Okay guys now that I have you both here... what do you think about a function that let you set a threshold value for the reference brightness and then selects all images above or below that threshold. This would give you the chance to delete images that are too dark/too bright in relation to the reference. Certainly this won't do any good to most of the time lapse sequences - but regarding a construction time lapse it would for example allow you to filter out subsequences where clouds shade the area.
Certainly this would lead to "gaps" in the sequence especially visible in the clouds. But I could imagine that for the construction it wouldn't matter.

I'd love to hear your thoughts!

Okay guys now that I have you both here... what do you think about a function that let you set a threshold value for the reference brightness and then selects all images above or below that threshold. This would give you the chance to delete images that are too dark/too bright in relation to the reference. Certainly this won't do any good to most of the time lapse sequences - but regarding a construction time lapse it would for example allow you to filter out subsequences where clouds shade the area.
Certainly this would lead to "gaps" in the sequence especially visible in the clouds. But I could imagine that for the construction it wouldn't matter.

I'd love to hear your thoughts!

I certainly think it would be neat to play with, but I've been having decent results with it as-is, so absolutely don't go out of your way on my account, especially since I haven't even donated to the project (yet!).

I did notice that when it's mapping out the brightness curve on the graph, now and then a cloud or something might come across the sun, causing a sudden plummet in brightness for a frame or two, which, I suppose, sort of affects the deflicker calculations especially?

I'll be happy to test out anything you're putting together, but just know that I am very much a rookie when it comes to time-lapse shooting and processing, and rather new to LR3/4 in general (started playing with it in January) so take my comments and thoughts with a grain of salt)

Hi Gwegner,

Now that is creative thinking.
I have to say that for the construction timelapses I have made, i dont use LRT, cause functions like deflickr just dont really are up for a job where so much light changes constantly. I used GB Timelapse deflickr in post on some renovation sequences which were indoors and were frequency was less than a minute.

A function to get out the very worst changes in brightness would be appreciated. I have done that normally just on eye-sight in Lightroom. (every 100 or so images shutters dont do what they supposed to, or something that just passes by, makes a very different lightning situation).

So, if it is not to much of a programming job, I would love to try it out for you. I can send you some construction samples as well, like a week with 10 minutes interval. You can play around with that alot

On this thing you can see what clouds do to a 1-minute frequency, annoying, but in my point of view, also rather charming cause that is what a day is.
Problem can be that the camera ofcourse accounts for the difference in cloud lightning, and puts out the same luminance, allthough in a entirely different sort of image?

[Video: http://www.youtube.com/watch?v=3cOwkWL95KM]

Btw, this was on a fixed daylight WB, and I did not alter it later. You see the grass turn somewhat blue in cases of clouds. In the auto WB I used in some post projects that really turns out better.

. sucbscribing post

Very cool video, Bas, I enjoyed that.

This post was last modified: 2012-06-17, 23:44 by matt314159.

Just an update on where I'm going with this time lapse project, the Suction Cup + S90 + CHDK Intervalometer Script + Eye-Fi card transferring to PC is going to get replaced sometime in the next week with a solution that will hopefully become more tenable.

The Suction Cup is getting replaced with a Sharpics clamp mount (http://amzn.com/B002H733OE ) that will (I hope) clamp to the window frame fairly sturdily and lock into place much better than the suction-cup mount.

The S90 + CHDK intervalometer + Eye-Fi is getting replaced with a G10 + GBtimelapse (older version that still supported the powershot series) shooting tethered via USB directly to a PC.

The problem I was running into was mainly the eye-fi card. It is not meant to run for days on end and more often than not, it crashes. Sometimes it will simply stop transferring to the computer but still save the photos as the camera shoots them, while other times, it will simply throw a memory card error on the camera and freeze everything up until power-cycled. This is not going to work at all.

Hopefully the GBtimelapse program will allow me to program times of day and day of week...if not I may have to find another tethering software to allow me to do that.

Sounds like your getting on the right track. The clamp looks a lot better then the suction cup (really, a suction cup?
But, if there is a chance I would just nail it on a wooden plank, and that plank nailed to the windowsides. Tight for a year then.
Clamps are good for a day, but I wouldnt bet my life that after a year they will not have moved, just by loosing grip and, how do you say that, loosing tention on the knobs.

O, dont forget to hang a black carpet or sheet behind it, so no light from the inside can reflect in the window to your camera.

It's a metal window ledge. The window is probably 4' x 4' with four, 2' panes, with a metal 2" ledge that divides everything into four quadrants. I'm shooting out the top right-hand quadrant of that window.

After the S90 I did get a light damping system that seems to work quite well. I fashioned a hood out of the sleeve of a black long-sleeved shirt that I cut the arm off of. The cuff of the arm goes up to the point where it touches the window all the way around the camera lens. The camera lens is as close to the window as possible. The rest of the sleeve covers all the sides of the camera. I cut a slit up the back to give easy access to the camera controls.

Even despite that, the double-paned windows were getting some sort of minor reflections in through the sides, so I took a piece of matte black construction paper and cut about a 2" hole in it, and taped that right to the glass, so that blocks out reflections coming in the side from an angle.

Bas, you said before you've used gbdeflicker, would it be in bad taste for me to ask, on the LRtimelapse forum, what you think of that software? Would it integrate better for the GBtimelapse software to make my workflow a lot easier? I like the results I get form LRtimelapse and Lightroom but it's kind of a cumbersome process to do every week for 60 weeks straight.

Just for kicks, here's a render of the 2nd week:

http://youtu.be/YKA47-TYbZo

The eye-fi card threw a memory error on thursday aroudn 3PM so i missed out the rest of thursday and all of friday, so It's only about 3.5 days of shooting. I notice more noticeable flicker this time around, though I used about the same process as before, not sure why that would be. I did mess with clarity settings a little and that kind of thing in LR so maybe that was part of what did it. But also just dealing with clouds going overhead, etc, is difficult, so I suppose it's not terrible.

.. Hm I thought I posted a long reply, but appearantly not.

My experience is that deflickring in constructionprocesses is really not necessary and possible.Maybe on the 3 minute shutter it will a improve a slight bit, but on the 10 minute count, there are to many variables that already make a flickr in a shot. Clouds completely different, shade on different parts, etc etc. Your camera is the best deflickr that there is. Deflickring in post production, with either GBD or LRT both need a sort of similar before and after frames. If that is not the case (1 seconde, 25 frames = 1 hour) then they will go and adjust the luminance in frames that didn't need that changes because the lightning change was natural. They will go and alter perfectly good shots.

Seeing your results so far, I think you should be fairly happy with the natural flickr that comes with such construction projects. People will accept them, for the leap in time is understandable. Like you said, the effort takes up a lot of time, and recources, and the results are not that great to justify that.

A sound and smooth whitebalance is a better goal for me. That annoyes me much more.

If you leave your emailadres I can send you some tests with WB and deflickr tests from one camera. (not for public eyes in this case)

Regards,

Bas