Dedicated Tmelapse application

[fma38]

Dedicated Tmelapse application

by fma38 » Sat Mar 06, 2010 10:04 am

I did a test using 0.1 Yaw and 0.1 Pitch for 1440 shots, its amazingly smooth. I am just wondering if smaller increments would be possible. As it at least not visible in Papywizard window.

As the Merlin accuracy is about 0.3-0.5° (because of backlash), I limited the moves to 0.1° in Papywizard (smaller increments won't move the head at all).

In fact, it is possible to be more accurate, as long as you don't change direction. For timelapse application, it could be interesting. But for pano photo, you often change the direction, and it is too hard to manage the backlash (depending how the camera is mounted, if there is some wind, or so).

That's why timelapse must be a dedicated application: needs are very different.

In the latest Papywizard releases, I moved all hardware low-level routines in separate modules, so it is possible to use them in your own program. If someone wants to start a timelapse application, he can make something without GUI at first, to see what can be done. I will help as much as I can.

[claudevh]

by claudevh » Sat Mar 06, 2010 12:04 pm

Hello Fréderic,

Could-you identify those "low-level routines" modules and how we can use them (from terminal by exemple) for testing purposes ?

Thanks

[fma38]

by fma38 » Sat Mar 06, 2010 12:45 pm

They are in this dir:

http://www.papywizard.org/browser/trunk/papywizard/hardware

Examples how to use them are in the related plugins:

http://www.papywizard.org/browser/trunk/papywizard/plugins

There are maybe remaining annoying dependencies (we should avoid the need to use PyQt for that part), but I will fix this if people start to use it.

[olihar]

by olihar » Thu Mar 11, 2010 3:53 am

Here is the test...

I don't know how smooth the compression is at Vimeo.

http://www.vimeo.com/10077334

[Photosbykev]

by Photosbykev » Thu Mar 11, 2010 5:06 am

Here is the test...

I don't know how smooth the compression is at Vimeo.

http://www.vimeo.com/10077334

good start it does looks like there maybe some lost motion/backlash in the drivehead though. There is an occasional pause in the motion which I don't think is down to the streaming?

[olihar]

by olihar » Thu Mar 11, 2010 7:07 am

I tried using a RAM player to look at the shots and it is super smooth.

I will try and export motion curve from it.

Last edited by olihar on Thu Mar 11, 2010 7:16 am, edited 1 time in total.

[olihar]

by olihar » Thu Mar 11, 2010 8:15 am

Here is the motion curve from it made in Nuke.



Did the straight green line in PS to compare it to..

This seems to me to be pretty consistent movement. I guess the little curve in the line would be from 0.1 movement and a little backlash in the Merlin.

[Photosbykev]

by Photosbykev » Thu Mar 11, 2010 8:52 am

I tried using a RAM player to look at the shots and it is super smooth.

I will try and export motion curve from it.

It must be at my end lol I'll be running a timelapse test this morning with just a pan motion on it, I just need to find my 10 stop ND fliters to get the shutter speed down to a reasonable level.

[claudevh]

by claudevh » Thu Mar 11, 2010 11:03 am

Hello Olihar and all others ...

Your test demonstrate one time more that the Merlin head is "capable" for timelapse ...

I have made numerous trials with Merlin and now with the Milapse ... for comparison purpose.
I have learn that the " Shoot-Move-Shoot" method (like the TL preset generator) isn't the best ... it generate sometime quite big differences between the differents frames (due to the fact that the frames are too sharp !)and will oblige to "deflicker" the timelapse during the confection ...

With the Milapse, we don't use the " Shoot-Move-Shoot" method but the "Continuous Motion" method ... (the camera is trigered by an external device during the "continuous move of the head) this generate frames with "blur"!
The assembly of those "blured" frames give the impression of "more smoth video" for our human eyes !!!
The search for "perfection" isn't allways the best way ...

If you have a external device for trigering the camera you could try yourselve.

For all of you, if you are not "very familair" with TL technics, please read "Motion Techniques and Terminology" herafter :
http://openmoco.org/node/93

This is of big interest and will take only a few minutes to read and understand ....

Last edited by claudevh on Thu Mar 11, 2010 11:07 am, edited 1 time in total.

[fma38]

by fma38 » Thu Mar 11, 2010 11:34 am

"Shoot-Move-Shoot" vs "continuous move": Papywizard is not designed at all for that way of shoting; that's why a different application is better.

The other problem is with the Merlin: it is not possible to trigger the camera while moving...

[claudevh]

by claudevh » Thu Mar 11, 2010 12:16 pm

Hello Fréderic,

that's why a different application is better.

I fully agree with you !!!

The other problem is with the Merlin: it is not possible to trigger the camera while moving

I just try to say that "probably" the best method is the "Continuous Motion" method and this mean an "independant" triger system. I mean independant from the Merlin "inbound" controller!

A "PapyLapse" application could, in a first approach, just control "different" continuous speeds with "different" curves profiles ...
(The camera being triggered by a simple external "intervallometer")

I think that we could be happy with that at the beginning....

and later with the development of the Arduino for replacing the "motor board" of the Merlin, this can be extended with "shutter release"

[Photosbykev]

by Photosbykev » Thu Mar 11, 2010 2:34 pm

I'd be more than happy with a speed controller inside Papywizard and use an intervalometer plugged into the camera

For a test this morning I shot a 480 exposure sequence at 0.1 degree intervals using the timelapse generator posted on the forum. I wanted a delay between shots so I upped the stabilise delay upto 2 seconds. When the sequence started I was getting a shot every 12 seconds or so which is too long for most uses. By zerong the stabilise delay the 'frame rate' was 6 seconds/ frame, still pretty long.

With a continuous adjustable rate pan or tilt drive the intervalometer can be programmed to shoot whatever frame rate you like within reason. e.g on the 7D it would be possible, but not that useful, to shoot jpgs continuously at 8 fps until the card was full. It might be that the hand controller would be a better option for single axis movement?

The 480 files are currently being processed, if the sequence is any good I'll post it up somewhere

Last edited by Photosbykev on Thu Mar 11, 2010 2:35 pm, edited 1 time in total.

[fma38]

by fma38 » Thu Mar 11, 2010 3:00 pm

A quick-and-dirty solution could be to use a very simple controller, which only drives the motors, without any position/speed feedback. This one could be nice:

http://www.pololu.com/catalog/product/767

it even allow you to add a RC receiver to drive the head (Pololu has a multiplexer, allowing you to switch between the controller or the RC receiver)

There are other nice controllers, like this one:

http://www.pololu.com/catalog/product/1392

with position/speed feedback, but you need 2 of them (they can be chained).

PS: Pololu has amazing devices!

[Photosbykev]

by Photosbykev » Thu Mar 11, 2010 6:12 pm

a link to the test this morning using the Merlin head/N800 amd Papywizard with one of the Timelapse presets giving me 480 moves @ 0.1 degrees. http://www.youtube.com/watch?v=MEIFaG9fKu8 . At full size the move - stop - shoot - move is visible.

1 hour 40 minutes compressed to 16 seconds. 480 separate images taken on a Canon 7D, 0.5 second@f22 iso 100. The camera was mounted on a programmable motorised pan/tilt head. A 3 stop ND filter was used to get the longer shutter speed and a 2 stop graduated ND filter was used to balance the sky and foreground. Converted from small RAW files to jpg, cropped in PS to 16:9 aspect ratio. Video processed in Adobe AE and Premiere Pro

[claudevh]

by claudevh » Thu Mar 11, 2010 6:51 pm

Not so bad Kevin;)
In you want to smooth more, you can "deflicker" !

[Photosbykev]

by Photosbykev » Thu Mar 11, 2010 7:03 pm

Not so bad Kevin;)
In you want to smooth more, you can "deflicker" !

I applied a directional blur in AE to smooth the image up and deflickered it in Premiere pro. Without losing more significant detail I wouldn't want to do much more to it. Bearing in mind this was shot with a Canon 10-22 lens @ 18mm so it is a wide angle low magnification sequence. If I wanted to increase the focal length then the steps would become very obvious. Realistically you need continuous tracking to get a really smooth panning action.

I'll try and get out in the morning and shoot something similar to this video using the x4 sideral rate (60 degrees/hour) on the hand controller and an intervalometer on the camera.

[claudevh]

by claudevh » Thu Mar 11, 2010 7:18 pm

I'll try and get out in the morning and shoot something similar to this video using the x4 sideral rate (60 degrees/hour) on the hand controller and an intervalometer on the camera.

Ok Kevin, I thing that this will be the right way !

[Photosbykev]

by Photosbykev » Fri Mar 12, 2010 5:40 pm

Using the same post processing as the previous test but no directional blur effect was required I shot another test this morning. I've put both tests on the same page here

http://www.photosbykev.com/wordpress/2010/03/11/cemaes-bay-harbour-timelapse/

This test uses the hand controller of the Merlin so the pan motor is running at x8 i.e 120 degrees/hour, it was meant to be x4 but I just couldn't see the handset lights so it was shot in fast rather than slow mode. The camera was controlled with an intervalometer shooting 680 exposures in about an hour (5 second intervals). 0.5 seconds @ f22 iso 100.

Last edited by Photosbykev on Fri Mar 12, 2010 6:45 pm, edited 1 time in total.

[mikehellers]

by mikehellers » Sat Mar 13, 2010 1:51 am

Having a simple application which allows us to simply create a path to move the Merlin between two defined points as a continuous move, if possible allowing you to define the curve used for the movement, and using an intervalometer to trigger the camera seems like something that would be an excellent solution to start with an provide the flexibility which is needed but which you can't have using the actual external merlin manual controller.
I guess creating a basic application to do this wouldn't be too complicated, or?

..mike

[Photosbykev]

by Photosbykev » Mon Mar 15, 2010 12:42 pm

Added for completeness. This is the final timelapse video I shot using the Merlin head.

http://vimeo.com/10171155

One thousand nine hundred and fifty 0.3 second exposures taken every 4 seconds for over 2 hours. The camera is mounted on a motorised pan/tilt head running at x1 sideral rate (15 degrees/hour).

The camera was connected to a laptop running DSLR Remote Pro which controlled the camera and downloaded the photographs to the hard drive. After 1,950 exposures the camera battery was only down to 40% capacity.

For the final sequence a 2 stop graduated Neutral Density filter and polarising filter was used.

[fma38]

by fma38 » Mon Mar 15, 2010 2:51 pm

The problem with the Merlin is you can't change the speed while moving: you have to stop it first, and resend a command with a new speed. If the com is fast enough, the head won't have time to decelerate a lot, but it will be difficult to follow an accurrate curve, as you don't know about inertia and so (it will depend of the load)...

[claudevh]

by claudevh » Mon Mar 15, 2010 4:59 pm

Nice results Kevin for your firts trials! Really promising !!!
And you chose a "difficult subject" who is quite "non-statics" !
This is more easy with landscapes, stars and in general with "distant" subjects ...

[claudevh]

by claudevh » Mon Mar 15, 2010 5:13 pm

Fréderic, I believe that, if the Merlin should do a "stop and go" in order to change the speed, this will not really be a problem as the pictures are in general only taken every 3 to 5 seconds or longer !

For the "Milapse head" this is exactly the same and this is not disturbing ...

Having a simple application which allows us to simply create a path to move the Merlin between two defined points as a continuous move, if possible allowing you to define the curve used for the movement

Really this will be the "Nirvana" ! Today, we are obliged to use the buttons of the hand-pad ...
We will be happy if we can get only the first part

Having a simple application which allows us to simply create a path to move the Merlin between two defined points as a continuous move

:(:(I am really disapointed that I am "so stupid" that I can't write this kind of small program !!!:(:(:(

Last edited by claudevh on Mon Mar 15, 2010 5:14 pm, edited 1 time in total.

[Photosbykev]

by Photosbykev » Mon Mar 15, 2010 5:33 pm

Nice results Kevin for your firts trials! Really promising !!!
And you chose a "difficult subject" who is quite "non-statics" !
This is more easy with landscapes, stars and in general with "distant" subjects ...

The lighting was bouncing around a lot as there were clouds behind me covering the sun and the sun was still coming up it was only 8am when I started. Because of that I had to 'guess' what the best average expsoure would be for the whole sequence, I wasn't too far out

Converting the 1,950 files from sRAW to Jpg then cropping in PS to 16:9 and then processing in After Effects took over 8 hours, most of that was waiting time It confirms I need to upgrade my computer lol

Last edited by Photosbykev on Mon Mar 15, 2010 5:38 pm, edited 1 time in total.

[fma38]

by fma38 » Mon Mar 15, 2010 5:40 pm

Claude, the shutter needs to be triggered out of these stop/go phase. So, it can't be an external timer, or you may see some strange moves...

If someone wants to start a simple application, this is really easy with Python. You already have all the low-level parts (just use the code from Papywizard; I can provide a version without any dependencies to Papywizard). As a first step, just write a script; you don't need a GUI. There are plenty of good tutorials arround, in all languages GURL started from zero, and wrote a nice generator...