Good afternoon followers,
after having successfully shot and processed some time lapses during daylight conditions I wanted to try myself on the HG-technique so I shot a sunset and a sunrise sequence during my holidays. To prepare myself I’ve read a lot about how to set up my camera and about the processing afterwards as well.
I used a Nikon D5300 in M-mode and fixed aperture at 3,5, started with a shutter speed of 0,8 and an ISO setting of 400 and decreased it to ISO100 before I gradually decreased the shutter time. I controlled all that manually via the qDslrDashboard app.
My problem now is that after having gone through all the post processing steps that are introduced in Gunter’s tutorials, it seems that for any reason the match of the two- and three-star keyframes does not really work. The curves in my first screenshot seem to be normal but after the whole process through Lightroom (5) and the following auto transition in LRTimelaps (4) the Liminance curve still shows the exact shape like the one before having had applied the holy grail wizzard….
I tried to deflicker the whole sequence but the output was only a little bit better and still not acceptable.
I also tried the whole processing with the sequence of the sunset that I took but I was faced with exactly the same problem. Cause after having rendered the video one could exactly see the jumps where I had adjusted the camera settings.
It seems that there must be a big mistake/problem either with my camera settings, the way I process the sequences or even both together but I just cannot find the reason and any solution for it…
Has anyone maybe experienced the same or similar a problem or can tell me how to overcome it by analyzing the provided data? Would be really great..!
Cheers
Johannes
Did you use the sync script in Lightroom like explained in the instructions? Something seem to have gone wrong there.
Please start over, and try the process again. Start by initialzing (Metadata->Initialize) in LRTimelapse.
Hi Gunther,
Thanks a lot for your quick reply!
I must admit that I was already faced with exactly the same problem in LRT3. I just updated it to the latest version cause I thought it could get any better but unfortunately the result stayed exactly the same. I had already watched your tutorials for LRT4 before but as the version was new for me I watched it again while I was simultaneously working on my own HG sequence following you step by step and so did also use the sync script in Lightroom as explained by you….
Hi Gunther,
So I went through the whole process once again like instructed and this time I used only 4 key frames, definitely used the sync script and made only minor adjustments in Lightroom and even did not use any filters as I had read that big adjustments could lead to problems... but still, my luminance curve looks weird and one can clearly see where the adjustments had been made in the rendered video...
I must admit that I am at a loss with this problem...
Hmm, I'm surprised that the green deflicker reference curve is so far away from the pink one?!
Can I have a screenshot from the curves in the normal mode, with visual previews off?
Maybe you could try as well upgrading to the latest version that I've relesed today, I did some adjustments to the holy grail algorithm. With the new version, just launch the holy grail wizard once more on this sequence then save and turn on the visual previews again.
Unfortunately I do not have the sequence with that strange curves anymore cause I was constantly trying to improve this time lapse. But I figured out that if I was not doing any adjustments in Lightroom I got already a nice and smooth transition between the two- and three- star keyframes just using the HG-wizzard. So there probably must have been a bug which prevented a proper communication between LRT and LR. Deinstalling and getting the latest versions of LR & LRT did not help so I reconfigured a completely new operating system.
Now it seems to work better but you still can see a big difference between the luminance curve that is shown when I've made some adjustments in Lightroom and the one without any adjustments. After having used the deflicker tool for 4 times I finally got a quite smooth transition but I think that I would have had no chance getting this result in LRT3.
So Gunther, or maybe anybody else... have you got any idea about what is going wrong with my HG- sequences? Am I doing something wrong in the shooting part or am I doing something wrong during the post processing? I mean, in your tutorials you use the deflicker as well but after having used the HG-wizzard your curves never have such strong amplitudes at the positions where any adjustments originally had been made at the camera.
Pic 1: Normal workflow (including LR) original preview
Pic 2: Normal workflow visual preview - no deflicker
Pic 3: Normal workflow - 4 times deflickered with value 20
Pic 4: Without LR-workflow original preview
Pic 5: Without LR-workflow visual preview
Thanks a lot already in advance! It is unique to get supported by you, the one who's videos made me getting more and more fascinated by time lapses that I finally bought a DSLR to shoot time lapses myself
I think this looks pretty normal. The exif data of the cameras is mostly not really accurate, so the visual luminance curve before deflickering might still look a bit "zigzag". Apart from this, the LR tools introduce flicker as well, if heaviliy used.
I'd set a reference area to the sky for sequences like this as well, before deflickering, this might give better results too.
Apart from this, I fine tuned the HG Wizard in LRT4.0.3 - so make sure you have installed that version.
Hi,
I got the exact same problem with the visual luminance of an holy grail sequence going crazy. But I think, LRTimelapse 4.0.3 has nothing to do with it. For me, it only comes from the fact that my EXIF files seems to have been written complety wrong when I shot the sequence. I've shot it with the AETTR technique with Magic Lantern on a Canon 550D.
If you look at the luminance curve from LRTimelapse, you can see many bumps on the luminance curve (I didn't apply the deflicker yet but I think it will correct everyting).
[
attachment=970]
I've done a picture comparison (these are the 2 succesive original pictures with the luminance bump where the slider is in the LRTimelapse preview) to show that even if the EXIF data are the same (especially the 1/200 sec shutter speed which is the wrong written metadata I think), the 2 successive pictures don't have the same luminosity even the weather didn't change at that time (I've done the twist lens technique, so it's not aperture flicker).
[
attachment=971]
I'll try to find why these EXIF data are not well written when shooting or maybe someone here have an idea ?
EDIT : after one pass visual deflicker (value around 15 with reference area on the sky), everything looks fine as expected (really awesome that new visual deflicker Gunther even with that crazy luminance curve
)
[
attachment=975]
Yannick
Magic Lantern is known for writing wrong exif values. You might search the forum for Magic Lantern, other users already experienced that problems too.
Yes that's a problem with ML not writting the right value for the shutter speed into the EXIF Data (when ETTR uses 1/2 EV and 1/3 EV interval shutter speed). I've found a way to make it right with exiftool by replacing the "EXIF:ExposureTime" by the "Makernotes:ExposureTime" value which seems to be more accurate.
But there's still a problem when using the ETTR method with LRTimelapse. When the shutter speed changes each 1,2,3 or 4 pictures, the HG correction will still let appear some big bumps in luminance (as you can see on the center)
[
attachment=978]
I think I don't have to worry about it cause the deflicker will correct the problem easily.