When shooting a sunrise to practice my Holy Grail techniques, I found myself in a position of having to replace the battery in the camera. Because I had not practiced this before (sigh), I inadvertently missed two shots.

The rendered timelapse looks really good, so I can't just write it off as a practice session to be forgotten about. The two missing shots, however, cause a noticeable (to me) jump in the flow of the clouds.

I would like to repair this jump somehow, and I was wondering if anybody had any suggestions.

My first thought is to duplicate the two frames around the gap, inserting the duplicates in the gap. Although this may still leave a bit of a stutter, I suspect that the flow of the clouds won't be interrupted and the jump will be much less noticeable.

If I were to pursue this route, I have a question on how to get LRTimelapse to put these additional shots in the right spot. Do I need to hack the capture time in the raw file somehow? Can I just name the duplicates in such a way that LRTimelapse will insert them in the sequence when I sort by filename?

Thanks in advance.
Rob.

I did this once or twice as well, it's indeed the best option. When you render then, best use the new "Motion Blur" feature, that will do a great job smoothing that out.

Just duplicate the frame before the missing shot, and rename it with ending 1

So for example
DSC1234 <- original
DSC12341 <- copy of 1234
DCS12342 <- copy of 1234
DSC1235 -< original

Normally this should work with the normal sort by datetime, because the timestamps are equal, but if not, try sort by filename.
Please make sure that the sorting is correct in Lightroom as well before exporting!

Let me know, I didn't try the naming...

Thanks Gunther. I did as you recommended, with one small change:

DSC1234 <- original
DSC1234a <- copy of 1234
DSC1235 -< original
DCS1235a <- copy of 1235

Instead of having three frames of one image, I have two frames of two images.

LRT had no problem ordering the files based on capture time, so I didn't need to try using file order.

Lightroom appears to have not had a problem exporting the files in the right order either.

Unfortunately, I now see that as I changed the battery I shifted by view by about two pixels. This means that the perception of the jump is still there, no matter what I do with the missing frames. I'll have to do some more Lightroom surgery with the cropping rectangle to remove the shift from the experiment before I can tell if the additional frames helps with the perception of continuity at all.

I'll let you know how it goes.

Thanks Gunther. I did as you recommended, with one small change:

DSC1234 <- original
DSC1234a <- copy of 1234
DSC1235 -< original
DCS1235a <- copy of 1235

Instead of having three frames of one image, I have two frames of two images.

LRT had no problem ordering the files based on capture time, so I didn't need to try using file order.

Lightroom appears to have not had a problem exporting the files in the right order either.

Unfortunately, I now see that as I changed the battery I shifted by view by about two pixels. This means that the perception of the jump is still there, no matter what I do with the missing frames. I'll have to do some more Lightroom surgery with the cropping rectangle to remove the shift from the experiment before I can tell if the additional frames helps with the perception of continuity at all.

I'll let you know how it goes.

Try the Time Blending in After Effects. Experiment around from 50 to 90%. I had also a slight shift in one of my TL - actually I even replaced the lens and It worked very well, no noticeable shift.

Try the Time Blending in After Effects. Experiment around from 50 to 90%. I had also a slight shift in one of my TL - actually I even replaced the lens and It worked very well, no noticeable shift.

I don't have After Effects at this point, but as many of my experiments are suffering from jitter from the wind I expect I'll need to pick it up at some point. You would recommend it?

Try the Time Blending in After Effects. Experiment around from 50 to 90%. I had also a slight shift in one of my TL - actually I even replaced the lens and It worked very well, no noticeable shift.

I don't have After Effects at this point, but as many of my experiments are suffering from jitter from the wind I expect I'll need to pick it up at some point. You would recommend it?

I use it only for certain effects and corrections. If the jitter is not too bad you may correct it also with After Effects. The disadvantage now is the creative cloud nonsense of Adobe, means that you may only rent the software. I still stick to an older version hoping that Adobe will change his mind. AE is very powerful and complex for certain needs. I prefer Motion and Final Cut fi
or editing.

I just downloaded a trial of Premiere and used it to apply warp stabilization to my problematic little sequence. I've tried to create a video with four different experiments in it:

http://sdrv.ms/1iTk0eq

(Please let me know if you cannot see the video from this link, and I will post it to youtube or something.)

There are four sequences:
1. A video from the LRTimelapse renderer, stabilized in Premiere. This sequence is missing two frames.
2. A video from the LRTimelapse renderer, stabilized in Premiere. This sequence duplicates the frames on either side of the gap to preserve temporal continuity (if you know what I mean).
3. A video created from a JPG sequence exported from Lightroom, stabilized in Premiere. This sequence is missing two frames.
4. A video created from a JPG sequence exported from Lightroom, stabilized in Premiere. This sequence duplicates the frames on either side of the gap.

You can see the jump best if you follow the small cloud from centre of frame as it moves to the right.

I find that the stabilization in 1 and 2 is not fully successful (likely due to me not knowing what I'm doing) and the jump is still noticeable with or without the duplicate frames.

The stabilization of the sequence created from the JPG is much smoother, and so I think 3 and 4 are a good experiment on the effects of duplicating frames to fill gaps.

Surprisingly, I find that just leaving the gap is less noticeable. I think that when the frames are duplicated, the duplication looks like a stop, and the subsequent jump is even *more* noticeable than if the missing frames are left out. This is not what I expected.

What do you all think?

I just downloaded a trial of Premiere and used it to apply warp stabilization to my problematic little sequence. I've tried to create a video with four different experiments in it:

http://sdrv.ms/1iTk0eq

(Please let me know if you cannot see the video from this link, and I will post it to youtube or something.)

There are four sequences:
1. A video from the LRTimelapse renderer, stabilized in Premiere. This sequence is missing two frames.
2. A video from the LRTimelapse renderer, stabilized in Premiere. This sequence duplicates the frames on either side of the gap to preserve temporal continuity (if you know what I mean).
3. A video created from a JPG sequence exported from Lightroom, stabilized in Premiere. This sequence is missing two frames.
4. A video created from a JPG sequence exported from Lightroom, stabilized in Premiere. This sequence duplicates the frames on either side of the gap.

You can see the jump best if you follow the small cloud from centre of frame as it moves to the right.

I find that the stabilization in 1 and 2 is not fully successful (likely due to me not knowing what I'm doing) and the jump is still noticeable with or without the duplicate frames.

The stabilization of the sequence created from the JPG is much smoother, and so I think 3 and 4 are a good experiment on the effects of duplicating frames to fill gaps.

Surprisingly, I find that just leaving the gap is less noticeable. I think that when the frames are duplicated, the duplication looks like a stop, and the subsequent jump is even *more* noticeable than if the missing frames are left out. This is not what I expected.

What do you all think?

In my opinion number 3 looks best. The one with the gap.

In my opinion number 3 looks best. The one with the gap.

Thanks anna! That appears to be the consensus of the people I'm able to corner into watching the clip as well.

I've had a thought as to whether maybe the duplicated frames are being treated in a different order between LRT (when applying the interpolated settings) and Lightroom (when exporting). I'm only able to use Capture Date as a sort in LRT because the capture number of the shot rolled over to IMG_0000.CR2 and I didn't rename when importing. I'm wondering if maybe the two applications are switching the order of the duplicated frames and there is enough of a ramp between them that when switched the transition effect is even more disconcerting.

Sounds like another experiment is needed...