This post was last modified: 2014-01-15, 03:19 by dreamingof8a.

Hi,

I recently tried a Holy Grail timelapse (day to night) with LRT (great program!) and ran into a strange issue after following the workflow. Here is what I did in detail:

- import and initialize the sequence
- run keyframe and HG wizards (here I adjusted the yellow line using the sliders so that the very first and the very last frame ended up exposed properly, that's how you're supposed to do it, right?), save XMP
- inspect results in LR

In LR I noticed that instead of getting darker, the frames were actually getting brigther in between HG-keyframes.
So a *** Keyframe would actually be darker then the next ** Keyframe a few shots down the line.
In the screenshot below you can see that by looking at the exposure compensation from the HG wizard across those short sequences between keyframes. The get smaller very quickly, eventually leading to an increase in total brightness rather than decrease.

[Image: http://image-upload.de/image/VpHWxN/0144f51cf7.png]


When I had a closer look at the keyframes I noticed another very strange thing:

As you can see in the next screenshot, the two keyframes were one EV apart (1.6 and 3.2 sec), however when I underexposed the *** keyframe the picture gets noticably darker although it should pretty much match the ** keyframe.

[Image: http://image-upload.de/image/VHlVz9/a7618df93c.png]


So then I started to play around with the exposure slider ( I know you should not mess with that but I was trying to find a best match) and strange enough, the best setting seemed to be an exposure compensation of -0.75 for the *** KF?!

[Image: http://image-upload.de/image/x2UfuF/fdbb5b6b60.png]


Obviously LRT calculates a -1 EV compensation for each Keyframe pair from the EXIF info and this SHOULD be correct. Do you have any idea why in my case this seems so far off?
The intervall between photos was 1 minute, could this be the reason? Since it is a rather long time and around sunset, the light changes quickly, maybe it has already gotten another 0.25 EV darker, hence I only need to compensate for the remaining -0.75 to match the other keyframe?
Have you ever noticed something similar?
And more important, do you see a solution to this other than manually editing all the keyframe pairs in LR?
And how do I tell LRT to keep the manual settings from LR but calculate the transitions in between the keyframes?

I'm also not entirely convinced that this is the reason for my first observation, but it could be. Since I had to adjust the camera very quickly to follow the light, the sequences between KF are pretty short.


EDIT: I also tried a second pass to deflicker the sequence using part of the sky as a reference area but there was basically no change.
Thanks!

Hi, regaarding your first question: that's why there is the "rotate" slider - after the first check in Lightroom, if you notice, that the overall brightness change is not correct, just rotate the curve and try again. Please make sure to watch my holy grail tutorial: http://lrtimelapse.com/tutorial

Regarding the fact that sometimes the matching is not perfect: this can have several reasons. LRT just calculates the required offset from the exif data - but sometimes those values are not accurate, if you have intervals of one minute for example and it gets darker in that meantime, the camera won't notice that - so this might be the reason.

It is totally okay to manually fine tune the matching in those cases. Just make sure not to touch the Auto HG slider anymore after doing this, because that would reset your changes.

Hope I could help!

Thanks for the reply!
So in my case I'm a bit in trouble since if I rotate the curve, either the starting or the end frame will be to dark or too bright ... is there a rule, e.g. should the yellow curve always decrease towards the right? As you can see in my case it's almost horizontal.

I'll watch the tutorial once again when I get back home, maybe I missed some things.