much just placeholder cubes with a colour for each character. I've also got text which appears for the speaking parts visible when they talk to get timings. When I import the new rigged characters, they'll just replace the blocks. Normally, I'd know roughly what I want them to do and wing it as it goes along, which means animating pretty much everything, then finding out it doesn't work for camera angles or interactions or something. This way, I can knock it together in a few hours and do all the other stuff like getting the camera paths set, walking speeds, stuff like opening doors and so on. And it means only animating what needs to be animated with less going back to remove parts. Another positive is that if you don't know where the cameras will be, you animate everything, so you can get a shot wherever. If the camera never goes below the waist, there's no need to spend the extra time making sure the feet don't slide during the walkcycles. This pre-planning of movement and shots should hopefully optimise the process and make it so I only have to animate what needs to be animated. Luckily, it won't need to be rendered in one go. Normally what I do is use switches between cameras as parts to render, so any minor changes between one render to the next is less noticeable. This could be tweaks to poses where it looks good from the first camera but slightl weird from the next, so it would be adjusted to the best pose between shots. The main culprit is Efa's breasts. Because they're baked physics, and they sometimes take a while to bake, they get done on a per-shot basis. This means that sometimes they jiggle a little differently between each shot, so the cut hides that. For this longtake, it just means I'll have to bake everything first, then ensure every setting is spot on so it looks the same. It'll be challenging but not impossible!
Methodia Rascal
2021-02-24 06:45:13 +0000 UTCI'm a little confused, are you splitting the first part into two videos (e.g. 1-A/1-B) or just cutting the file in two for rendering purposes? Or possibly expanding the initial 8-part concept into a 9-part? Any way it is, always glad to hear you've made progress. Using blocks sounds a bit difficult, but whatever works! Hopefully the walk cycle works smoothly without too much tweaking, the more automation you can get the better. That's a pretty crazy take indeed! In the context of the larger multi-part episode, I could see it as a relaxing scene between actions as long as the camera isn't static the whole time. Unfortunate that it'd all be rendered in one go, though, lots of room for little problems to sneak in. Could you perhaps copy the existing work and alter things at your pleasure? Not sure on the workflow with these >_< Regardless, thank you for the update!Things don't always work out perfectly, all we can do is try to keep our chin up!
Dargon
2021-02-24 05:37:13 +0000 UTC
