I'll try again, hopefully not leaving out any information.

I was shooting a drone hyper raw DNG files from a Mavic Pro 2. I manually did 2 exposure jumps. LRT recognises that.

Import into LRT, Key frame wiz, holy, save, drag to LR.

In LR, I select all files in grid library view, read metadata from files.

I generally keep it to the tone, and WB, with the allocated grad filters. I use scripts to copy to the next keyframe (image shows what I have changed).

Go back to library view and select all, save metadata to files.

LRT, reload, auto transition, Vis preview. It fixes the second exposure change, but not the first. Saying all that, I think the issue is happening on the initial import.

Even before I do anything, LRT recognises the 2 exposure changes as shown by the orange and lavender lines (1/3 stop change). The aqua line however ( I've gone over with red line to stand out easier), shows only 1 significant exposure change.

I loaded the files off the disk and tried again, same issue.

Could I possible have a glitch in the original photo metadata?

I'm out of ideas.

Thanks,

Operating from a Macbook pro 16" 2019, Catalina 10.15.5. LRT 5.4

This post was last modified: 2020-06-14, 13:12 by Stef Kuipers.

Maybe a stupid idea: what if you just ‘delete’ that specific frame?
(Not that experienced in the software jet)
It looks to me you take the normal steps...

I assume you've done some time-lapses. In the middle of these time-lapses, remove one of the frames. Even better, if you have done a hyper-lapse, remove one of the frames from the centre of the shot. Watch the play back.

Seeing the result is a lot better and easier than me trying to explain it.

When you start a holy grail time-lapse, you're pot committed. I won't buy a SONY again as that wigs out when trying to roll exposure at Holy grail lapses. That's four hours wasted when it doesn't work.

We pay money for things to work like our clients pay us money for results. Simple.

Sorry, thank you for your reply though.

Have you tried to process the sequence without the „Holy Grail“ Wizzard? Just as if it was a „normal“ sequence without manual exposure changes?
Maybe give it a try, und just compensate the exposure jumps during the deflicker-Process?
I often use this method, if i have just a few manual exposure changes.
And it works often better in my experience.

It could be a problem with the metadata. Becaus you can see, that both jumps are the same size (blue curve). But the adjustment that the HG Wizard does is double the size on the first jump than on the second. That means, it over compensates for the first jump while doing it well on the second.
Unfortunately I cannot see the values for shutter/iso/aperture in your screenshot. But that's where you should have a look.

But even in that case, a couple of deflicker/refine steps should get rid of the jump.

Thanks Daniel,

I'd like to try that. How do you manually do these adjustments?

Cheers,

Thanks,

I noticed last night, that I should have added to this post, that with the first exposure change, the HG Lum levelling column has two bumps or two distinct jumps. Unlike the second exposure change that has only one.

There was no ISO or shutter changes made.

I also tried to run it through DNG converter first to see if that would help. Any idea how this happens or how to tidy up errors in metadata?

These are all tests on drone hyper-lapses at the moment.

Thanks,

Please don't run it through the DNG converter!! Use the original sequence!
No one knows what the Mavic did there, Metadata handling by DJI is often a mess, this is not the first problem we experienced in that area.
Again, just ignore that bump and do a couple of Deflicker/Refine steps and you'll have a smotth sequence. Check out the deflicker tutorial if you don't know how to do it: https://lrtimelapse.com/tutorial/

If you really want to dig in deep, send me the part with the double jump at the beginning (10 images around the jump will be enough) then I'll have a look. Send it via wetransfer or https://fromsmash.com/ to support(at)lrtimelapse.com.

Hi,

Thanks, I did deflicker/refine and it did fix it. That's good to know it can bail me out.

I'll probably do another test this evening, see if it occurs again. Thanks for the offer, it's not necessary though, unless you're curious and then sure, I can put the whole sequence up for you.

Thanks again