Ramped time lapse

[David Regan]

I'm looking into doing a short time lapse shot later this week on super16 on an SR3. What the director and I really want is for the time lapse to be ramped in speed, i.e. it would start as a moderate time lapse of clouds reflected in a car window. The time lapse should last for about 2-3 seconds, then we would slow the time lapse down, doing shorter and shorter intervals, so finally arrive at full speed, which would be a dolly back from the reflection to see the whole car. Does anyone have any suggestions for doing this well, or any reccommended intervals? Since there is no feature in the SR3 to automatically do it, I think we would just have to press record every minute or so, then start rolling evek half minute, every 15 seconds, every 7 seconds, and finally just roll full speed and do the dolly. Hopefully we'd get in in one take, I don't know if we reset quickly it would be too much of a jump to start it again, we'd probably have to start back at every 15 seconds, and pick it up from there.

 

Anyway, any thoughts or tips on doing this much appreciated.

 

Thanks

[Chris Keth]

Yikes, that's a good one. I don't think you can ramp the SR3 with what you have available there are school.

 

Here's my recommendation: Set the camera up on the dolly and get the move all set to happen. Set everything to first position. Shoot time lapse stuff at 1 fps. That is the lowest running speed the SR3 will do. Do plenty of it, because you will time remap that footage in post to get the really sped up stuff for the beginning of the shot (the really fast time lapse stuff). You'll be speeding it up a lot so it will eat up footage fast. If you want 10 seconds of finished ramped time lapse, shoot a couple minutes of it to be safe. Shooting that whole segment at the highest frame rate of the "ramp" will let you do speed changes in post without calculating any frames and losing image quality. Once you have a good bit of that, quickly switch to 24fps without moving the camera and hold your frame for long enough to get some material to work with and then do the pull out. The 24fps footage at the beginning of that shot will be time remapped in a similar way to the first part. The idea is that you will start the second segment at 1/24th normal speed to match the end of the 1fps shooting and it will blend in and look continuous. It's VERY important that the frame doesn't change when you switch speeds or it will be a huge pain in the ass.

 

If you have any questions, well, you know. :lol:

Edited January 21, 2008 by Chris Keth

[David Regan]

Thanks for the tips, incidentally, its for the pickups for the trailer.

 

So yeah the 1 fps sounds really good, I looked at the references of what sort of time lapse he wants, some clips from rumblefish, and I think the start will be pretty fast, so I'll probably do the 1fps thing for maybe 15-20 minutes or so. Then I'll probably run maybe 15 feet off at 24fps, and then start the move. I think thats a good idea of yours since it should allow us to reset quickly and do the shot again, without having to redo the entire time lapse.

 

As for moving the camera its not a huge deal, since the starting frame is a non-descript section of tinted window, we don't need to see anything outside of it. (Unless Neal wants it of course, and then it will be tricky lol)

 

So yeah, I think it can work. Rehearse like crazy.

 

Thanks for the good tips.

[Paul Bruening]

Dude,

 

I continue to be haunted by Rumblefish. I can't stop it. It just pops into my head when I think of how I want to handle a scene.

 

 

 

Don't box me in.

[Chris Keth]

I would go even longer than 20 minutes if Neal's thinking of stuff that fast. Remember that if you were to go like 1000% for the beginning, that would be eating up 1 minute of footage for ever 6 seconds of finished footage!

 

Since you want more than you'll need, maybe go with like 40 minutes at 1fps. I guess see what kind of finished length he is thinking of than you can do some rough math and figure it out.

[A. Whitehouse]

20 minutes is a lot of footage at 1 fps. How long does the director want the shot to be 3 seconds? A half minute time lapse in the middle of a short film sounds pretty indulgent but speeding up a lot of time lapse material into a quick shot might ens up with so little detail in anything moving that it looks nearly static . You can get external digitachs from a local rental house which at the flick of a switch will return you to sync or even a ramping device for your camera which can be programmed and incorporated into your move. How were you going to handle the exposure change or was it going to be a 1/48 exposure during your initial 1 fps run? Depending on how quick the reflected clouds are moving you might find that 1/48 exposures might be to jittery and require some work in post or even in telecine. Ive done a lot of time lapse on 16mm using digitachs and they're great devices. You could ramp down using different combinations of intervals and numbers of exposures. It could even be incorporated into the move with practice. Don't forget to tape the view finder.

[Chris Keth]

20 minutes is a lot of footage at 1 fps. How long does the director want the shot to be 3 seconds? A half minute time lapse in the middle of a short film sounds pretty indulgent but speeding up a lot of time lapse material into a quick shot might ens up with so little detail in anything moving that it looks nearly static . You can get external digitachs from a local rental house which at the flick of a switch will return you to sync or even a ramping device for your camera which can be programmed and incorporated into your move. How were you going to handle the exposure change or was it going to be a 1/48 exposure during your initial 1 fps run? Depending on how quick the reflected clouds are moving you might find that 1/48 exposures might be to jittery and require some work in post or even in telecine. Ive done a lot of time lapse on 16mm using digitachs and they're great devices. You could ramp down using different combinations of intervals and numbers of exposures. It could even be incorporated into the move with practice. Don't forget to tape the view finder.

 

We're talking shooting time for my recommendation of up to 40 minutes is really only a 1:40 of footage at projection speed.

[David Regan]

I figure we'll be ok on footage. The director said he really only wants about 3 seconds of timelapse so if my math is correct, if we run 30 minutes at 1fps, that will give us 45 used during that time lapse. Then he can compress that 30 minutes into 3 seconds, if he wants something faster, or he can take only a portion of it into 3 seconds for something slower.

 

A. Whitehouse, you make good points, I plan to set exposure based at 1fps so a 1/2 second exposure time. Since in a typical timelapse there are fluctuations in exposure, I will find my initial exposure, and base everything off of that, and just adjust stop when I go back to 24fps or any other speed. I wasn't aware of problems in post but thanks for the heads up. I'd love to do some tests but since this shoot is a bit of a whim, and we shoot in 2 days, there isn't time as I'd like. If you do know of any problems I am sure or likely to run into that would be great if you could warn me about them. Again, I'd love to try with the digitach, but for the above reasons, and budget I don't think rental of one is going to happen.

 

Thanks for the help

[A. Whitehouse]

We're talking shooting time for my recommendation of up to 40 minutes is really only a 1:40 of footage at projection speed.

 

Yeah I get the math but my point is that for a 3 second shot, of say clouds, to be created from 1:40 of footage you're not left with much in terms distinguishable action. From my experience doing this in the past, compression of time and frame blending (which is the sort of process this will go through in order to achieve the shots projected length in post) mean you require some handles to work with but too much and the action just turns into a transparent blur. I mean we're talking about compressing 2400 frames into 72. If they start to go about the process of removing every odd frame or something then the gaps in time become problematic and the frames are too dissimilar causing jitter. Apart from all that if this is low budget then we're talking about burning through 20% of a roll each take not to mention the amount of time out of the

schedule. If the directors certain they want roughly 3-5 seconds of time lapse I would look at shooting for 3 or 4 minutes real time or 200 frames ish. If the clouds are moving so slowly in that space of time that not much changes then do as Chris suggests, it would put the most frames down to work with.

 

I don't think a digitach would be that much to rent and they're very easy to use. Even book rate is probably well within your reach. I think you could use it to get a better result, more time between exposures and experimenting with bursts. Have you thought about changing the shutter angle on your Sr3? Maybe a shorter exposure could help? Good luck, these types of things are really good fun. Maybe instead of changing aperture you could pull out the appropriate amount of NDs before ramping to Sync, that way there'll be less of a shift in DOF and make the two shots more seamless. If you get the opportunity post your results here, Id be interested to see them.

[David Regan]

Yeah, your points are good, and having never done this before, I realize it may not be perfect. Again I'd love to test...but...no time :/ So I think you are saying that is better to have more frames and throw some away, than record only a few? Since we may have scrounged up some more short ends to have, maybe I'll try running for a shorter extended period at like 12 fps.

You say 3-4 minutes real time, or 200 frames ish. So still at 1fps? I'm trying to get my head around my calculations to figure if this will work or not. So lets say I just wanted an ordinary time lapse, without ramping it, what would your suggestion be? My biggest concern is if we will be able to retake this without redoing the entire first time lapse. Or if we just redo the ending part with say 20' of 24fps full speed footage, might we be able to get away with it?

 

Thanks

[A. Whitehouse]

Well at 1 fps thats 60 in a minute. Since the end result needs 72 frames then its about trying to cram as much movement into those frames. Rolling for 4 minutes real time (at 1fps) means 240 frames which I think will give you more than enough to put something together in post but the problem is if the clouds are moving slowly and not much happens in that 4 minutes then it kind of eliminates the time compression from the time lapse because theres no indication of motion or time passing. In that case you will need to extend the period of your time lapse in real time and shoot more frames. Ideally, using something like a digitach, you could space the time out between exposures (like 1 frame every 4 seconds) and compress time even more. It really depends on how much you want the clouds to move through frame and that means you'll have to judge it for yourself on the day. It could well mean that in the end you do have to wait for 40mins so that the clouds have enough time to change and move for your shot.

In terms of footage, I don't think having 2400 frames is going to help you more than say 240 but because the camera can only do 1fps you may have to shoot that much and edit out every tenth frame to make it usable.

I would do the time lapse at the beginning of each take but if you had too you might be able to fudge it in post. Do you know anyone who's handy in After Effects?

I would suggest something similar even you weren't doing the ramp.

[David Regan]

Yeah I think we might have to do what you said, shoot 2400 frames, and just edit out certain ones. Of course if we have a good windy day, and the clouds are really blowing over fast, as you say, I'll just thave to 'judge it on the day.'

 

Thanks for your help really appreciate it. I'll post results if I can when we get the footage back.

[Chris Keth]

Of course if we have a good windy day

 

You mean Rochester has days that aren't windy? :blink: