This post was last modified: 2013-02-27, 12:41 by qtluong.

I am trying to use LRT to salvage a night to day sequence which I'll be the first to admit wasn't well shot. It was done on top of a 4000m peak on which I spent the night with some altitude sickness which affects the brain :-)

As I am using the recommended standard workflow with LR4, I run into a problem which is driving me crazy after two late nights. Please help !

After "auto transition", I get the values shown in attachment lr1, which are plausible.
   

After "deflicker", I get the values shown in attachment lr2, which again look ok.
   

However, when I click "XMP Save", I get the values shown in attachment lr3. Those have changed a lot compared to the two previous one, and are totally off. See for instance line 48, where the "Exposure 2012" value has gone from -2.5 (the value I set manually in LR for a keyframe) to a crazy -3.936.
   

What went wrong ?

Could you please try this with the latest beta from http://lrtimelapse.com/download
I fixed some issues there for holy grail in combination with deflicker. Please let me know if that helps!

Could you please try this with the latest beta from http://lrtimelapse.com/download
I fixed some issues there for holy grail in combination with deflicker. Please let me know if that helps!

This seems better, at least the exposure of the keyframe 48 doesn't change, although I see large jumps elsewhere.

If I understand correctly, when I click "XMP Save" after "Deflicker", then "Exposure 2012" is replaced by "Exposure (Deflick.)", is it correct ?

If so, that's not very intuitive, because one doesn't expect "Save" to affect values. After clicker "Deflicker", the two columns "Exposure (Deflick.)" and Exposure 2012 should probably be swapped, with titles such as "(Old Exposure 2012)" and "New Exposure (Deflick.)".

It's difficult to say anything without knowing your sequence. Maybe you can post a screenshot of your preview panel with the curves.

Yes, when saving the deflicker will be applied to the exposure. That's by design, otherwise it wouldn't work. However, if you later try to apply deflicker again, you will be asked to remove the old deflicker and can so even remove deflicker completely (from the exposure column) if you turn it off again. Just try it!

This post was last modified: 2013-03-02, 06:56 by qtluong.

OK, this answers half my question (why there are strange values appearing when you click save). The other half of the question is about the values themselves.

Here is a complete screen shot after "auto transition". If I render that sequence, it is somewhat acceptable with some flicker.

   

Here is complete screen shot after "deflicker". If I render that sequence there are some huge exposure jumps. You can see that on the line 53, the exposure went from 1.850 to 3.442, although the Flicker was estimated only at -0.250 which does look wrong to me (clearly I made an exposure mistake on that frame, but this was corrected manually in LR and shouldn't have any bearing on the algorithm).

   

If you are interested in investigating further, I can make this sequence available to you by download for research purposes.

Hi, just from the first image I can see that that won't work out. You already messed it up when shooting, you did't work consistently in manual mode and thus now there are so many Holy grail Keyframes (orange ones) one beneath the other - that's not how that approach works!

Check my Holy Grail tutorial again - usually you change Shutter/ISO once and then you leave the camera alone for at least 30-50 frames, then you chage again. I don't know exactly what you did and if you used the wizard to create the keyframes or did it manually. But that way it won't work out well, sorry.

As far as I can see at least at the beginning of your sequence you worked in automatic mode. The keyframes wizard normally detects that and then doesn't create Holy Grail keyframes at all or it asks you. It could however be that the smooth ending of the sequence confused him and lead to the assumption it was a HG-Shoot.

What exactly is causing that spikes in the beginning of the sequence?
I noticed that you didn't set a reference are as well. You might want to set that to the sky. Did you watch my tutorials?

Sorry, I think you have two options:
1.) remove all thos orange keyframes and work with some normal keyframes (ignoring the spikes) - then use deflicker to try to sort that out. But I can assure you this won't get perfect as well.
2.) delete the first part of the sequence with those spikes ans at least save the last part.

Good luck!

Thanks again for looking at the data and for your advice.

Are you saying that for the HG keyframes (marked with ** and ***) to be processed properly by LRT you need to work in manual mode and have at least 30 frames without exposure parameter changes ? And therefore, if I didn't work in that way, then I shouldn't have any keyframe marked **/*** ?

Again, I acknowledge that the sequence wasn't well shot. At the beginning, I was operating in manual mode because it wasn't bright enough for automatic. When it got brighter, I switched to Auto. The spikes at the beginning of the sequence are due to some bad adjustments I made. I read your PDF but have not watched the tutorials.

OK, I just watched the tutorial. Very well made I must say !

However I see that in it, you are dealing with 3 exposure jumps. In my sequence, I am going from starry night (15s f/1.4 ISO 3200) to early morning (1/500s f4 ISO 100). This represents a difference of 21 EV !

I am wondering how practical your suggested workflow would be when you have to do so many adjustments. Besides, those adjustments take place over about 440 frames, so it means that on average you'd have to do an adjustment every 22 frames, probably less in some cases, short of the 30-50 frames you mention.

Since this question is larger in scope than my initial one, I'll start a new thread for it.

I've now tried to rework the sequence using only normal keyframes. If I do not use deflicker, the results are OK with some flicker present (as expected). If use deflicker, the smooth parts of the sequence look nicer (less flicker), but some of the exposure values for the keyframes go wild, making the sequence unusable. I thought that the deflicker algorithm was supposed not to affect keyframes ?

When working with deflicker, you have to carefully set your reference area. Especially when using many "jumps" deflicker will be a sensitive thing because it will deflicker inbetween the jumps, this might introduce some small changes even if the keyframes are held constant.

Normally If you shoot those types of sequences (from day into night) you know what you do and shoot in a way you don't have flicker at all. I recommend doing the shooting right and omitting deflicker whenever possible in the HG-shots.